McGraw-Hill

ELECTRICAL AND ELECTRONIC

ENGINEERING SERIES

FREDERICK EMMONS TERMAN, Consulting Editor

ELECTRON-TUBE CIRCUITS
 :McGraw-Hill Electrical and Electronic Engineering Reries
FREDERICK EMMONS TEHMAN, Consulting Editor

BAILEY AND GAULT' Alternating-current Machinery

CAGE . Theory and Application of Industrial Electronics
CUCCIA' Harmonics, Sidebands, and Transients in Communication
Engineering
EASTMAN' Fundamentals of Vacuum Tubes
FITZGERALD AND KINGSLEY' Electric Machinery
GEPPERT' Basic Electron Tubes
HESSLER AND CAREY' Fundamentals of Electrical Engineering
HILL . Electronics in Engineering
JOHNSON' Transmission Lines and Networks
KRAUS . Antennas
LEPAGE . Analysis of Alternating-current Circuits
LEPAGE AND SEELY' General Network Analysis
MILLMAN AND SEELY' Electronics
RUDENBERG . Transient Performance of Electric Power Systems
SEELY' Electron-tube Circuits
SISKIND' Direct-current Machinery
SKILLING . Electric Transmission Lines
SKILLING' Transient Electric Currents
SPANGENBERG' Vacuum Tubes
TERMAN . Radio Engineering
TERMAN AND PETTIT' Electronic Measurements
ELECTRON -TUBE
CIRCUITS

BY
SAMUEL SEELY, PH.D.
Professor of Electrical Engineering
Syracuse University

FIRST EDITION
FOURTH IMPRESSION

NEW YORK TORONTO LONDON

1vlcGRA W-HILL BOOK COMPANY, INC.
1950
 ELECTRON-TUBE CIRCUITS
Copyright, 1950, by the McGraw-Hill Book Company, Inc. Printed in the United
States of America. All rights reserved. This book, or parts thereof, may not be
reproduced in any form without permission of the publishers.

THE MAPLE PRESS COMPANY, YORK, PA.

 PREFACE
This book is the outgrowth of several courses that were organized by
the author on electron-tube circuits and applications and that covered
many of the important circuits in widespread use during the Second World
War. It seeks to achieve the following: (1) to develop in the student a
clear analytical method in the study of electron-tube circuits; (2) to
present and study the various classes of circuits which find widespread
application; (3) to indicate with examples how one proceeds to combine
circuits of various types to achieve either one or a multiplicity of opera-
tions. It is not intended to include either a comprehensive discussion of
all aspects of any given field or all possible circuits or methods for achiev-
ing a given result. It is hoped that representative circuits have been
included and studied and that the reader may find some suggestions that
will be of assistance to him in a particular development.
The choice of material to be included and the detail of coverage were
the subject of much thought. The principle adopted was to present a
coordinated account of each broad field of application, with the main
emphasis on the operation of many of the significant circuits. No claim
is made to completeness of coverage, nor are the fine points of any given
field discussed in detail. Where general practice has favored a given type
of circuit, the major emphasis is on these.
Approximately one-half the content is of a radio-engineering character,
the remaining material being extensively used in radar, television, pulse
communication, and general electronic control. Sufficient material is
contained for a course in radio-engineering circuits and for one in non-
radio electronic circuits. Sufficient diversity exists for the instructor to
choose topics to satisfy almost any course requirements. It has been
assumed that the student has completed his basic studies in a-c circuit
theory and in basic electronics before undertaking a study of the text.
The order of presentation of the Il).aterial has been dictated by the
character of the analysis rather than by the application. Because of this,
circuits of diverse application may be found in a given chapter. How-
ever, the book divides itself quite naturally into a number of major
sections. The first part of the volume is devoted to a review of the
fundamental properties of electron tubes and their basic circuit applica-
tions. The next part of the volume includes a discussion of a variety of
amplifiers of the untuned variety. This includes circuits in which the
V
VI PREFACE

tube is operated as a linear device and other circuits in which the non-
linear capabilities of the tube are employed. The former include simple
voltage amplifiers, video amplifiers, power amplifiers, and electronic com-
puting circuits. The latter include circuits which utilize the tube as a
switch. The third part of the book contains a discussion of circuits of the
tuned variety and discusses such topics as tuned voltage amplifiers,
tuned power amplifiers, and oscillators. A comprehensive treatment of
power rectifiers, filters, and regulators is followed by a discussion of
amplitude modulation and demodulation and frequency modulation and
detection. The latter part of the text includes a treatment of circuits
that have been largely extended by developments in radar applications
during the course of the war. This discussion is considerably more
detailed and more extensive than has heretofore appeared in any general
text.
An effort has been made to include sufficient analysis of the operation
of the circuits to indicate clearly the operation and the various factors on
which the operation depends. This has a twofold purpose, one of which
is to indicate the procedure that must be adopted in effecting an analysis
and the second of which is to indicate the factors on which the operation
depends. This is considered to be very important, since in some instances
the tube plays a direct part in the operation of the circuit, whereas in
others it may serve simply in the capacity of a switch. However, the
mathematical developments are only a part of the analysis, since the
discussion attempts to introduce the physical aspects of the problem and
then to incorporate the mathematical results into the complete analysis.
A rather regrettable situation will be found to exist in the matter of
notation. This arises from the author's desire to conform to the Institute
of Radio Engineers standards on vacuum-tube notation. However, such
single-subscript notation in electron-tube circuits is often inadequate, and
double-subscript notation is employed, except for those particular cases
where no confusion is likely to arise. The result is a mixed single-sub-
script and double-subscript system of notation, the single-subscript terms
generally conforming to the IRE notation.
A controversial matter is also to be noted. Throughout the text the
symbols a-c and d-c are used as adjectives. Purists might object that the
word current in a-c current is redundant and that the phrase a-c voltage is
fundamentally meaningless. However, the use of the symbols a-c and
d-c as descriptive adjectives is becoming increasingly widespread and does
provide a clear and convenient abbreviation.
A number of problems have been included at the end of each chapter.
These have been formulated in a way that requires an understanding of
the subject matter. As a result, all text assignments may be Eupple-
 PREFACE Vll

mented by problem assignments. Problems which entail nothing more

difficult than the substitution of numbers into equations have been kept
to a minimum. Wherever possible, the problems are based on practical
data in order to familiarize the student with such practical details.
To provide proper acknowledgment of the source of much of this
material proves to be an impossible task. Much of the material that is
principally of a radio-engineering character has appeared in one form or
another in a wide variety of sources over many years, and the significant
original sources seem to have been generally neglected. The principal
source of many of the circuits which were extended for use in radar
applications was the M.LT. Radiation Laboratory, of which the author
was a staff member during the war. However, it is known that many of
these circuits were adapted from existing circuits of diverse origin, whereas
some were developed at other laboratories, including British laboratories.
In only a few cases is the identity of the groups who did some of this work
known.
Special mention must be made of the freedom with which the author
drew on his earlier text, "Electronics" (by J. Millman and S. Seely,
McGraw-Hill Book Company, Inc., New York, 1941). Certain of the
material closely parallels that in the earlier book.
The author wishes to acknowledge many helpful discussions with a
number of his colleagues. He is particularly indebted to Professors
David K. Cheng and Glenn M. Glasford, both for such discussions and for
their assistance in proofreading portions of the text. Thanks are also
due to the General Electric Co. and the RCA Manufacturing Co. for
freely supplying many photographs and tube characteristics.

SAMUEL SEELY
SYRACUSE, N. Y.
November, 1949
 j
j
j
j
. j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
 CONTENTS
PREFACE . . . . . . . . . v

CHAPTER I-INTRODUCTION 1
2-CHARACTERISTICS OF ELECTRON TUBES. 9
3-VACUUM TUBES AS CIRCUIT ELEMENTS. 39

4 -BASIC AMPLIFIER PRINCIPLES. . 52

5-UNTUNED VOLTAGE AMPLIFIERS. 68

6- UNTUNED VOLTAGE AMPLIFIERs-Continued. 95

7-SPECIAL AMPLIFIER CIRCUITS. . . 123

8-ELECTRONIC COMPUTING CIRCUITS. 146

9-UNTUNED POWER AMPLIFIERS. 167

10--TUNED . VOLTAGE AMPLIFIERS . 192

II-TuNED POWER AMPLIFIERS. 213

12-0SCILLATORS 244

13-RECTIFIERS. 271
14-RECTIFIER FILTERS AND REGULATORS 292

IS-AMPLITUDE MODULATION. 320

16-DEMODULATION . . . • . 343
17-FREQUENCY MODULATION AND DETECTION 363

18-RELAXATION OSCILLATORS . • . . • . 395

19-HEAVILY BIASED RELAXATION CIRCUITS 414

20-SWEEP GENERATORS • . . . . . 442

21-SPECIAL SWEEP GENERATORS . 460

22-ELECTRONIC INSTRUMENTS 473

ApPENDIX A-MILLMAN THEOREM . . . 489

B-PLATE CHARACTERISTICS OF RECEIVING-TYPE TUBES . 491

C-CHARACTERISTICS OF TRANSMITTING TUBES . . . 508

D-TABLE OF BESSEL FUNCTIONS OF THE FIRST KIND 511

INDEX . • . • • . • • • • • • • • • • . . . . . . . . . . 513

ix
 CHAPTER 1

INTRODUCTION

BEFORE one undertakes a study of circuits that incorporate electron

tubes, it might be well to examine, even superficially, certain of the funda-
mental physical principles which govern the operation of these tubes.
There are two important basic questions that relate to such tubes. One
relates to the actual source of the electrons and their liberation, and the
second relates to the control of the electron beam. A brief discussion of
these matters will be included.
According to modern theory, all matter is electrical in nature. The
atom, which is one of the fundamental building blocks of all matter,
consists of a central core or nucleus which is positively charged and which
carries nearly all the mass of the atom. Enough negatively charged
electrons surround the nucleus so that the atom is electrically neutral in
its normal state. Since all chemical substances consist of groups of these
atoms which are bound to each other, then all matter, whether it is in the
solid, the liquid, or the gaseous state, is a potential source of electrons.
All three states of matter do, in fact, serve as sources of electrons. A
number of different processes serve to effect the release of electrons, those
which are of importance in electron tubes being (1) thermionic emission,
v~ (2) secondary emission, (3) photoelectric emission, (4) high field emis-
sion, and (5) ionization. These processes will be considered in some detail
in what follows.
With the release of the electrons, a means for their control must be
provided. Such control is effected by means of externally controlled
electric or magnetic fields, or both. These fields perform one or both
of the following functions: (1) control of the number of electrons that
leave the region near the emitter; (2) control of the paths of the electrons
after they leave the emitter. Control method 1 is the more common, and
such a control method is incorporated in almost all electron tubes, except
those of the field-deflected variety. The cathode-ray tube is a very
important example of a field-deflected tube. However, even in this
latter case, a control of type 1 is incorporated to control the electron-tube
current, even though the subsequent motion is controlled by means of an
electric or a magnetic field, or both.
1
2 ELECTRON - TUBE CIRCUITS [CHAP. 1

1·1. Thermionic Emission. Consider matter in the metallic Btate.

Metals are most generally employed in the form of a wire or ribbon
filament. If such a filament contains electrons and if these are relatively
free to move about in the metal (and this is the case since the application
of a small potential difference between the ends of the wire will result in a
current flow), it might be expected that some electrons might "leak"
out of the metal of their own accord. This does not occur, however.
Consider what happens to an electron as it seeks to escape from a metal.
The escaping, negatively charged electron will induce a positive charge
on the metal. There will then be a force of attraction between the
induced charge and the electron. Unless the escaping electron possesses
sufficient energy to carry it out of the region of influence of this image
force of attraction, it will be returned to the metal. The minimum
amount of energy that is required to release the electron against this
attractive force is known as the work function of the metal. This
requisite minimum amount of energy may be supplied by anyone of a
number of different methods. One of the most important methods is to
heat the metal to a high temperature. In this way, some of the thermal
energy supplied to the metal is transferred from the lattice of the heated
metal crystals into kinetic energy of the electrons.
An explicit expression relating the thermionic-emission current density
and the temperature of the metal can be derived.l·* The expression so
derived has the form

(1-1)

where Ao is a constant for all metals and has the value of 120 X 10 4
amp/(m2)CK2) and bo is a constant that is characteristic of the metal.
The quantity bo is related to the work function Ew of the metal by
bo = 1l,600E w OK (1-2)
It has been found experimentally that Eq. (1-1) does represent the form
of the variation of current with temperature for most metals, although
the value obtained for Ao may differ materially from the theoretical value
of 120 X 10 4 amp/(m2) (OK2).
It follows from Eq. (1-1) that metals that have a low work function
will provide copious emission at moderately low temperatures. Unfor-
tunately, however, the low-work-function metals melt in some cases and
boil in others, at the temperatures necessary for appreciable thermionic
emISSIOn. The important emitters in present day use are pure tungsten,
thoriated-tungsten, and oxide-coated cathodes. The thermionic-emis-
sion constants of these emitters are contained in Table 1-1.
* Superior numbers refer to references at the end of each chapter.
 SEC. 1-1] INTRODUCTION 3
TABLE 1-1
THE IMPORTANT THERMIONIC EMITTERS AND THE THERMIONIC-
EMISSION CONSTANTS

Emitter A o, amp/(m 2)CK2) Ew, volts

Tungsten ............................. . 60 X 10 4 4.52

Thoria ted-tungsten .................... . 3 X 104 2.63
Oxide-coated .......................... . 0.01 X 10 4 1

Tungsten is used extensively for thermionic filaments despite its rela-

tively high work function. In fact, this material is particularly impor-
tant because it is virtually the only material that can be used successfully
as the filament in high-voltage tubes. It is used in high-voltage X-ray
tubes, in high-voltage rectifier tubes, and in the large power-amplifier
tubes that are used in radio and communication applications. It has the
vdisadvantage that the cathode emission efficiency, defined as the ratio of
the emission current in milliamperes to the heating power in watts, is
small. Despite this, it can be operated at a sufficiently high temperature,
between 2600 and 2800 o K, to provide an adequate emission.
It has been found that the application of a very thin layer of low-work-
function material on filaments of tungsten will materially reduce the
work function of the resulting surface. A thoriated-tungsten filament is
obtained by adding a small amount of thorium oxide to the tungsten
before it is drawn. Such filaments, when properly activated, will yield
an efficient emitter at about 1800o K. It is found desirable to carbonize
such an emitter, since the rate of evaporation of the thorium layer from
the filament is thus reduced by about a factor of 6. Thoriated-tungsten
filaments are limited in application to tubes that operate at intermediate
voltages, say 10,000 volts or less. Higher voltage tubes use pure tungsten
filaments.
The oxide-coated cathode is very efficient (about twenty times as
efficient as tungsten) and provides a high emission current at the rela-
tively low temperature of 1000o K. It consists of a metal sleeve of konal
(an alloy of nickel, cobalt, iron, and titanium) or some other metal, which
is coated with the oxides of barium and strontium. These cathodes are
limited for a number of reasons to use in the lower voltage tubes, say
about 1,000 volts or less. They are used almost exclusively in receiving-
type tubes and provide efficient operation with long life.
Curves showing the relative cathode efficiencies of tungsten, thoriated-
tungsten, and oxide-coated cathodes are illustrated in Fig. 1-1. It will
be seen that tungsten has a considerably lower efficiency than either of
the other two emitters.
 ELECTRON - TUBE CIRCUITS [CHAP. 1

The thermionic emitters in their practical form in electron tubes may

be of the directly heated, or filamentary, type or of the indirectly heated
type, and in the case of gas and vapor tubes the cathode may be of the
22 heat-shielded type. Typical fila-
.... 20 mentary cathodes are illustrated in
118
~ 1&
~14
g 12
)/='1-
~~§
\:)~~~
\l- ~

~~~L----
Fig. 1-2. These filamentary cath-
odes may be of the pure tungsten,
thoriated-tungsten, or oxide-coated
.~ 10 type.
tt: 8 ,,'00~
'"
~ &
Y
:E 4
L /
.3 2
o
/
./' L ~~- -
o 10 20 30 40 50 60
Per cent mted fil~ment power
FIG. 1-1. Cathode efficiency curves of FIG. 1-2. Typical directly
an oxide-coated, a tho ria ted-tungsten, heated cathodes.
and a pure tungsten filament.

The indirectly heated cathode for use in vacuum tubes is illustrated in

Fig. 1-3. The heater wire is contained in a ceramic insulator which is
enclosed by the metal sleeve on which the oxide coating is placed. A
cathode assembly of this type has such a high heat capacity that its tem-

~ ~
FIG. 1-3.
§
g~
(~
Typical indirectly heated
~ [J
FIG. 1-4.
I
Different types of heat-
cathodes. shielded cathodes.

perature does not change with instantaneous variation in heater current

when alternating current is used.
Heat-shielded cathodes, which can be used only in gas-filled electron
tubes for reasons to be discussed in Chap. 2, are designed in surh a way
as to reduce the radiation of heat energy from the cathode. This mate-
rially increases the efficiency of the cathode. Several different types of
heat-shielded cathodes are illustrated in Fig. 1-4.
1-2. Photoelectric Emission. The energy that is required to release
an electron from a metal surface may be supplied by illuminating the
surface with light. There are certain restrictions on the nature of the
surface and the frequency of the impinging light for such electron emission
to take place. That is, electron emission is possible only if the frequency
~EC. 1-2J INTRODUCTION 5

of the impinging light exceeds a certain threshold value that depends on

the work function Ew of the surface according to the equation
= eEw (1-3)
f C h
where e is the charge of the electron and h is Planck's constant. The
corresponding threshold wave length beyond which photoelectric emission
cannot take place is given by
h = ~ = 12,400 A (1-4)
C eEw Ew
where A is the angstrom unit (10-8
cm). For response over the entire
visible region, 4000 to 8000 A, the
work function of the photosensitive
surface must be less than 1.54 volts.
The essential elements of a photo-
tube are the photosensitive cathode
surface and a collecting electrode, con-
tained in a glass envelope that either
is evacuated or contains an inert gas
at low pressure. A photograph of
such a phototube is shown in Fig. 1-5.
The number of photoelectrons per
square millimeter of area of a photo-
cathode is small, and it is customary
to use photocathodes of large area, as FIG. 1-5. A typical phototube.
shown.
The current characteristics of such phototubes for different collecting
potentials between the cathode and the collecting anode, with light
20
200 Toofcandles
+CO)",15 /
"'.... ....'"
5Uo;~IO I /00

.....
",0
'--
II
r
lS.~ 5
O:E
o
o 10 m ro ~ w w w W
PI"te potentiod, volts
FIG. 1-6. The volt-ampere characteristics of a type PJ-22 vacuum phototube, with
light intensity as a paramcter.

intensity as a parameter, are illustrated. Figure 1-6 shows the curves of

a vacuum phototube with light intensity as a parameter. Note that the
6 ELECTRON - TUBE CIRCUITS [CHAP. 1

current reaches near saturation values for very low values of applied
potential.
The presence in the glass envelope of an inert gas, such as neon or argon,
at low pressure materially alters the volt-ampere eurves. A set of char-
acteristic curves for a gas phototube are given in Fig. 1-7. The presence
of the gas in a phototube increases the sensitivity of the phototube, the
20

f~ 15
<)1..
L..
5U
Q)
(lOleV
.,0
~.~ 5
flO
IS
~~
30 ,..........-
V
C:::E I 15 __ I--"'"
00 10 20 30 40 50 60 70 80
PIVlte potentiVlI,Yolts
FIG. 1-7. The volt-ampere characteristics of a type PJ-23 gas-filled phototube, with
light intensity as a parameter.

current output for a given light intensity increasing with increased plate
potential, whereas the output remains sensibly constant in the vacuum
phototube.
20
15
V/ ......- /
+-- I - - - -.~
::i 15
t:
tID ~ PJ;;/f'
a I V E b =80v
/ /"
~ 5
IS
a: f-,-V
00 25 50 75 100 125 150 175 200
I1luminVlti on, footcOlndles
FIG. 1-8. Photocurrent as a function of illumination for a PJ-22 vacuum phototube,
and a PJ-23 gas-filled cell.

A significant comparison of the output from two phototubes, one of the

vacuum type and the other of the gas-filled type, other characteristics
of the tubes being the same, is contained in Fig. 1-8. Note that the
photocurrent for the vacuum phototube is a linear function of the illumi-
nation, whereas that for the gas-filled cell shows deviations from the
linear at the higher illuminations. However, the greater sensitivity
of the gas-filled cell is clearly evident.
1-3. Secondary Emission. It is possible for a particle, either an
electron or a positive ion, to strike a metallic surface and transfer all or a
part of its kinetic energy in this collision to one or more of the internal
SEC. 1-5] INTRODUCTION 7

electrons. If the energy of the incident particle is sufficiently high, some

of the internal electrons may be emitted. Several tubes have been
designed which incorporate secondary-emission surfaces as part of the
device, and highly sensitive phototubes have such auxiliary elements in
them. Frequently the secondary emission that exists is of a deleterious
nature. This matter will be discussed in explaining certain features of
the characteristics of tetrodes.
1-4. High Field Emission. The presence of a very strong electric field
at the surface of a metal will cause electron emission. Ordinarily the
field in the average electron tube is too small to induce such electron
emISSIOn. This process has been suggested to account for the electron
emission from a mercury-pool cathode in a mercury rectifier.
1-5. Ionization. The process in which an atom loses an electron is
known as ionization. The atom that has lost the electron is called a
positive ion. The process of ionization may occur in several ways.
Electron Bombardment. Consider a free electron, which might have
been released from the envelope or from any of the electrodes within the
tube by any of the processes discussed above. Suppose that this free
electron has acquired enough energy from an applied field so that upon
collision with a neutral atom, it removes an electron. Following this
action, two electrons and a positive ion exist. Since there are now two
electrons available, both may collide with gas particles and thus induce
further ionization. Such a process as this may become cumulative, with
consequent large electron release. This process is very important and
accounts for the successful operation of gas- and vapor-filled rectifier
tubes. It is also the basis of the gas amplification in gas-filled phototubes.
Photoelectric Emission. If the gas is exposed to light of the proper
frequency, then this radiant energy may be absorbed by the atom, with
resulting electron emission. This process is important in initiating
certain discharges.
Positive-ion Bombardment. The collision between a positive ion and a
neutral gas particle may result in electron release, in much the same
manner as by electron bombardment. This process is very inefficient
and is usually insignificant in normal gas tubes.
Thermal Emission. If the temperature of the gas is high enough, some
electrons may become dislodged from the gas particles. However,
the gas temperature in electron tubes is generally low, and this process
is normally unimportant.
REFERENCE
1. Millman, J., and S. Seely, "Electronics," Chap. V, McGraw-Hill Book Company,
Inc., New York, 1941.
8 ELECTRON- TUBE CIRCUITS [CHAP. 1

PROBLRMS
1-1. A tungsten filament, 0.0085 in. diameter, 3 1 YI 6 in. long, is operated at
2650°K. What is the temperature-limited current? If the temperature is
increased by 50o K, by what percentage does the emission current increase?
1-2. The filament of an FP-400 tungsten-filament tube is 1.25 in. long and
0.005 in. in diameter. If the total emission current is 30 ma, at what temperature
is the filament operating?
1-3. A simple inverted-V oxide-coated cathode is made of tungsten ribbon
0.125 by 0.020 in. and is 1.4 in. long. It is maintained at a temperature of
1I00°K. What is the thermionic-emission current?
1-4. An oxide-coated emitter is operating at 1I00°K. Calculate the relative
thermionic-emission currents if bo has the value 12,000; the value 11,000.
1-5. At what temperature will a tho ria ted-tungsten filament give as much
current as a tungsten filament of the same dimensions which is maintained at
2650 0 K?
1-6. At what temperature will an oxide-coated cathode give the same emission
as a thoriated-tungsten filament of the same physical dimensions which is main-
tained at 17500 K?
1-7. Monochromatic light of wave length 5893 A falls on the following surfaces:
a. Cesium, with a work function 1.8 volts.
b. Platinum, with a work function 5.3 volts.
Is photoelectric emission possible in both cases? Explain.
1-8. A PJ-22 vacuum photocell is to be used to sound an alarm when the light
at a given region of a room falls below 40 ft-c or increases above 120 ft-c. What
are the corresponding photocurrents? A collecting potential of 45 volts is used.
 CHAPTER 2

CHARACTERISTICS OF ELECTRON TUBESl

CONSIDER an isolated thermionic source situated in a vacuum. This

cathode will emit electrons, most of which have very little energy when
they emerge. Those electrons that first escape will diffuse throughout
the space within the envelope. An equilibrium condition will soon be
reached when, because of the mutual repulsion between electrons, the
free electrons in the space will prevent any additional electrons from
leaving the cathode. The equilibrium state will be reached when the
space charge of the electron cloud produces a strong enough electric field
to prevent any subsequent emission.
The inclusion of a collecting plate near the thermionic cathode will
allow the collection of electrons from the space charge when this plate is
maintained at a positive potential with respect to the cathode; the higher
the potential, the higher the current. Of course, if the thermionic emis-
sion is limited, then the maximum current possible is the temperature-
saturated value.
In addition to such a simple two-element device, which is the diode,
grids may be interposed between the cathode and plate. If a single grid
is interposed, the tube is a triode. If two grids are present, the tube is a
tetrode; three grids yields a pentode, etc. Details of the characteristics
and operation of such devices will be considered in some detail in the
following pages.

THE HIGH-VACUUM DIODE

2-1. The Potential Distribution between the Electrodes. Consider a
simple diode consisting of a plane cathode and a collecting plate, or anode,
which is parallel to it. It is supposed that the cathode can be heated to
any desired temperature and that the potential between the cathode and
anode may be set at any desired value. It is desired to examine the
potential distribution between the tube elements for various cathode
temperatures and fixed anode-cathode applied potential.
Suppose that the temperature of the cathode is high enough to allow
some electrons to be emitted. An electron space-charge cloud will be
formed in the envelope. The density of the electrons and the potential
9
10 ELECTRON - TUBE CIRCUITS [CHAP. 2

at any point in the interelectrode space are related by Poisson's equation

d2E p
(2-1)
dx 2 - ~

where E is the potential in volts, p is the magnitude of the electronic-

charge density, in coulombs per cubic meter, and E = 10-9 /3611" is the
permittivity of space. A study of
this expression will yield significant
,,
information.
,,,
Q)
,,
It is supposed that the electrons
,,' that are emitted from the cathode
g
<e:,
r:b have zero initial velocities. Under
,, these conditions, the general char-
, acter of the results will have the
-L forms illustrated in Fig. 2-1. At the
----------£i---------- /, temperature T I , which is too low for
FIG. 2-1. The potential distribution
between plane-parallel electrodes, for
any emission, the potential distribu-
several values of cathode temperature. tion is a linear function of the dis-
tance from the cathode to the anode.
This follows from Eq. (2-1), since, for zero-charge density,
./

dE
or -dx = const

This is the equation of a straight line.

At the higher temperature T 2 , the charge density p is not zero. Clearly,
the anode-cathode potential, which is externally controlled, will be
independent of the temperature, and all curves must pass through the
fixed end points. Suppose that the potential distribution is somewhat
as illustrated by the curve marked T 2• All curves must be concave
upward, since Eq. (2-1), which may be interpreted as a measure of the
curvature, is positive. A positive curvature means that the change in
slope dE/dx between two adjacent points must be positive. Moreover,
the curvature is greater for larger values of p, corresponding to the higher
temperatures. It is possible to justify that the maximum current that
can be drawn from the diode for a fixed plate voltage and any temperature
is obtained under the condition of zero electric field at the surface of the
cathode. Under these optimum conditions

_ dE = 0 at x=o (2-2)
dx
This condition is valid under the assumption of zero initial velocities of
emission of the electrons.
SEC. 2-2] CHARACTERISTICS OF ELECTRON TUBES 11

2-2. Equations of Space Charge. An explicit relation between the

current collected and the potential that is applied between the anode
and cathode is possible. In general, the current density is a measure
of the rate at which the electrons pass through unit area per unit time
in the direction of the field. If v denotes the drift velocity in meters
per second, N is the electron density in electrons per cubic meter, ande
is the electronic charge in coulombs, then the current density in amperes
per square meter is
J = Nev = pv (2-3)
Also, neglecting the initial velocity, the velocity of the electron at any
point in the interelectrode space is related to the potential through which
it has fallen by the following expression, which is based on the conserva-
tion of energy:
Y2mv 2 = eE (2-4)
By combining the foregoing expressions, there results
d 2E JE-'/J
(2-5)
dx 2 - Eo(2e/m)'/J
This is a differential equation in E as a function of x. The solution of it
is given by
amp/m 2 (2-6)

For electrons, and in terms of the boundary conditions, there results

J = 2.33 X 10- 6 ~b~ amp/m 2 (2-7)

This equation is known as the Langmuir-Childs or three-halves-power

law. It relates the current den-
ib
sity, and so the current, with the Space charge
[K' /tinifed
applied potential and the geom-
etry of the tube. It shows that
the space-charge current is inde- ,_ Temperafure
pendent of the temperature' and / lim/fed
the work function of the cathode.
Thus, no matter how many elec-
trons a cathode may be able to eb
supply, the geometry of the tube FIG. 2-2. The volt-ampere characteristics
and the applied potential will de- of a typical diode.
termine the maximum current that can be collected by the anode. If the
electron supply from the cathode is restricted, the current may be less
than the value predicted by Eq. (2-7). The conditions are somewhat as
represented graphically in Fig. 2-2.
12 ELECTRON- TUBE CIRCUITS [CHAP. 2

For the case of a tube that possesses cylindrical symmetry, a similar

analysis is possible. The results of such a calculation lead to the follow-
ing expression for the current,

h = 14.6 X 10-6 i E{3b~ amp (2-8)

ra
where l is the active length of the tube and {32 is a quantity that is deter-
mined from the ratio ru/rk, the ratio of anode to cathode radius. For
ratios ra/rk of 8 or more, {32 may be taken as unity.
2.4 7ype8/
_ 2.0 L/ Oxide-/.4
7ype/O
"E!? 1.6 Th.W-/.J /~ /'
// V
""'
.'? I. 2 TyperPS5
~0.8 /. V L W- !J
o
~ L
i°.4 /
V
-0.4°
o 0.5 1.0 1.5 2.0 2.5 3.0
I0910Eb (log1ovolts)
FIG. 2-3. Experimental results to verify the three-halves power law for tubes with
oxide-coated, thoriated tungsten, and pure tungsten filaments.

Attention is called to the fact that the plate current depends upon the
three-halves power of the plate potential both for the plane parallel and
also for a diode possessing cylindrical symmetry. This is a general
relationship, and it is possible to demonstrate that an expression of the
form Ib = kEi" applies for any geometry, provided only that the same
restrictions as imposed in the above developments are true. The specific
value of the constant k that exists in this expression cannot be analytically
determined unless the geometry of the system is specified.
The dependence of the current on the potential for any tube may be
determined by plotting the results obtained experimentally on a loga-
rithmic scale. Theoretically one should find, if the expression Ib = kEi'.!
is valid, that
(2-9)
The logarithmic plots for three commercial tubes are shown in Fig. 2-3.
The type 10 tube is a triode and was converted into a diode by connecting
grid and plate together. The other tubes are diodes. It will be observed
that the logarithmic plots are straight lines, although the slopes of these
lines are all slightly less than the theoretical 1.5.
2-3. Rating of Vacuum Diodes. The current and potential ratings of
a diode, i.e., the maximum current that the tube may carry and the maxi-
SEC. 2-31 CHARACTERISTICS OF ELECTRON TUBES 13
mum potential difference that may be applied between anode and cathode,
are influenced by a number of factors.
1. A limit is set to the tube current by the cathode efficiency of the
emitter. Thus, for a given input power to the filament, a maximum cur-
rent is specified.
2. There is a maximum temperature limit to which the glass envelope
of the tube may be safely allowed to rise. This is the temperature to
which the tube ,vas raised during the outgassing process. This is about
400°C for soft glass and about 600°C for pyrex. For higher temperatures,
the gases adsorbed by the glass walls may be liberated. Owing to this
limitation, glass bulbs are seldom used for vacuum tubes of more than
about 1 kw capacity.
3. A very important limitation is set by the temperature to which the
anode may rise. In addition to the fraction of the heat radiated by the
cathode that is intercepted by the anode, the anode is also heated by
the energy carried by the anode current. The instantaneous power
carried by the anode current and supplied to the anode is given by cbi b,
where Cb is the anode-cathode potential and ib is the anode current. The
temperature to which the anode rises will depend upon the area of the
anode and the material of its construction.
The most common metals used for anodes are nickel and iron for
receiving tubes and tantalum, molybdenum, and graphite for transmit-
ting tubes. The surfaces are often roughened or blackened in order to
increase the thermal emissivity. The anodes of many transmitting tubes
may be operated at a cherry-red heat without excessive gas emission.
To allow for forced cooling of the anode, cooling coils may be provided, or
the tube may be immersed in oil. The newer type of transmitting tubes
are frequently provided with radiator fins for forced-air cooling. Several
different types of transmitting tubes are illustrated in Fig. 2-4.
4. The voltage limitation of a high-vacuum diode is also dependent
on the type of its construction. If the filament and anode leads are
brought out side by side through the same glass press, some conduction
may take place between these leads through the glass. This effect is
particularly marked if the glass is hot, and the resulting electrolysis will
cause the glass to deteriorate and eventually to leak. The highest
voltage permissible between adjacent leads in glass depends upon the
spacing and upon the type of glass but is generally kept below 1,000 volts.
Higher voltage tubes are usually provided with filament leads at one end
of the glass envelope, with the anode at the other end.
The glass envelope must be long enough so that flashover on the outside
of the tube will not occur. In a diode as a rectifier, no current will flow
during the time that the anode is negative with respect to the cathode.
14 ELECTRON-TUBE CIRCUITS [CHAP. 2

The maximum safe rating of a rectifying diode is known as the peak

inverse voltage rating.
Commercial vacuum diodes are made which will rectify current at high
voltages, up to 200,000 volts. Such units are used with X-ray equip-
ment, with high-voltage cable-testing equipment, and with the high-

FIG. 2-4. Photographs of two transmitting tubes. (RCA Mfg. Co.)

voltage equipment for nuclear-physics research. The dimensions and

shape of the glass envelope will depend upon the current capacity of the
tube and the type of cooling to be used, oil-cooled tubes being generally
smaller than air-cooled types.

THE TRIODE
2-4. The Grid. The introduction of a third element between the
cathode and plate of the diode by DeForest in 1907 was the start of
the extensive developments involving vacuum tubes. This new elec-
trode, called the control grid, consists of a wire mesh, or screen, which
surrounds the cathode and is situated close to it. The potential applied
SEC. 2-4] CHARACTERISTICS OF ELECTRON TUBES 15

to the grid in such a tube is usually several volts negative relative to the
cathode, whereas the plate is usually maintained several hundred volts
positive with respect to the cathode. Clearly, the electric field resulting
from the potential of the grid tends to maintain a large space-charge
cloud, whereas the field of the plate tends to reduce the space charge.
However, owing to its proximity to the cathode, a given potential on the
grid will exercise a greater effect on the space charge than the same poten-
tial on the plate. This would seem to imply that a strict proportionality
should exist between the relative effectiveness of the grid and plate
potentials on the space charge and that the plate current should be
represented approximately by the equation

ib = k (e c + ~)% (2-10)

where eb is the plate-cathode potential, ec is the grid-cathode potential,

and the factor J.L is a measure of the relative grid-plate potential effective-
ness on the tube current. The factor J.L is known as the amplification
factor of the grid.
The validity of Eq. (2-10), which is simply a natural extension of the
three-halves-power space-charge equation of the diode, has been verified
experimentally for many triodes. No simple, rigorous theoretical deriva-
tion of this equation is possible, even for a triode of relatively simple
geometry. However, the value of the amplification factor I-' can be
calculated with a fair degree of accuracy from equations that are based
on electrostatic considerations.
By maintaining the grid at some negative potential with respect to the
cathode, it will repel electrons and will, in part, neutralize the attractive
field of the anode, thus reducing the anode current. If the grid potential
is made positive, the electron
stream will increase because of the
combined action of both the grid
and the plate potentials. But,
with a positive potential on the
grid, some of the space charge will
be attracted to it, and a current in
the grid will result. The grid
structure must be designed to dis-
FIG. 2-5. Total space, plate, and grid
sipate the grid power if the grid current in a triode, as a function of grid
potential is to be maintained posi- voltage, with fixed plate voltage.
tive; otherwise the grid structure
may be seriously damaged. Generally the grid is maintained negative,
although positive-grid triodes for power-amplifier applications are
available.
The variations of the plate and grid currents with variations of grid
16 ELECTRON-TUBE CIRCUITS [CHAP. :2

potential are illustrated in Fig. 2-5. In this diagram, the plate potential
is maintained constant. For sufficiently negative grid potential, cutoff
of the plate current occurs. As the grid potential is made less negative,
the plate current follows a smooth curve, the variation being expressed
analytically by Eq. (2-10). As the grid potential is made positive, grid
current flows, the magnitude of this current increasing rapidly with
increasing grid potential.

FIG. 2-6. The plate characteristics of a FIG. 2-7. The transfer characteristics of
triode. a triode.

For positive grid potentials, and with the consequent grid current,
Eq. (2-10) no longer'represents the plate current, although it does give a
good representation of the total space current. With increasing grid
potentials, the grid current increases, and the plate current decreases.
2-5. Triode Parameters. In view of Eq. (2-10), the dependence of the
plate current on the plate and the grid potentials may be expressed
functionally by the expression
ib = f(eb,e c ) (2-11)
Of course the plate current also depends upon the heater temperature,
but as the heater current is usually maintained at rated value (this is such
as to provide perhaps five to ten times
the normal required current), this term
usually does not enter into the functional
relationship. If Eq. (2-11) is plotted on
a three-dimensional system of axes, a
space diagram representing the function
j(ib,eb,e c) = 0 is obtained. The projec-
L------=""""'~~'-----e6 tions of these surfaces on the three co-
FIG. 2-8. The constant-current ordinate planes give three families of
eharacteristics of a triode. characteristic curves. These curves are
given in Figs. 2-6, 2-7, and 2-8.
The curves of Fig. 2-6 are known as the plate characteristics since they
show the variation of the plate current with plate voltage for various
SEC. 2-5] CHARACTERISTICS OF ELECTRON TUBES 17

values of grid bias. The main effect of making the grid more negative
is to shift the curves to the right, without changing the slopes appreciably.
This is in accord with what would be expected from consideration of
Eq. (2-10).
If the grid potential is made the independent variable, the mutual, or
transfer, characteristics of Fig. 2-7 result. The effect of making the plate
potential less positive is to shift the curves to the right, the slopes again
remaining substantially unchanged.
The simultaneous variation of both the plate and the grid potentials
so that the plate current remains constant gives rise to a third group of
characteristics illustrated in Fig. 2-8. These show the relative effects
of the plate and grid potentials on the plate current of the tube. But
from the discussion of Sec. 2-4 it is the amplification factor that relates
these two effects. Consequently, the amplification factor is defined as
the ratio of the change in plate voltage to the change in grid voltage for a
constant plate current. Mathematically, J.I. is given by the relation

aeb) (2-12)
J.I. = - ( ae e Ib
The negative sign takes account of the fact that a decreasing grid poten-
tial must accompany an increasing plate potential if the plate current is
to remain unchanged.
Consider the variation in the plate current. This is obtained by
expanding Eq. (2-11) in a Taylor's expansion. But it is here assumed
that the variation is small and that it is adequately represented by the
first two terms of the expansion. Subject to this limitation, the expres-
sion has the form
(2-13)

This expression indicates simply that changes both in the plate voltage
~eb and in the grid voltage ~ec will cause changes in the plate current.
The quantity (aebjaib)E, expresses the ratio of an increment of plate
potential to the corresponding increment of plate current, for constant
Ec. This ratio has the units of resistance, is known as the plate resistance
of the tube, and is designated by the symbol rp. Clearly, rp is the slope of
the plate characteristics of Fig. 2-6 and has been indicated there.
The quantity (aib/aechb) which gives the ratio of an increment of plate
current to the corresponding increment of grid potential for constant
plate potential E b, has units of conductance. It is known as the plate-
grid transconductance or mutual conductance and is designated by the
symbol gm. The mutual conductance gm is the slope of the mutual-, or
transfer-, characteristic curves of Fig. 2-7. .
18 ELECTRON - TUBE CIRCUITS [CHAP. 2

To summarize, the triode coefficients have the forms

plate resistance

mutual conductance (2-14)

amplification factor

It is easy to show that J.L is related to rp and gm by the expression

(2-15)
This is obtained by setting t:.ib = 0 in Eq. (2-13) and then using the defini-
tions of Eq. (2-14).
The variations of these parameters for a fixed value of plate potential
for the 6C5 tube are shown in Fig. 2-9. It is noticed that the plate
c resistance varies over rather wide
,g ... 22 limits, being very high at zero
~ 0
~ 1>20 ~-p-- 3000~ plate current, and approaches a
o..~
E 18 ~ constant value at the higher plate
~ <>
'E currents. The transconductance
2000 ~'
c varies from a very small value at
13 zero plate current and tends to-
~
1000 5u
ward a constant value at the
2 V)
<: higher plate currents. The am-
<S lS
c: t-!: plification factor remains reason-
o 0
o 10 20 ably constant over a wide range
PIOIte current, mOl
of currents, although it falls off
FIG. 2-9. The parameters /J., T p , and gm of rapl'dl y a t th e I ow curren t s. The
a 605 triode as a function of plate current.
corresponding values for other
values of Eb may differ numerically, but the general variations will be
similar.
High-power triodes are used extensively in tranl3mitters. The grid of
such a tube is driven positive with respect to the cathode during part
of the cycle, and the current is cut off during part of the cycle. The
characteristics of importance of such tubes are the plate curves and the
constant-current curves. The variations over normal operating limits
are as illustrated in Figs. 2-10 and 2-11 for a type 889A tube.
MULTIELECTRODE TUBES
2-6. Tetrodes. In the tetrode a fourth electrode is interposed between
the grid and the plate. This new electrode is known as the screen grid, or
grid 2, in order to distinguish it from the "control" grid of the triode.
SEC. 2-6] CHARACTERISTICS OF ELECTRON TUBES 19
AVERAGE PLATE CHARACTERISTICS

Ef = II VOLTS A.C.

8
C\I

o
o
0,.....

j
0>
o
o
Ow
<XlI-
<I:
...J
no

o
o
o
10

o
o
o
'I"

o
o
o
C\I

GO 10 'I" o
PLATE AMPERES
FIG. 2-10. The plate characteristics of a power-triode type 889A.
20 ELECTRON - TUBE CIRCUITS [CHAP. 2

AVERAGE CONSTANT-CURRENT CHARACTERISTICS

itJ±IJN:l.IIIL rm:r-
]1 I:j:,.,
~ E1' - II VOLTS A.C.
:J GRID AMPERES - Ie
fj, PLATE AMPERES =Ib
'~i' ,- I-l"r' ~t

JJ tfi-I1} :t ~it
rU ~:

.T -,I: .

f-H

'=iT
~=+ ~,

H-J.

it-
H-

I-~'
-i -

!rj:f f

_I
:-1~

o 0
o
o .0
00
GRID VOLTS (Ee ) I

FIG. 2-11. The constant-current characteristics of the power triode of Fig. 2-10.
SEC. 2-7] CHARACTERISTICS OF ELECTRON TUBES 21
Physically, it almost entirely encloses the plate. Because of its design
and disposition, the screen grid affords very complete electrostatic shield-
ing between the plate and the control grid. This shielding is such that
the grid-plate capacitance is reduced by a factor of about 1,000 or more.
However, the screen mesh does not interfere appreciably with the electron
flow. The reduction of the grid-plate capacitance is a very important
improvement over the triode, and this matter will be considered in some
detail in Chap. 3.
Because of the electrostatic shielding of the plate by the screen, the
potential of the plate has almost no effect in producing an electric field
at the cathode. Since the total space current is determined almost
wholly by the field near the cathode surface, the plate exerts little or no
effect on the total space charge drawn from the cathode. There is,
therefore, a significant difference between the triode and the tetrode. In
a triode, the plate performs two distinct functions, that of controlling the
total space current, and that of collecting the plate current. In a
tetrode, the plate serves only to collect those electrons that have passed
through the screen.
The passive character of the plate makes the tetrode a much better
voltage amplifier than the triode. This follows from the fact that in the
triode with a resistance load an increase in load current is accompanied
by a decreased plate-cathode potential, which results in a decreased space
current. In the tetrode, the decreased plate-cathode potential still
exists, but owing to the secondary role of the plate the space current is
not materially affected.
The disposition of the cathode and the control grid is nearly the same
in both the tetrode and the triode, and therefore the grid-plate trans-
conductance is nearly the same in both tubes. Also, the plate resistance
of the tetrode is considerably higher than that of the triode. This fol-
lows from the fact that the plate voltage has very little effect on the plate
current. Thus, with the high plate resistance and with a gm that is
about the same as for the triode, the tetrode amplification factor is very
high.
2-7. Tetrode Characteristics. In the tetrode with fixed control-grid
and screen-grid potentials, the total space current is practically constant.
Hence, that portion of the space current which is not collected by the
plate must be collected by the screen; where the plate current is large,
the screen current must be small, and vice versa. The general character
of the results is illustrated in Fig. 2-12.
Although the plate potential does not affect the total space current to a
very great extent (although a slight effect is noted in the curve at the
lower plate potentials), it does determine the division of the space current
between plate and screen. At zero plate potential, few of the electrons
22 ELECTRON - TUBE CIRC UITS [CHAP. 2

have sufficient energy to reach the anode, and the plate current should be
small. As the plate potential is increased, a rapid rise occurs in the plate
current, with a corresponding reduction of the screen current. When the
plate potential is larger than the screen potential, the plate collects
almost the entire space current and the screen current approaches zero
or a very small value.
An inspection of the curves of Fig. 2-12 shows that the plate current
rIses very rapidly with increasing plate potential, but this increase is
followed by a region of plate-poten-
tial variation in which the plate
current decreases with increasing
I
I plate potential. This region is one
" : of negative plate resistance, since
Ec:k--_______ ~c~ an increasing plate potential is ac-
L----::.:.J.------~eb companied by a decreasing plate
FIG. 2-12. Curves of total space current, current. The kinks, or folds, in the
plate current, and screen current, in a curves are caused by the emission of
tetrode.
electrons from the plate by the
process of secondary emission. This results from the impact of the pri-
mary electrons with the plate. That is, secondary electrons will be
released from the anode, and if this is the electrode with the highest posi-
tive potential, the electrons will be collected by the anode, without any
noticeable effect. If, however, secondary electrons are liberated from the
anode, and if these electrons are collected by some other electrode, then
the anode current will decrease, whereas the current to the collecting elec-
trode wiII increase. It is this latter situation which exists in the tetrode
when the plate potential is low and the
screen is at a high potential.
When the plate potential is higher than ISO
the screen potential, the secondary elec-
trons from the plate are drawn back, with- SO
100
out appreciable effect. If under these
potential conditions secondary electrons
are liberated from the screen, these will be
collected by the anode. The correspond- ec
ing plate current will be greater than that FIG. 2-13. The transfer char-
in the absence of secondary emission from acteristics of a tetrode, for a fixed
the screen. screen potential, and with the
plate potential as a parameter.
2-8. Transfer Characteristics. Sin c e
the plate of a tetrode has no appreciable influence on the space current, it
is expected that the cathode, the control grid, and the screen grid should
possess characteristics not unlike those of a triode. This is actually the
case, as illustrated in Fig. 2-13. These curves show the effect of varia-
SEC. 2-9] CHARACTERISTICS OF ELECTRON TUBES 23
tions of plate potential on plate current, for fixed E c2 • Because of the
slight influence of the plate, the transfer curves are bunched together.
These curves should be compared with those of the triode in Fig. 2-7,
where the transfer curves are widely separated.
The transfer curves for plate voltages below the screen potential, and
this is the region of operation which is generally avoided in practice,
become separated. In fact, the transfer characteristic for Eb = 100
volts actually falls below that for Eb = 50 volts. This anomalous
behavior is directly the result of the secondary-emission effects discussed
above.
2-9. Tube Parameters. It is expected, on the basis of the foregoing
discussion, that the plate current may be expressed as a function of the
potential of the various electrodes by an expression of the form
(2-16)

where eel is the potential of the first, or control, grid, ec 2 is the potential of
the second, or screen, grid, and eb is the potential of the plate, all with
respect to the cathode. This functional relationship is just a natural
extension of that which applies for triodes. In fact, an approximate
explicit form of the dependence is possible. This form, which is an
extension of Eq. (2-10), may be written as

. _ k ( eel
tb -
+"e~ + e )%
- -
C2
(2-17)
J.l.l J.l.2
where J.l.l and J.l.2 are the control-grid and screen-grid amplification factors,
respectively.
The variation in the plate current, second- and higher-order terms in the
Taylor expansion being neglected, is given by

(2-18)

Generally, the screen potential is maintained constant at some appro-

priate value, and hence ~ee2 = O. The third term in the expansion may
be omitted under these conditions. The partial-differential coefficients
appearing in this expression furnish the basis for the definitions of the
tube parameters. These are

plate resistance

mutual conductance (2-19)

amplification factor
24 ELECTRON- TUBE CIRCUITS [CHAP. 2

The two subscripts associated with each term indicate the parameters
that are maintained constant during the partial differentiation. It can
be shown that here too the relation }.t = rplJm is valid. Nominal values
for the various parameters that appear in this relationship are rp = 10 5 to
2 X 106 ohms, IJm = 500 to 3,000 }.tmhos, and }.t = 100 to 1,200.
2-10. Pentodes. Although the insertion of the screen grid between the
control grid and the anode in a triode serves to isolate the plate circuit
from the grid circuit, the range of operation of the tube is limited owing to
the effects of secondary emission. This limitation results from the fact
that, if the plate-potential swing is made too large, the instantaneous
plate potential may extend into the region of rapidly falling plate current,
with a resulting marked distortion in the output.

Tefrode
Pentacle
EC2 -cons/anf

L---------------------- e6 ----~----~L-----~ec

FIG. 2-14. The characteristics of a tube FIG. 2-13. The transfer curves of a pen-
when connected as a tetrode and as a tode for a fixed screen potential and with
pentode. the plate potential as a parameter.

The kinks, or folds, that appear in the plate-characteristic curves and

that limit the range of operation of the tetrode may be removed by
inserting a coarse suppressor-grid structure between the screen grid and
the plate of the tetrode. Tubes that are provided with this extra grid are
known as pentodes. The suppressor grid must be maintained at a lower
potential than the instantaneous potential reached by the plate at any
time in its potential excursions. Usually the suppressor is connected to
the cathode, either externally or internally. Now since both the screen
and the anode are positive with respect to the suppressor grid, secondary
electrons from either electrode will be returned to the emitting electrode.
The main electron stream will not be materially affected by the presence
of the suppressor grid. The effects of the insertion of the suppressor grid
are shown graphically in Fig. 2-14.
The pent ode has displaced the tetrode in radio-frequency (r-f) voltage
amplifiers, because it permits a somewhat higher voltage amplification at
moderate values of plate potential. Likewise it permits a greater plate-
voltage excursion without distortion. Tetrodes are used extensively in
high-power tuned amplifiers.
The transfer curves of a pent ode are shown in Fig. 2-15. It is noted
that the curves are almost indepe;ndent of the plate voltage.
SEC. 2-13] CHARACTEHISTICS OF ELECTRON TUBES 25
2-11. Remote-cutoff Tubes. If in a pentode the grid-cathode spacing,
the spacing between grid wires, or the diameter of the grid wires is not
uniform along the entire length of the control-grid structure, the various
portions of the grid will possess different degrees of electrostatic control
over the plate current. That is, one portion of the grid may cause
electron-f.ow cutoff, ,vhereas an appreciable current might pass through a
more widely spaced section of the grid.
As a result, the plate-current control by
the grid is considerably less effective than
in a conventional pentode. The general
character of the results is illustrated in
Fig. 2-16. Owing to its construction, a
given grid-voltage increment results in
a platc-current change that is a function
of the bias. This means that the mutual FIG. 2-16. The transfer curves of a
conductance is a function of the bias. 6SJ7 sharp-cutoff pentode and a
For this reason, these tubes are called 6SK7 remote-cutoff pentode.
varia/)le-rnu tubes. They are also known
as rernote-eutoff and supercontrol tubes. They have applications in radio
receivers and may be used in f-m transmitters. These applications will
be considered in later chapters.
2-12. Hexodes, Heptodes. A number of special-purpose tubes con-
taining more grid elements than the pentode are used extensively.
These tubes possess a wide variety of characteristics, depending upon the
grids to which fixed potentials are applied and those to which signals
might be applied. These tubes are used extensively as converters in
superheterodyne receivers and find f-m transmitter and other applica-
tions. More will be said about these in the sections having to do with
such applications.
2-13. Beam Power Tubes. The suppressor grid is introduced into the
pentode in order to extend the range of operation of these tubes beyond
that of the tetrode. These tubes are quite satisfactory over wide limits,
and the range of operation is limited when the instantaneous plate
potential falls to the rapidly falling plate-current region at low potentials.
This rapid change in plate current for small changes in plate potential
in the region of low plate voltage results from the overeffectiveness of the
suppressor grid at these low plate potentials.
Because of this, the shape of the suppressor grid in some modern
pentodes has been so dimensioned that the effects of secondary emission
are just suppressed or only admitted slightly at the low anode voltages.
This results in an improved plate characteristic and is manifested by a
~harper break in the plate characteristic.
26 ELECTRON -TUBE CIRCUITS [CHAP. 2

The pentode or tetrode beam power tube waB deBigned with these
considerations specifically III mind, and the plate characteristics are
illustrated in Fig. 2-17. The essen-
Beam power lube tial features of the beam power tube
are illustrated in the schematic view
of Fig. 2-18. One feature of the de-
sign of this tube is that each spiral
turn of the screen is aligned with a
'-------------el> spiral turn of the control grid. This
FIG. 2-17. The plate characteristic of a serves to keep the screen current
beam power tube and a power pentode. small and hence leaves the plate
current virtually unchanged. Other
features are the flattened cathode, the beam-forming side plates (main-
tained at zero potential), the shape of the plate, the curvature of the

FIG. 2-18. Schematic view of the shapes and arrangement of the electrodes in a beam
power tube. (RCA Mfg. Co.)

grids, and the spacing of the various elements. As a result of these

design characteristics, the electrons flow between the grid wires toward
the plate in sheets, or beams.
The region between the screen and the plate possesses features which
are somewhat analogous to those which exist in the space-charge-limited
diode. That is, there is a flow of charge between two electrodes. How-
SEC. 2-141 CHARACTERISTICS OF ELECTRON TUBES 27
ever, the electrons, when they enter this region do so with an appreciable
velocity. For such a case as this, the considerations of Sec. 2-1 would
have to be modified to take account of the initial velocity. If this is
done, it is found that a potential minimum will exist in the region between
the two electrodes. This potential minimum acts as a virtual suppressor
grid, and any secondary electrons that are emitted from either the plate
or the screen are returned to the emitting electrode.
The actual potential distribution in the screen-plate region will depend
on the instantaneous plate potential and the plate current, for a constant
screen potential. The resulting variable suppressor action proves to be
superior to that possible with a mechanical grid structure, as illustrated.
GAS TUBES
2-14. Electrical Discharge in Gases. There are two important types
of discharge in gases that play roles in electron tubes. One of these is
the glow discharge, and the second is the arc discharge. The glow dis-
charge utilizes a cold cathode and is characterized by a fairly high tube
drop and a low-current-carrying capacity. The voltage drop across the
tube over the operating range is fairly constant and independent of the
current. The arc discharge is characterized by a low voltage drop and a
high current capacity. For an arc tube with a thermionic cathode, the
temperature-limited cathode emission may be drawn with a tube drop
approximately equal to the ionization potential of the gas. For a mer-
cury-pool cathode, extremely high cur-
rent densities exist (of the order of ib
5 X 10 8 amp/m 2), with high currents
possible and a tube drop approximately
equal to the ionization potential of the
mercury atom.
Consider a gas tube which consists of
a cold cathode and a collecting anode,
~--------~------~e~
between which is connected a source of FIG. 2-19. Volt-ampere character-
potential through a current-limiting istic of a glow discharge.
resistor, and an indicating ammeter.
The volt-ampere characteristic of such a tube has the form illustrated in
Fig. 2-19. This curve shows that breakdown occurs at a potential which
is somewhat higher than the maintaining potential but that there is a
region where the tube drop remains substantially constant over an
appreciable range of currents. Visually, the discharge is characterized
by a colored luminous region, the color being a function of the gas present
in the tube.
It is desired to explain the mechanism of operation of these tubes.
Consider, therefore, that a free electron exists within the tube; such an
28 ELECTRON - TUBE CIRCUITS [CHAP. 2

electron might have been released by ionization due to collision between

a gas molecule and a cosmic ray or by photoelectric emission. With the
application of the potential between the electrodes, the electron will drift
toward the anode. If the field is large enough, the electron may acquire
enough energy to ionize a molecule when it collides with it. N ow two
electrons will be present, the original one and also the electron that has
been liberated by the process of ionization, and a positive ion. The two
electrons and the positive ion will move in the applied field, the electrons
moving toward the anode, and the positive ion toward the cathode. If
the field is large enough, the resulting cumulative ionization may continue
until breakdown occurs. Once breakdown occurs, the potential distribu-
tion within the tube is markedly modified, and most of the region of the
discharge becomes virtually equipotential or force-free, containing as
many positive as negative charges. This is the plasma of the discharge.
Almost the entire potential change occurs in the very narrow region near
the cathode. Normal values for cathode-fall voltage range between
about 59 volts (a potassium surface and helium gas) and 350 volts. The
presence of a low-work-function coating on the cathode will result in a
low cathode fall with any gas. Also, the use of one of the inert gases
(helium, neon, argon, etc.) results in a low cathode fall with any cathode
material. The cathode fall adjusts itself to such a value that each posi-
tive ion, when it falls through this field, will release an electron from the
cathode by secondary emission. The positive ion combines with this
electron and thus becomes neutralized.
Another feature of a normal glow discharge is that the current density
of the cathode remains sensibly constant. For higher currents, a greater
portion of the cathode is covered with glow, the area of the glow on the
cathode increasing directly with the magnitude of the current. Once
the cathode is completely covered with glow, any further current through
the tube depends on an excess of secondary emission from the cathode
over that required to neutralize the positive ions. This is accompanied
by a rising cathode fall. This is the "abnormal" glow and is generally
of small practical importance.
The dividing line between an arc and a glow discharge is rather indis-
tinct. The arc discharge allows for the passage of large currents at low
voltage, the current density at the cathode being high. Nevertheless
each discharge has associated with it the cathode fall, the plasma, and
the anode fall (which is of minor significance in both types of discharge).
The discharges differ in respect to the mechanism by which the electrons
are supplied from the cathode. In the glow discharge, as discussed, the
electrons are emitted from the cathode by the process of secondary
emission resulting from positive-ion bombardment of the cathode. In
the arc discharge, the emtssion of the electrons from the cathode occurs
SEC. 2-15] CHARACTERISTICS OF ELECTRON TUBES 29
through the operation of a supplementary mechanism other than by
positive-ion bombardment. In the thermionic arc, the electrons are
supplied by a cathode that is heated to a high temperature, either by the
discharge or externally by means of an auxiliary heating circuit. The
mechanism for electron release is not fully
understood in the arcs that employ a mercury-
Cafhode
pool cathode or an arc between metal surfaces.
However, in the discharges the primary func- ' •. -SlartJnq probe
tion of the gas is to supply a sufficient positive- ----- 'Anode
ion density to neutralize the electron space ---
charge. Because of this, the normal potential FIG. 2-20. Electrode
drop across an arc tube will be of the order of structure in a VR tube.
the ionization potential of the gas.
2-15. Glow Tube. A glow tube is a cold-cathode gas-discharge tube
which operates in the normal glow-discharge region. The voltage drop
across the tube over the operating range is fairly constant and independent
of the current. When the tube is connected in a circuit, a current-limiting
resistor must be used if serious damage to the tube is to be avoided.
One commercial type of tube consists of a central anode wire which is
coaxial with a cylindrical cathode, as illustrated in Fig. 2-20. The elec-

FIG. 2-21. Photographs of several low-capacity glow lamps. (General Electric Co.)

trodes are of nickel, the inner surface of the cathode being oxide-coated.
The cathode fall is sometimes lowered by sputtering some misch metal (an
alloy of cerium, lanthanum, and didymium) on the cathode. The gases
that are commonly used are neon, argon, and helium. The tubes con-
taining neon or helium usually contain a small amount of argon. The
presence of the argon lowers the starting voltage. These tubes are
available with normal output potentials of 75, 90, 105, and 150 volts and
bear th~ designations VR-75, VR-90, etc. The normal maximum current
30 ELECTRON - TUBE CIRCUITS [CHAP. 2

is 30 mao The starting probe that is attached to the cathode, as illus-

trated in Fig. 2-20, serves to lower the breakdown voltage of the tube.
Glow lamps are also available for pilot, marker, and test-lamp service.
Such tubes are available in several sizes from 725 to 3 watts capacity.
Photographs of these are given in Fig. 2-21.
2-16. Cold-cathode Triodes. A cold-cathode triode, or grid-glow tube,
contains three elements, the cathode, the anode, and a starter, or control,
anode. The control electrode is placed
400
close to the cathode. The spacing
~ of the electrodes is such that a dis-
~300
-- charge takes place from the cathode
1\
jo" 200 to the control electrode at a lower po-
tential than is required for a discharge
0.
cu
-gr: 100
"- 1'- r-... from the cathode to the anode. Once
« the control gap has been broken down,
however, it is possible for the discharge
o 0 25 50 75 100 125 150 175 200225 to transfer to the main anode. The
TrOlnsfer current, J.LOI
cathode-anode voltage that is required
FIG. 2-22. Transfer characteristic
of an RCA OA4G cold-cathode for this transfer to occur is a function
triode. of the transfer current, the current in
the control electrode-cathode circuit.
Such a "transfer," or "transition," characteristic is given in Fig. 2-22.
For zero transfer current, which means that the control electrode is not
connected in the circuit, the anode
voltage is equal to the breakdown
.,.. 1.0
voltage between cathode and
:I:
E
E
/
V
f?" 0.1
anode. It is observed from the
curve that the required anode-
~
V)

t
0-
/
/
cathode voltage falls rapidly as S 0.01
0.
the transfer current is increased.
An increased transfer current in-
g
>-
5 0.001
e
/
dicates the presence of greater
i oniz a tion. Regardless of the
OJ
~
0.0001
/
magnitude of the transfer current, o 20 40 60 80 100 120
temperAture of condensed mercury:C
however, the anode-cathode volt-
FIG. 2-23. Mercury-vapor pressure as a
age can never fall below the main- function of condensation temperature.
taining voltage for this gap. The
transfer characteristic approaches this sustaining voltage asymptotically.
2-17. Hot-cathode Gas-filled Diodes. These tubes are thermionic
cathode diodes in which there is an inert gas at low pressure or in which
mercury vapor is added. In the latter case a few drops of mercury are
added to the tube after evacuation. The pressure in the tube is then a
SEC. 2-18) CHARACTERISTICS OF ELECTRON TUBES 31
function of the mercury-vapor condensation temperature. The relation-
ship between the pressure and the temperature is shown in Fig. 2-23.
Under normal operating conditions, the temperature of the tube will be
15 to 20°C above that of the surroundings (ambient temperature).
As already discussed, the sole function of the gas in these tubes is to
provide ions for the neutralization of space charge, thus permitting the
current to be obtained at much lower voltages than are necessary III

FIG. 2-24. Commercial mercury-vapor diodes of different capacity. (RCA Mfg. Co.)

vacuum tubes. If more than saturation current is demanded by the

circuit, then gas amplification, resulting from positive-ion bombardment
of the cathode, will occur. Under these circumstances the cathode fall
increases. The tube drop should not be permitted to exceed the dis-
integration voltage of the cathode (about 22 volts for a mercury diode
with either oxide-coated or thoriated-tungsten cathodes); otherwise the
cathode may be seriously damaged by the positive-ion bombardment.
Two typical commercial mercury-vapor-filled diodes are illustrated in
Fig. 2-24.
2-18. High-pressure Gas Diodes. Diodes are available which contain
argon or a mixture of argon and mercury at a pressure of about 5 cm.
32 ELECTRON· TUBE CIRCUITS [CHAP. 2

The cathodes in such tubes consist of a short, heavy thoriated-tungBten

or oxide-coated filament and are located close to heavy graphite anodes.
These tubes, which are known as tungar or rectigon tubes, are used exten-
sively in chargers for storage batteries. •
The presence of the fairly high pressure gas serves a twofold purpose.
One is to provide the positive ions for reducing the space charge. The
second is to prevent the evaporation of the thorium or the coating from
the filament. This second factor is extremely important since the fila-
ment is operated at higher than normal temperature in order to provide
the large currents from such a simple cathode structure. The high-
pressure gas in such a tube imposes a limitation on these tubes, and they
are limited to low-voltage operation.
2-19. The Thyratron. The thyratron is a three-electrode tube which
comprises the cathode, the anode, and a massive grid structure between
them. The grid structure is so designed as to pro-
vide almost complete electrostatic shielding between
the cathode and the anode. In such a tube as
this, the initiation of the arc is controlled by control-
ling the potential of the grid. The grid usually con-
sists of a cylindrical structure which surrounds both
.::\ the anode and the cathode, a baffle or a series of
j...- -Cathode baffles containing small holes being inserted between
---I
1 the anode and the cathode. The electrode structure
1
of such a tube is illustrated in Fig. 2-25. The
shielding by the grid is so complete that the appli-
FIG. 2-25. The elec- cation of a small grid potential before conduction is
trade structure of a started is adequate to overcome the field at the
negative control cathode resulting from the application of a large
thyratron.
anode potential.
Once the arc has been initiated, the grid loses complete control over the
arc. Grid control is reestablished only when the anode potential is
reduced to a value less than that necessary to maintain the arc. Once
the arc has been extinguished by lowering the plate voltage, the grid once
more becomes the controlling factor which 'determines when conduction
will again be initiated. That is, if the grid potential is more positive
than that necessary for the controlling action to prevail, conduction will
take place; if more negative, no conduction will occur. The curve that
relates the grid ignition potential with the potential of the anode for
conduction just to begin is known as the critical grid curve. In fact, a
knowledge of this static curve is all that is required to determine com-
pletely the behavior of a thyratron in a circuit.
Typical starting characteristic curves of mercury-vapor thyratrons are
given in Fig. 2-26. Two distinct types of characteristics are illustrated,
SEC. 2-19] CHARACTERISTICS OF ELECTRON TUBES 33

viz., those in which the grid potential must always be positive, and those
in which the grid is generally negative, except for very low plate poten-
tials. The physical distinction between these positive and negative
control tubes lies essentially in the more complete shielding by the grid
in positive control tubes.
In the negative control tube where the shielding is far less complete
than in the positive control type, the effect of the plate voltage is clearly

FG-27A Negcllive control FG-33 Positive control tube

80050040' 3Qo tube 80~500W
1000

~800
1\\1\ K
.$i.
~ 600
\ l\\ \
Q)

-g 400 I~ ~
~
6
c
<S
o 200
o
-10 -8
-6 -4 -2
'"
~~
0 2 4 6 8 10 12
D-c grid potentiClI CIt st~rt of clischClrge
FIG. 2-26. Critical grid characteristics of a positive and a negative control thyratron
for different temperatures.
seen; the higher the plate potential, the more negative must the grid
potential be in order to prevent conduction from taking plate. For low
plate potentials, positive grid potentials must be applied before ionization,
and hence conduction, can begin. If the plate potential is reduced still
more, even below the potential neces- 320
sary for ionization, breakdown can still
280
be obtained by making the grid suffi-
ciently positive. Now, however, the
function of the tube may be destroyed,
......."'" " 240 cr
2DO{
OJ

since the arc may take place between

the cathode and the grid, with very
i"-,
"-
"'" I 60 i;
120
0..
Q)

80 ]
little current to the plate. The thy- a...
....... 4D
ratron will be converted into a gas di- o
ode under these conditions, the plate -40 -36 -32 -28 -24 -20 -16 -12 -8 -4 o
acting as a dummy electrode the cylin- Grid Potential
" . ' FIG. 2-27. Critical grid character-
dncal gnd now servmg as the anode. istic of an 884 argon-filled thyratron.
It is because of this that a large current-
limiting resistor is connected in the grid circuit, as it is unwise to draw 8.
large grid current.
In addition to the mercury-vapor- and gas-filled thyratrons of moder-
ate current capacity, small argon-filled low-current-capacity tubes are
available. The shielding between the cathode and the anode is not so
complete in these tubes as in the higher current units. Also, the critical
34 ELECTRON - TUBE CIRCUITS [CHAP. 2

grid curves are independent of temperature, since the number of gas

molecules in the glass envelope remains constant. A typical critical
grid curve for an 884 is given in Fig. 2-27.
2-20. Shield-grid Thyratrons. Before breakdown of the tube occurs,
the current to the grid of a thyratron such as the FG-27A is a few tenths
of a microampere. Although this current is entirely negligible for many
applications, it will cause trouble in circuits that require very high grid

Shiel~ grid potentiOiI

900 +5 0 -2 -3 -4

800 \ \
Q) 700 ~ 1 \ \
tn
E 600 \ 1\ \
Shield
"grid
'0
> 500 \ \ CONDENSED
OJ
11400 1\ \ \ \ HG TEMP
\ 40°C
Confrol, ~=:;-;=-r1.:-~...,
grid
c
~ 300
\ \ \
6200 \ \
l'-- ~ "'-..
" - Cafhode 100
o-16 -12 -8 -4 0 4
""
Control griel voltage
--.....
8 1'2 16 20

FIG. 2-28. Electrode structure of the FIG. 2-29. Control characteristics of an

FG-98 shield-grid thyratron. FG-98 shield-grid thyratron.

impedances. This is especially true in circuits that employ phototubes.

For this reason, a fourth electrode, or shield grid, has been added to the
thyratron. Such a shield-grid thyratron structure is illustrated in Fig.
2-28. The massive cylindrical shield-grid structure encloses the cathode,
control grid, and anode. Owing to the shielding, the grid current is
reduced to a small fraction of its original value, the preignition current
being of the order of 10-3 ,(.lamp.
The critical grid starting characteristics of such a tube are shown in
Fig. 2-29. It will be observed that these characteristics are functions of
the shield-grid voltage.
2-21. The Ignitron. The ignitron is a mercury-pool cathode diode
which is provided with a third electrode for initiating the discharge
between the cathode and anode. The third electrode, or ignitor rod, is
made of a suitable refractory material (such as silicon carbide, boron
carbide, and carborundum) which projects into the mercury-pool cathode.
Such a tube is illustrated in Fig. 2-30.
With an a-c potential applied between the cathode and the anode of
the pool-cathode diode, the arc would be extinguished once each alternate
half cycle, provided that the arc could be initiated regularly. The appli-
cation of a potential to the ignitor rod at the appropriate point in the
cycle will permit the regular ignition of the arc.
SEC. 2-22] CHARACTERISTICS OF ELECTRON TUBES 35
There is a fundamental difference between the control action in a
thyratron and that of the ignitor rod in an ignitron. In thyratrons, the
grid prevents the formation of an arc, whereas the ignitor initiates the
arc. In the former case the electrons already exist in the tube, owing
to the presence of an externally heated cathode, but the grid electrostati-
cally prevents the electrons from flowing to the anode until a critical

_-Anode

Ig'nlTo"
rod
I.----\~ Cafhode

FIG. 2-30. Sketch and photograph of a water-cooled metal ignitron. (General

Electric Co.)

voltage is reached. In the ignitron, the tube is in a nonconducting state

until the ignitor circuit is energized, when conduction is forced.
2-22. Tube Ratings-Current, Voltage, Temperature. Gas- and
vapor-filled tubes are given average rather than rms current ratings.
This rating specifies the maximum current that the tube may carry con-
tinuously without excessive heating of any of the parts. The time over
which the average is to be taken is also specified by the manufacturer.
That the average current is important in such a tube follows from the
fact that the instantaneous power to the plate of the tube is given by the
product of the instantaneous anode current and the instantaneous tube
voltage. Since the voltage is substantially constant and independent of
the tube current, the average power is the product of the tube drop and
36 ELECTRON- TUBE CIRC UITS [CHAP. 2

the average tube current. The tubeg 9,re 9,lgo given pe9,k-current ratings,
this rating specifying the maximum current that the tube should be
permitted to reach in each conducting cycle.
Such tubes are also given peak-inverse-voltage ratings. This is the
largest safe instantaneous negative potential that may be applied to the
tube without the possibility of conduction i~ the inverse direction arising
because of breakdown of the gas in the tube. This potential is also
referred to as the flash-back voltage. The variation of the inverse peak
voltage with temperature for an 866 diode is shown in Fig. 2-31.
10,000
9000
I \
8000
\
g,7000 \
"
~6000 \
>
~5000 \.

'"
OJ
0.4000

,- I~
OJ
60 ~ 3000
0.. > ~
e40.5 2000
"t5
~ 20 1000
~ 0 0
r--
o 20 40 60 80 100 120 140 160
Mercury condens"tion tempemture, ·C.
FIG. 2-31. Peak inverse voltage and tube drop of an 8GG diode as a function of
temperature.

The maximum peak forward voltage is a quantity that is significant

only for thyratrons. It specifies the largest positive potential that may
be applied to the anode before the grid loses its arc-initiating ability.
That is, for potentials higher than this, a glow discharge may occur
between anode and grid, which will immediately initiate the cathode-
anode arc.
The condensed-mercury temperature limits are specified for the safe
and efficient operation of mercury-vapor tubes. The range usually
extends from about 30 to 80°C. The upper temperature limit is deter-
mined by the allowable peak inverse voltage. The lower limit is set by
the allowable tube drop, which increases with decreasing temperature
and which may cause serious cathode disintegration, as well as a decreased
efficiency.
2-23. Deionization and Ionization Times. The ionization time of a
tube specifies the time required for conduction to be established once the
potentials have been applied. It seldom exceeds 10 !-,seconds and is
approximately 0.01 !-,seconds for the 884 thyratron.
SEC. 2-23] CHARACTERISTICS OF ELECTRON TUBES 37

The deionization time is a measure of the minimum time that is

required after removal of the anode potential before the grid of a thyratron
again regains control. It represents the time that is required for the
positive ions to diffuse away from the grid and recombine with electrons
to form neutral molecules. The deionization time depends on many
factors, buch as gas pressure, electrode spacing, and exposed areas. For
commercial tubes that are operated under rated conditions, it varies
between 100 and 1,000 ,useconds. This is considerably longer than the
ionization time and may offer a serious limitation to the use of such tubes
in many applications.
REFERENCE
1. For more details, see Millman, J., and S. Seely, "Electronics," Chap. VII,
McGraw-Hill Book Company, Inc., Ncw York, 1941.

PROBLEMS
2-1. Plot h vs. Eb of the 6H6 diode (see Appendix B) on log paper. From this
plot determine the quantities k and n in the expression Ib = kE;:.
2-2. The anode current in a type 5U4G diode with 54 volts applied between the
plate and cathode is 200 rna. What is the required potential for a current of
100 rna? The tube operates under space-charge conditions.
2-3. Suppose that the FP-400 tube is operating under rated filament power
input (see Prob. 1-2). The operating temperature is 2700oK; anode diameter
= 0.50 in.
a. Calculate the saturation current.
b. At what potential will the current become temperature saturated?
2-4. Plot h vs. Eb + ,uEe on log paper of the 6J5 triode (see Appendix B).
From this curve, find the quantities k and n in the expression Ib = k(E b + ,uK)n.
2-5. The 6J5 triode is operated with Eb = 135 volts. Determine and plot
curves of ,u, (1m, and rp as a function of Ee.
2-6. The rating of a certain triode is given by
h = 130 X 10- 6 (Ee + 0.125Eb)1.58
With Ke = -20 volts, Ebb = 350 volts, find I b, r p , (1m, ,u.
2-7. The plate and grid characteristics of a type 851 power triode are given in
Appendix C. Plot I, = Ib +Ie vs. Eb + ,uEe on log paper, and find the quanti-
ties k and n in the expression I, = k(Eb + ,uEe)n.
2-8. The current in a 6J5 triode for which,u = 20 and which is operating with
K = -8 volts, Eb = 250 volts is 8.7 mao Estimate the current when Eb = 200
volts and Ee = -6 volts.
2-9. A 6J5 triode for which,u = 20 is operating with Eb = 250 volts. What
grid voltage is required to reduce the current to zero?
2-10. Evaluate the value of ,u, (1m, and rp of the 6SJ7 pentode for K = -3,
Ree2 = 100, Eb = 150 volts.

2-11. Evaluate the values of ,u, (1m, rp of the 6SK7 supercontrol pentode for
E .. 2 = 100, Eb = 250 volts, with Ecc! = -2 volts; with Ee,! = -10 volts.
38 ELECTRON- TUBE CIRCUITS [CHAP. 2

2-12. Plot ll, CurVIl of Om VB. Ee! of ll, 6SK7 with Eee2 = 100, Eb = 250 volts.
2-13. Use the plate characteristics of the 6SJ7 and the 6SK7 pentodes to
construct mutual characteristics on the same sheet, with Eb = 200 volts for each
tube. Determine the maximum and minimum values of (}m for each tube in the
range of your sketch.
2-14. Plot a curve of (}m vs. Ee3 of a 6L7, with Eee! = -6, Eec2 = 150 volts.
2-15. Refer to Sec. 2-7 for a discussion of secondary emission caused by elec-
tron impact. What happens to the secondary electrons that are produced by the
impact of the primary current on the anode in a diode? In a triode?
 CHAPTER 3

VACUUM TUBES AS CIRCUIT ELEMENTS

T HE analysis of the behavior of a vacuum tube in a circuit may be

accomplished by two different methods, both of which are to be examined
in some detail. In one method, use is made of the static characteristics
of the tube. The second method achieves two forms. In one, the tube
is replaced by a constant-voltage generator, the potential of which
depends upon the input signal and the internal resistance of which
depends on the tube that is used. In the other, the tube is replaced
by a constant-current generator, the magnitude of the current depend-
ing on the input signal, the generator being shunted by a resistance
which is a function of the tube. Owing to the approximations that are
made in the second methods, they are inherently less accurate than the
direct use of the static characteristics. However, these methods allow
a clearer insight into the operation of the circuit. Because of this
feature, the equivalent-circuit methods of analysis are considerably more
important than that involving the tube characteristics. Moreover, it is
possible to estimate the inaccuracies in the method, thus allowing a
complete understanding of the
operation of the circuit.
The in trod ucti on of the
methods will be made in terms of
the operation of a triode and will
later be extended to the operation +~
£icC
of the other types of tubes. FIG. 3-1. The b!),sic circuit of a triode
3-1. Symbols and Terminology. amplifier.
The simplest triode amplifier is
illustrated in Fig. 3-1. Before proceeding with the analysis, it is neces-
sary to discuss the meaning of the symbols and the general terminology
of vacuum-tube circuits.
The input circuit of the amplifier usually refers to all the elements of
the circuit that exist between the grid and the cathode terminals of the
tube. Similarly, the output, or plate, circuit usually refers to the ele-
ments that are connected between the cathode and the plate terminals.
In the circuit illustrated, the input circuit comprises the input voltage
li!ource eo, the grid resistor Ro, and the bias battery Ec ,' The plate circuit
39
40 ELECTRON- TUBE CIRCUITS [CHAP. 3

consists of the load resistor R! and the plate supply battery Ebb. In
many applications, the input signal eg is a sinusoidally varying potential,
although the wave shape may be nonsinusoidal, and is frequently very
carefully chosen for a particular application.
A variety of potentials, both d-c and varying, are involved simul-
taneously in a vacuum-tube circuit, making it necessary that a precise
method of labeling such sources be established. In what follows, lower-
case letters will be used to designate instantaneous values, and capital
letters will denote either d-c or rms values of sinusoids. The subscripts
c and g will refer to the grid circuit, and the subscripts band p will refer
to the plate circuit. Examples of the notation follow:
Ecc = d-c grid, or C bias
Ebb = d-c plate supply, or B supply
Eg = rms value of the a-c input excitation voltage if this excitation is
sinusoidal
Ep = rms value of the a-c output potential for a sinusoidal output
eg = instantaneous input signal; measured with respect to the input
terminals
ec = instantaneous signal that appears between the grid and cathode
of the tube
ep = instantaneous signal that appears across the output element of
the circuit
eb = instantaneous potential between the plate and cathode of the
tube
i p = instantaneous signal component of plate current; positive in the
direction from the cathode to the plate through the load
ib = instantaneous total plate current; positive in the direction from
the cathode to the plate through the load
Ib = average or d-c current in the plate circuit
Eb = average or d-c potential from plate to cathode
Figure 3-1 illustrates the reference positive polarities and the reference
direction of current.
As a specific illustration of the notation, suppose that the input signal
voltage to the amplifier of Fig. 3-1 is
eg = v'2 Eg sin wt
Then the instantaneous grid-cathode potential is
ec = E cc + V2 E g sin wt
Circuits will be discussed in which no such simple relation between grid
driving signal and grid-cathode potential exists, owing to an involved
interconnection of circuit elements among the tube elements.
3-2. Graphical Analysis. Refer to Fig. 3-1, and suppose that the
grid input signal eg = o. Owing to the d-c sources Ecc and Ebb, it will be

f
SEC. 3-2] VACUUM TUBES AS CIRCUIT ELEMENTS

supposed that there is a current in the plate circuit. This is true only if
the plate supply Ebb and the grid supply Eec are properly chosen. The
value of this current may be found graphically. In fact, it is essential
that a graphical solution be used. This follows from the fact that the
plate circuit of Fig. 3-1 yields the
relation

However, this one equation is not

sufficient to determine the current
corresponding to the potential Ebb,
since there are two unknmyn quan-
tities in the expression, eb and i b.
A second relation between eb and FIG. 3-2. The operating point Q is
located at the intersection of the load
ib is given by the plate character- line and the plate characteristic for
istics of the triode. The simultane- e, = -E".
ous solution of Eq. (3-1) and the
plate characteristics will yield the desired current. This is accomplished
by drawing Eq. (3-1) on the plate characteristics, in the manner illus-
trated in Fig. 3-2. The line that passes through the points
ib = 0
. Ebb
1,b = -
Rz
is known as the load line. It is obviously independent of the tube char-
acteristics, for it depends only upon elements external to the tube. The
intersection of this line with the curve for ee = Eee is called the operating
or quiescent point Q. The grid-bias supply Eee is usually such as to
maintain the grid negative relative to the cathode. The Q current in
the external circuit is h, and the corresponding plate-cathode potential
is E b •
Suppme that the grid-cathode potential is
ee = E ce + y2 E g sin wt
T-he maXImum and minimum values of ee will be Eee + V2 Eg and
Eee - V2 Eo, respectively. The plate current and the plate voltage
will vary about the values of hand E b • The graphical construction of
Fig. 3-3 shmvs the details of the variations. The values of eb and ib for
any given value of ee are obtained from the intersection of the load line
and the ib-eb curve for the specified ee. The points ai, b' , c' , etc., of the
output current and the points a", btl, c", etc., of the output-voltage wave
correspond, respectively, to the points A, B, C, etc., of the input-grid-
signal wave form.
42 ELECTRON- TUBE CIRCUITS [CHAP. 3

~-L~~~~--~~--~eh

~-----+-=~~-+--~eb
bIt
I
.---- eh --I-
I
I
I
I
FIG. 3-3. The output current and voltage wave forms for a given input grid signal.
It is instructive to show the several wave shapes in their proper phase
relation. This is done in Fig. 3-4. It should be noted in particular that
the variations about the quiescent values
~---r--------~--~wt
have been labeled. The quantities so
labeled are

(3-2)

, These quantities give a measure of the

' -__-'i'-'lbO--_ _ _ _...l-_~wt amplification property of the amplifier,
as it is a direct measure of the a-c output
variations for a given a-c input variation.
The curves of Fig. 3-4 indicate the fol-
A lowing very significant results: If the
! current i p is sinusoidal, then i p and Cp are
~b
I
180 deg out of phase with each other.
L - - . . . L - - - _ L -__ wf Also, the grid driving voltage Cg and the
FIG. 3-4. The input grid wave plate current i p are in phase with each
shape, and the corresponding out- other. This simply states that, when a
put current and voltage wave positive-going signal is applied to the grid,
shapes. the tube current increases. Moreover,
with an increased current in the plate circuit, the potential of the plate
falls.
A curve of the intersection of the load line with the static-characteris-
tic curves, which is a measure of the current ib as a function of Cc for the
specified Eb& and load R z• is important. It is known as the "dynamic"
characteristic of the tube circuit and yields directly the output current
 SEC. 3-3] V ACUUM TUBES AS CIRCUIT ELEMENTS 43

for a given input signal. The construction is directly related to the con-
struction of Fig. 3-3, and is given in Fig. 3-5. The corresponding points on
both curves are similarly marked.
3-3. Voltage-source Equivalent
2'Circuit of a Triode. Ordinarily one
is i~terested in the "a-c response"
of the circuit, rather than in ob-
taining the total instantaneous var-
iation of the potentials and
currents. That is, the value of i p I
and of ep for a given eg are ordinar-
ily desired. It will be found pos-
n',
sible to obtain this information by
,
an analytic method, without direct
recourse to the graphical solution FIG. 3-5. The dynamic curve, and its
use in determining the output wave
of the foregoing section. shape for a given input signal.
To this end, again consider cer-
tain of the results of Sec. 2-5. According to Eqs. (2-13) and (2-14), the
variation in current about the quiescent point is given as

(3-3)

This may be written as

(3-4)

But as the changes about the quiescent values are, respectively,

l1ib = ib - Ib = i p }
l1e c = Cc - Ecc = eg (3-5)
l1eb = eb - Ebb = ep
then Eq. (3-4) becomes
(3-6)

or
(3-7)
This expression shows that the voltage ep comprises two components;
one is a generated emf which is J.i, times as large as the grid-cathode
voltage, and the second is a voltage across the tube resistance rp resulting
from the current i p through it.
These results are illustrated graphically in Fig. 3-6. The tube is
replaced by a fictitious generator, with a generated emf f.Le g and an internal
resistance rp. Confusion sometimes arises because of the failure to
44 ELECTRON-TUBE CIRCUITS [CHAP. 3

appreciate fully the polarities and the potentials in the circuit. Figure
3-6 is redrawn in a manner to stress this matter. This figure, when com-
pared with the original circuit of Fig. 3-1, shows that the equivalent cir-
cuit is substantially like the original circuit, except that the tube, relative
to the tube pins, is replaced by the equivalent voltage generator j.!egk, the
polarity of which is opposite to that associated with the grid driving
potential egk and an internal resistor rp. Since the equivalent circuits of
Figs. 3-6 and 3-7 give a measure of the variation about the Q point, no
d-c quantities appear in the figure.

G P
-.," --r-I

~
I

~
I I
I I
I I
ep
Rl ep Rt
I
I
I
I
--1.

FIG. 3-6. The equivalent circuit of a FIG. 3-7. The circuit of Fig. 3-6 redrawn
triode. in a manner to show the elements that
replace the tube, and the associated po-
tentials.

The equivalent circuit was derived without consideration of the char-

acter of the load. Consequently, it is valid for any type of load. The
only restriction is that the parameters j.!, rp , and gm must remain sub-
stantially constant over the operating range.
The technique of drawing the equivalent circuit of any tube circuit is a
straightforward process, although care must be exercised in carrying this
process out. To avoid error, the following simple rules will be found
helpful.
1. Draw the actual circuit diagram neatly.
2. Mark the points G, P, and K on this diagram. Locate these points
as the start of the equivalent circuit. Maintain the same relative
position as in the original circuit.
3. Between points P and K include a resistor rp and a generator of
potential j.!egk, where euk is the grid-cathode potential. The cathode of
this generator is marked +.
4. Omit the tube and all d-c sources from the diagram, but transfer all
circuit elements from the actual circuit to the equivalent circuit without
altering the relative positions of these elements.
The point of special importance is that, quite apart from the points of
application of the input signal, the equivalent fictitious generator must
always be j.!eak, where egk is the total grid-cathode potential.
Once the equivalent circuit has been drawn, it is no longer necessary to
SEC. 3-4] V ACUUM TUBES AS CIRCUIT ELEMENTS 45

refer to the original circuit. Moreover, since the equivalent circuit will
contain only generators and passive circuit elements (resistances, induct-
ances, and capacitances), the usual methods of a-c circuit analysis are
employed to calculate the desired quantities.
3-4. Measurement of Triode Coefficients. As several specific illus-
trations of the methods of analysis just discussed, the circuits for obtain-
p

Zearphones

FIG. 3-8. The Miller bridge and its equivalent circuit, for determining the a.mpli-
fica.tion fa.ctor of a triode under operating conditions.

ing the values of }.t, r p , and gm of a triode will be analyzed. It should be

recalled that the triode coefficients, first discussed in Sec. 2-5, were shown
to be related to the slope of the static-characteristic curves, according to
Figs. 2-5 to 2-7. However, the accuracy with which these quantities can
be measured in this way is not high. Not only do the methods now to be
discussed yield results which are made under dynamic conditions, but the
results are usually more accurate.
The amplification factor /-L is readily determined by means of the circuit
given in Fig. 3-8. The operations involved in balancing the bridge
consist simply in varying Rl and R2 until no signal from the oscillator
is heard in the earphones. When this condition prevails, the plate
current f p = O. Then the potential E gk = fR l • By applying Kirch-
hoff's law to the plate circuit,
-/-LEak + fR2 = 0
or

It follows from this that

_ R2
/-L - Rl (3-8)

This measurement may be effected for any desired d-c current in the tube
simply by adjusting the grid bias Ecc.
The transconductance gm is measured by means of a bridge circuit that
is a slight modification of Fig. 3-8. The addition of a resistor R3 between
the plate and cathode makes this measurement possible. The schematic
46 ELECTRON- TUBE CIRCUITS [CHAP. 3

and equivalent circuits of this bridge network are given in Figs. 3-9.
The measurement is accomplished by adjusting the resistors until no
signal is heard in the earphones.

6
-------.
P

Zearphones

FIG. 3-9. The Miller bridge from determining the transconductance of a triode under
operating conditions.

By applying Kirchhoff's law to the several meshes there results

IpRa + Iprp - p,Eok = 0
But the potential Eok is

Then
(3-9)
Also, it follows that

or
IRz = IpRa (3-10)
The ratio of Eq. (3-9) to Eq. (3-10) is
Ra + rp Rl
Ra = p, Rz
from which

rp = Ra(p,~~ -1) (3-11)

Although this bridge may be used to evaluate r p , the result would be

dependent on the measurement of p,. If, however, Rl is chosen in such a
way that p,R1/R z » 1, then approximately

or
P, R2 (3-12)
gm = ;:; = RaRl
The plate resistance rp of the tube can be measured directly by inc or-
SEC. 3-5] VACUUM TUBES AS CIRCUIT ELEMENTS 47
.porating the plate circuit of the tube as the fourth arm of a Wheatstone
bridge, as shown in Fig. 3-10. When the bridge is balanced,

Tp
R2Ra
= ]'[; (3-13)

The above circuits do not yield perfect balance owing to the capacitive
effects of the tube, and it is sometimes
necessary to provide a means for bal-
ancing these effects. Basically, how-
ever, the circuits are those given.
3-5. Current-source Equivalent Cir-
cuit. The current-source equivalent
circuit differs from the voltage-source
equivalent circuit discussed in Sec. 3-3
principally in replacing the tube by a
current generator which supplies a cur- FIG. 3-10. A Wheatstone bridge for
rent gmegk in the direction from plate- determining the plate resistance of n.
triode under operating conditions.
cathode within the tube, and with the
plate resistance Tp across the generator terminals. The result, illus-
trated in Fig. 3-11, follows directly from Eq. (3-6),
. ep
g",fg = tp - - (3-14)
Tp

or in more complete notation

gmegk =
.
tp - r
epk
p
(3-15)

Equation (3-15) is interpreted to mean that the tube acts as a generatOl

that supplies a current gmegk. This current divides between Tp and thp

---.r
I
I

R/ ~.po
I
I

' - -_ _ _~--_--l~-

FIG. 3-11. The schematic and current-source equivalent circuit of a simple triode
amplifier.

load current i p • Note that, if the load impedance is small compared

with the plate resistance, then, clearly, the load current may be considered
to be substantially constant and equal to gmegk' It will be shown later
that this condition applies for a pent ode.
18 ELECTRON - TUBE CIRCUITS [CHAP. 3

3-6. Harmonic Generation in a Tube. The analysis of Sec. 3-3

usually permits an adequate solution of an amplifier circuit 'when the
limitations of the method are not exceeded or if relatively slight differ-
ences are considered of no importance. There are occasions when it is
desirable to examine critically the effects of the assumptions.
The assumption of linear operation, which is implied in Eq. (3-3) and
which assumed that higher-order terms in the Taylor expansion of the
current ib(eb,e c ) are negligible, is not always valid. This assumption,
which allowed the graphical solution of Fig. 3-5 to be replaced by the
analytical one of Fig. 3-6, requires that the dynamic characteristic of
the amplifier circuit be linear over the range of operation. Actually, the
dynamic characteristic is not linear in general but contains a slight
curvature. This nonlinear characteristic arises because the ib-eb static
characteristics (see Fig. 3-3) are not equidistant lines for constant
ec intervals over the range of operation. The effect of this nonlinear
dynamic characteristic is a nonsinusoidal output wave shape when the
input wave is sinusoidal. Such an effect is known as nonlinear or ampli-
tude distortion.
It is possible to obtain a measure of the degree of nonlinearity that
results from the existence of the nonlinear dynamic curve. To do this,
it is observed that the dynamic curve 'with respect to the Q point may be
expressed by a power series of the form
(3-16)
Clearly, if all terms in this series vanish except the first, then the linear
assumptions of the equivalent-circuit concept result. It \vill be found
that triodes, when operated under normal conditions, may be adequately
expressed by retaining the first two terms in the expansion. When
a triode is operated with such a large signal that the instantaneous
grid-cathode potential becomes positive, or if the triode is operated with
such a bias that the very curved portions of the plate characteristics must
be employed, more than two terms must be retained in the expansion.
Lik8\yise, it is found that the parabolic approximation is not adequate to
represent the dynamic curve of a tetrode or a pentode under normal
operating conditions. If the dynamic curve contains an extreme curva-
ture or if the operation is over an extreme range, it is sometimes found
preferable to devise special methods of analysis. Such special methods
will be used in the analysis of a tuned class C amplifier in Chap. 11.
Suppose that the dynamic curve may be represented as in Eq. (3-16),
and consider that the input wave is a simple cosine function of time, of
the form
eg = Egm cos wt (3-17)
SEC. 3-6] V ACUUM TUBES AS CIRCUIT ELEMENTS 49
By combining this expression with Eq. (3-16) and expanding the higher-
order powers of the cosine that appear in the resulting series, the result
may be shown to have the form
ib = h + B o + Bl cos wt + B2 cos 2wt + Ba cos 3wt + . .. (3-18)
If it is assumed that the excitation voltage is a sine function of the time
instead of the cosine form chosen, the resulting Fourier series representing

-+-+-----f---U1~

FIG. 3-12. The construction for obtaining the plate-current values to be used in thc
five-point schedule for determining the Fourier coefficient!!.

the output current will be found to contain odd sine components and
even cosine components.
A number of different methods exist for obtaining the coefficients Bo,
B l , B 2 , etc. One of the more common methods is best discussed by
reference to Fig. 3-12. It will be assumed for convenience that only five
terms, Bo, B l, B 2 , B a, B 4 , exist in the resulting Fourier series. In order to
evaluate these five coefficients, the values of the current at five different
values of Cg are required. The values chosen are lmax, h" h, 1-'/2, and
lmin and correspond, respectively, to the following values of Cg : the maxi-
mum positive value; one-half the maximum positive value; zero; one-half
the maximum negative value; the maximum negative value.
It is evident from the figure that the currents are those chosen as shown
at the angles
wt = 0 ib = 1ma.
wt = ?:: ib = IVJ
3
wt = ?:: ib = lb (3-19)
2
27r
wt = 3 ~b =L'/2
wt =7r ib = lmin

By combining these results with Eq. (3-18), five equations containing five
50 ELECTRON - TUBE CIRCUITS [CHAP. 3

unknowns are obtained. The simultaneous solution of these equations

yields
Bo = }~ (Imax + 21"", + 2L"", + lmin) - Ib
Bl = H (Imax + I"", - Ly. - lmin)
B2 = X (Imax - 2h + lmin) (3-20)
B3 = Y6 (Imax - 21y. + 2Ly. - lmin)
B4 = 712 (Imax - 41y. + 6h - 4Ly. + fmin)
The percentage of harmonic distortion is defined as
B3
D3 = Bl X 100% (3-21)

where DB (8 = 2, 3, 4, . . . ) represents the per cent distortion of the

8th harmonic and the total distortion is defined at'>
D = V D~ + D~ + D~ + ... (3-22)

For the case where a three-point schedule is sufficient, and, as already

indicated, this would apply for a triode under normal operating conditions,
the analysis yields the expressions
Bl = Y2 (Imax - fmin) }
(3-23)
B2 = Bo = X (fmax - 2h + lmin)
PROBLEMS
3-1. A 6C5 triode is used in the circuit of Fig. 3-1, the plate characteristics of
which are given in Appendix B.
a. With Ebb ~ 300 volts, Eoc = -8 volts, Rz = 2{)k, draw the load line, and
locate the operating point. Plot the dynamic characteristic.
b. If eg = 6 sin 10,000t, determine graphically the output current, and plot
the curve as a function of wt.
c. From these curves, determine and plot the instantaneous plate voltage for
the same interval. Check the phase relation between a-c components of
grid voltage, plate current, and plate voltage .
../' 3-2. The characteristics of a given triode may be represented by the expression
Ib = 8.8 X 1O-3(Eb + 16E )l.6
c rna

It is to be operated at a plate potential Eb = 250 volts and a grid-bias voltage

Ec = -9 volts.
a. Calculate the plate resistance of the tube.
b. If this tube is used in the circuit of Fig. 3-1 with a load resistor Rz = 10k ,
determine the plate-supply voltage necessary for the tube to be operating
under the specified conditions.
c. Suppose that the grid driving source applies a voltage eg = 8 sin wt to the
grid. Determine the a-c potential across the load resistor.
 V ACUUM TUBES AS CIRCUIT ELEMENTS 51
\.. 3-3. Draw the voltage-source equivalent circuits of the electron-tube circuits
in the accompanying diagram.

t I
I

O~fpuf
I
I
I

--~1~~--~----~--1
(a)

I
I

oufPUf
Ra ,,,
---.t
(0) (d)
3-4. A type 6A3 triode is used in an amplifier circuit to supply power to a
3,000-ohm resistor. In this circuit Ecc = --15 volts, Ebb = 350 volts. A 45-volt
peak a-c signal is applied to the grid.
a. Plot the dynamic curve of the tube.
b. Assume that only the fundamental and a second harmonic exist in the
output. Determine the magnitudes of each.
c. Plot a curve showing the output current for the sinusoidal input. On the
same sheet, plot the corresponding results from the calculations in part b.
3-5. It is possible to obtain a five-point schedule for determining the coefficients
Bo, B l , B 2 , B a, B4 by almost any sensible choice of angle. Determine the five-
point schedule for determining the coefficients B in terms of I max, I o. 707, lb, 1_0.707,
1min.
 CHAPTER 4

BASIC AMPLIFIER PRINCIPLES

T HE classification of an amplifier is usually somewhat involved, owing

to the fact that a complete classification must include information about
the tubes that are used, the conditions of the bias, the character of the
circuit elements connected to the tubes, the function of the circuit, and
the range of operation. Certain of these factors will be discussed here,
but many will be deferred for later discussion.
4-1. Classification of Tubes and Amplifiers. Apart from the wide
variety of vacuum tubes of the diode, triode, tetrode, pentode, beam,
hexode, heptode, and multiunit types and the varied power capacities
of each type, it is possible to classify the tubes according to their principal
applications. Tubes may be classified roughly into five groups, viz.,
voltage-amplifier tubes, power-amplifier tubes, current-amplifier tubes,
general-purpose tubes, and special-purpose tubes.
1. Voltage-amplifier tubes have a relatively high amplification factor
and are used where the primary consideration is one of high voltage gain.
Such tubes usually operate into a high impedance load, either tuned or
untuned.
2. Power-amplifier tubes are those which have relatively low values of
amplification factor and fairly low values of plate resistance. They are
capable of controlling appreciable currents at reasonably high plate
potentials.
3. Current-amplifier tubes are those which are designed to give a large
change of plate current for a small grid potential; i.e., they possess a high
transconductance. These tubes may be required to carry fairly large
plate currents. Such tubes find application as both voltage and power
amplifiers, depending on the tube capacity.
4. General-purpose amplifier tubes are those whose characteristics are
intermediate bet\veen the voltage- and the power-amplifier tubes. They
must have a reasonably high amplification factor and yet must be able to
supply some power.
5. Special-purpose tubes include a wide variety of types. The hexode,
heptode, and multiunit tubes are of this type.
Amplifiers are classified according to their frequency range, the method
of tube operation, and the method of interstage coupling. For example,
52
SEC. 4-1] BASIC AMPLIFIER PRINCIPLES 53
they may be classed as direct-coupled amplifiers, audio-frequency (a-f)
amplifiers, video amplifiers, or tuned r-f amplifiers if some indication of
the frequency of operation is desired. Also, the position of the quiescent
point and the extent of the tube characteristic that is being used will

~
ib

~b
__ _

j - -
I
- I - -
ec
~IOISSA
F-{ : - ~IOISSAB
Nl .

QjL-n-
~CIOISSC •
Hi
FIG.4-1. Amplifier classification in terms of the position of the quiescent point of the
tubes.

determine the method of tube operation. This will specify whether the
tube is being operated in class A, class AB, class B, or class C. These
definitions are illustrated graphically in Fig. 4-1.
1. A class A amplifier is an amplifier in which the grid bias and the a-c
grid potentials are such that plate current flows in the tube at all times.
2. A class AB amplifier is one in which the grid bias and the a-c grid
potentials are such that plate current flows in the tube for appreciably
.more than half but less than the entire electrical cycle.
3. A class B amplifier is one in which the grid bias is approximately
equal to the cutoff value of the tube, so that the plate current is approxi-
mately zero when no exciting grid potential is applied, and such that
plate current flows for approximately one-half of each cycle when an a-c
grid voltage is applied.
4. A class C amplifier is' one in which the grid bias is appreciably
g~eater than the cutoff value, so that the plate current in each tube is zero
when no a-c grid voltage is applied, and such that plate current flows for
appreciably less than one-half of each cycle when an a-c grid potential is
applied.
To indicate that grid current does not flow during any part of the input
cycle, the suffix 1 is frequently added to the letter or letters of the class
identification. The suffix 2 is added to denote that grid current does
flow during some part of the cycle. For example, the designation class
54 ELECTRON- TUBE CIRCUITS [CHAP. 4

ABI indioates that the amplifier operates under class AB conditions and
that no grid current flows during any part of the input cycle.
Voltage amplifiers, whether tuned or untuned, generally operate in
class A. Low-power audio amplifiers may be operated under class A and,
with special connections, under class AB or class B conditions. Tuned r-f
power amplifiers are operated either under class B or under class C condi-
tions. Oscillators usually operate under class C conditions. A detailed
discussion is deferred until the appropriate point in the text. When a
tube is used essentially as a switch, no classification is ordinarily specified.
4-2. Distortion in Amplifiers. The application of a sinusoidal signal
to the grid of an ideal class A amplifier will be accompanied by a sinus-
oidal output wave. Frequently the output wave form is not an exact
replica of the input signal wave form because of distortion that results
either within the tube or from the influence of the associated circuit.
The distortions that may exist either separately or simultaneously are
nonlinear distortion, frequency distortion, and delay distortion. These
are defined as follows:
1. Nonlinear distortion is that form of distortion which occurs when
the ratio of voltage to current is a function of the magnitude of either.
2. Frequency distortion is that form of distortion in which the change
is in the relative magnitudes of the different frequency components of a
wave, provided that the change is not caused by 'nonlinear distortion.
3. Delay distortion is that form of distortion which occurs when the
phase angle of the transfer impedance with respect to two chosen pairs of
terminals is not linear with frequency within a desired range, the time of
transmission, or delay, varying with frequency in that range.
In accordance with definition 1, nonlinear distortion results when new
frequencies appear in the output which are not present in the input signal.
These new frequencies arise from the existence of a nonlinear dynamic
curve and are discussed in Sec. 3-6.
Frequency distortion arises when the components of different fre-
quency are amplified by different amounts. This distortion is usually
a function of the character of the circuits associated with the amplifier.
If the gain vs. frequency characteristic of the amplifier is not a horizontal
straight line over the range of frequencies under consideration, the circmt
is said to exhibit frequency distortion over this range.
Delay distortion, also called phase-shift distortion, results from the fact
that the phase shift of waves of different frequency in the amplifier is
different. Such distortion is not of importance in amplifiers of the a-f
type, since delay distortion is not perceptible to the ear. It is very
objectionable in systems that depend on wave shape for their operation,
as, for example, in television or facsimile systems. If the phase shift is
SEC. 4-3] BASIC AMPLIFIER PRINCIPLES 55

proportional to the frequency, a time delay will occur although no dis-

tortion is introduced. To see this, suppose that the input signal to the
amplifier has the form
eg = Em! sin (wi + ( + Em2 sin (2wi + ( + . . .
1) 2) (4-1)
If the gain K is constant in magnitude but possesses a phase shift that is
proportional to the frequency, the output will be of the form
ep = KErni sin (wi + 8 + if;) + KEm2 sin (2wt + 8z + 2if;) + ...
1

This output voltage has the same wave shape as the inlJut signal, but a
time delay between these two waves exists .• By writing
wt' = wt + if;
then
ep = KEmi sin (wt' + (h) + KEmz sin (2wt' + ( + 2) (4-2)
This is simply the expression given by Eq. (4-1), except that it is referred
to a new time scale t'. Delay distortion, like frequency distortion, arises
from the frequency characteristics of the circuit associated with the
vacuum tube.
4-3. The Decibel; Power Sensitivity. In many problems where two
power levels are to be compared, it is found very convenient to compare
the relative powers on a logarithmic rather than on a direct scale. The
unit of this logarithmic scale is called the bel. A decibel, which is abbre-
viated db, is YIo bel. By definition, two power sources are in the ratio

!
of N bels, according to
z
Number of bels = logio PI
or P (4-3)
Pz
Number of db = 10 logio PI

It should be emphasized that the bel or the decibel denotes a power ratio.
Consequently the specification of a certain power in decibels is meaning-
less unless a reference level is implied or is explicitly specified. In com-
munication applications it is usual practice to specify 6 mw as the zero
reference level. However, any power may be designated as the zero
reference level in any particular problem.
Suppose that these considerations are applied to a power amplifier,
with P z the output power and PI the input power. This assumes that
Jthe input circuit to the amplifier absorbs power. If the grid circuit does
not absorb an appreciable power, then the term decibel gain of the amplifier
means nothing. Under such conditions, it is customary to speak of
'power sensitivity, which is defined as the ratio of the power output to the
56 ELECTRON - TUBE CIRCUITS [CHAP. 4

gqullre of the input signal voltage. Thus

.. . P
P ower sensItIvIty == E2 mhos (4-4)
g

where P is the power output in watts and Eg is the input signal rms volts.
If the input and output impedances are equal resistances, then
P 2 = EUR and P l = Ei/R, where E2 and El are the output and input
potentials. Under this condition, Eq. (4-3) re.duces to

Number of db = 20 loglo ~2 (4-5)

• 1

In general, the input and output resistances are not equal. Despite
this, this expression is adopted as a convenient definition of the decibel
voltage gain of an amplifier. It is essential, however, when the gain of an
amplifier is discussed, that it be clearly stated whether one is referring to
voltage gain or power gain, as these two figures will be different, in general.
Many of the considerations of the foregoing sections are best illustrated
by several examples.
Example 1: Calculate the gain of the amplifier circuit of Fig. 4-2.
'6 P
.-----,---,

FIG. 4-2. Schematic and equivalent circuits of a simple amplifier.

Solution: The equivalent circuit of the amplifier is drawn according to the rules
of Sec. 3-3 and is that of Fig. 4-2b. The application of the Kirchhoff voltage law
to the equivalent circuit yields
Jp(RI + rp) - p,Egk = 0
But the potential difference Eok is simply Eg. The output current is then

J - p,Eg (4-6)
p - Rl + rp
and the corresponding output potential is

(4-7)

The gain, or voltage amplification, of the amplifier is the ratio of the output,
to the input potentials and is

K = E pk = -p,R I (4-8)
E gk Rl + Tp
SEC. 4-3] BASIC AMPLIFIER PRINCIPLES 57
A plot of this expression as a function of the load resistance is given in Fig. 4-3.
Note that the maximum possible gain is III although this gain cannot be achieved
with finite values of R z• However, the
1.0
gain K increases rapidly with R z, and
an amplifier with significant gain is
easily achieved.
0.8
_![ 0.6
...v
Example 2: A type 6J5 triode for J.l 0.4
which J.I. = 20, Tp = 7,700 ohms is em-
j
0.2 1/
ployed in an amplifier, the load of 0 0 I 2 3 4 5 6 '1 8 9 10
which consists of an inductance for Rz/rp
which Rz = 1,000 ohms and L = 1 FIG. 4-3. The gain of the amplifier of
henry. Calculate the gain and phase Fig. 4-2 as a function of load resistance.
shift of the amplifier at w = 2,000
rad/sec and w = 10,000 rad/sec. Draw the complete voltage sinor* diagram of
the system. The input signal is 6 volts rms.
Solution: The schematic and equivalent circuits are shown in the accompanying
diagram. At w = 2,000 rad/sec,

;k
;h

120 + jO .
Jp = 7,700 + (1,000 + j2,000) = 13.1 - )3.01 rna

The output potential is

E pk = -(1,000 + j2,000) (13.1 -j3.01) X 10- 3

+ j23.2)
- (19.1 = 30.1/ -129.5°

The gain is given by

K = E pk = 30.1~9~
E gk 6/0
= 5.01/ -129.5~

The voltage sinor diagram has the form

shown in the sketch. At w = 10,000
l"'.1d/sec,
120 + jO
J p = 7,700 + (1,000 + jlO,OOO) = 5.94 - j6.83 rna

* A number of different terms have been used to describe the rotating line segment,
the projection of which generates a sinusoid. The term sinor is used in this text.
Other terms that appear in the literature are: vector, complex vector, complexor, and
phasor.
58 ELECTRON - TUBE CIRCUITS [CHAP. 4

Th{\ output voltng{\ is

E pk = - (1,000 + j10,OOO) (5.94 - j6.83) X 10- 3
= - (74.2 + j52.6)
= 90.8 / -144.7°
The gain is given by
90.81 -144.7°
K= 6LQ =15.1/-144.7°

The voltage sinor diagram has the form of the accompanying diagram.
Egk

The results are tabulated for convenience. An examination of the results indi-
cates the presence of frequency distortion, since the gain at w = 2,000 rad/sec is
different from that at w = 10,000 rad/sec. Also, phase-shift distortion exists in
this amplifier.

w Gain and phase Voltage db gain

2,000 5.01 /-129.5° 14 db

10,000 15.1 /-144-=7° 23.6 db

4-4. Interelectrode Capacitances in a Triode. It was assumed in

the foregoing discussions that, with a negative bias on the grid, the grid
driving-source current was negligible. This is generally true if one
examines only the current intercepted by the grid because of its location
within the region of the electron stream. Actually though, owing to the
physical proximity of the elements of the tube, interelectrode capacitances
between pairs of elements exist. These capacitances are important in
the behavior of the circuit, as charging currents do exist.
Owing to the capacitance that exists between the plate and the grid, it
is not true that the grid circuit is completely independent of the plate
circuit. Since the capacitance between plate and grid is small, the
approximation that the plate circuit is independent of the grid circuit is
valid at the lower frequencies. However, at the higher frequencies,
interelectrode capacitances may seriously affect the operation.
A more complete schematic diagram and its equivalent circuit are given
in Fig. 4-4. In this circuit, Cgp denotes the capacitance between the grid
SEC. 4-4] BASIC AMPLIFIER PRINCIPLES 59

and the plate, Cgk is the grid-cathode capacitance, and Cpk is the capaci-
tance between the plate and the cathode. The solution for the gain of this
circuit is readily effected with the aid of the Millman theorem (see Appen-
dix A). The point 0' in Eq. (A-I) corresponds to the plate terminal P,
and the point 0 is the cathode terminal K. Four branches must be con-
sidered between these points: the load impedance with zero potential; the
G I~C3 P

Cj=Cgk
FIG. 4-4. Schematic and equivalent circuits of an amplifier, including the inter-
electrode capacitances.

capacitor C2 with zero potential; the potential rise /lEg in series with rp;
the potential Ey in series with Ca. The capacitor C1 which exists across
the input Eg does not appear in the equation. This latter follows from
the fact that the exact circuit connections between points 0, 1, 2, and 3
of Fig. A-I need not be specified. The result is
E - -J.LEgYp + EuYa (4-9)
pk - Yp + Yl + Y 2 + Ya
where = l / rp is the admittance corresponding to Tp
Yp
Y 2 = jwC 2 is the admittance corresponding to C2
Y 3 = jwC a is the admittance corresponding to C a
Y l = IIZl is the admittance corresponding to Zl
E pk = potential difference between P and K, or the voltage across
the load impedance
The vector voltage gain is given by
K = output voltage = E pk = E pk
input voltage E gk Eg

and may be written in the form

K = Y a - gm (4-10)
Yp + Yl + Y 2 + Ya
In this expression, use has been made of the fact that gm = /l/rp.
In this analysis, a number of factors have been neglected. It has been
assumed that no conduction or leakage currents exist between tube
terminals. Such leakage current will depend upon many variable fac-
tors-the spacing between electrodes, the material of the base, the condi-
tions of the surface of the glass and the tube base, and perhaps the surface
60 ELECTRON- TUBE CIRCUITS [CHAP. 4

lell.kllge between connecting wires. Ordinarily the error IS small in

neglecting the effects of this surface leakage. If this assumption is not
true, the effect can be taken into account by writing for each interelec-
trode admittance g8 + jwC s instead of jwCs, where gs takes account of the
leakage current and also dielectric losses. Interwiring and stray capaci-
tances must be taken into account. This may be done by considering
them to be in parallel with C I , C2 , and Cs.
The error made in the calculation of the gain by neglecting the inter-
electrode capacitances is very small over the a-f spectrum. These
interelectrode capacitances are usually 10 j.J,j.J,f or less, which corresponds
to an admittance of less than 2 j.J,mhos at 20,000 cps. This is to be com-
pared with the mutual conductance of the tube of, say, 1,500 j.J,mhos at
the normal operating point. Likewise Y 2 +
Y 3 is usually negligible
compared with Yp +
Y I • Under these conditions, the expression for the
gain [Eq. (4-10)] reduces to Eq. (4-8).
4-5. Input Admittance of a Triode. Owing to the presence of the
J
interelectrode capacitances, the grid circuit is no longer isolated from the
plate circuit. In fact, with a positive-going signal on the grid and with
the consequent negative-going potential on the plate, an appreciable
change of potential appears across the capacitance Cgp, with a consequent
appreciable current flow. Also, the potential change across the capaci-
tance Cgk is accompanied by a current flow. Clearly, therefore, the input
signal source must supply these currents. To calculate this current, it is
noted from the diagram that
II = Eg Y I
and
Is = EgpYa = (Eg - E pk )Y3
But from the fact that
Epk = KEgk = KEg
then the total input current is
Ii = II + Is = [Y I + (1 - K)Ys]Eg
The input admittance, given by the ratio Y i = Ii/E g, is
(4-11)

If Y i is to be zero, evidently both YI and Y amust be zero, since K cannot,

in general, be 1/0 deg. Thus, for the system to possess a negligible input
admittance over a wide range of frequencies, the grid-cathode and the
grid-plate capacitances must be negligible.
SEC. 4-5] BASIC AMPLIFIER PRINCIPLES 61
Consider a triode with a pure resistance load. At the lower frequencies,
the gain is given by the simple expression [Eq. (4-8)]

K =
Rl-f.LR
+ rp
In this case, Eq. (4-11) becomes

Y i = jw [C1 + (1 + R;!lr)C3] (4-12)

Thus the input admittance is that from a capacitor between grid and
cathode of magnitude
)

(4-13)

Owing to the magnitude of the gain, the input capaci-

tance is considerably higher than any of the interelec-
trode capacitances. The presence of this capacitance FIG. 4 - 5. The
equivalent input
will be found to affect the operation of the amplifier. circuit of a triode.
For the general case when the gain of the amplifier
K is a complex quantity, the input admittance will consist of two terms,
a resistive and a reactive term. For the case of an inductive load, the
gain K may be written in the form (see Sec. 4-3, Example 2)
K = - (k1 + jk 2) (4-14)
and Eq. (4-11) becomes
Y i = -wC 3k2 + jw[C + (1 + k
1 1 )C 3] (4-15)
This expression indicates that the equivalent input circuit comprises a
resistance (which is negative in this particular case, although it will be
positive for a capacitive load) in parallel with a capacitance Ci, as shown
in Fig. 4-5. The equivalent elements have the form
1
Ri = - wC 3k2 (4-16)

and the capacitor

(4-17)
As indicated in the above development, it is possible for the term k2 to
be negative (with an inductive load). Under these circumstances the
effective input resistance will be negative. Physically, this means that
power is being fed back from the output circuit into the grid circuit
through the coupling provided by the grid-plate capacitance. If this
feedback reaches an extreme stage, the amplifier will oscillate. These
62 ELECTRON - TUBE CIRCUITS [CHAP. 4.

feed-back effects in an amplifier will be examined in some detail in Sec.

5-S.
4-6. Input Admittance of a Tetrode. The basic equivalent circuit of
the tetrode is essentially that of the triode, even though a screen grid
exists in the tetrode. A schematic diagram of a simple amplifier circuit
employing a tetrode is given in Fig. 4-6. In drawing the equivalent
circuit, the rules given in Sec. 3-3 have been appropriately extended and

FIG. 4-6. Schematic and equivalent circuits of a tetrode in an amplifier circuit.

employed. This requires the introduction of a point S, the screen ter-·

minaI, in addition to the points K, G, and P.
Notice that the screen potential is maintained at a fixed d-c potential
with respect to cathode and is at zero potential in so far as a-c variations
about the Q point is concerned. As indicated in the figure, this effectively
places a short circuit across Cks and puts Cgk and CyS in parallel. This
p parallel combination is denoted C1 •
G
The capacitance Cp8 now appears from
plate to cathode and is effectively in par-
c2 Zz allel with Cpk. This parallel combina-
tion is denoted C2• Also, from the
discussion in Sec. 2-6, the shielding
action of the screen is such that the
FIG. 4-7. The ideal equivalent cir- capacitance Cpo between grid and plate
cuit of a tetrode amplifier.
is very small. If this capacitance is as-
sumed to be negligible, and it is less than 0.001 p,p,f in the average voltage
tetrode, then Fig. 4-6b may be redrawn in the form shown in Fig. 4-7. In
this figure, the capacitances have the values

C1 :: Cyk + Cgs } (4-1S)

C2 - Cps + Cpk
The input admittance of the tube is then
Yi = jwC 1 (4-19)
The mere substitution of a tetrode for a triode may not result in a very
SEC. 4-7] BASIC AMPLIFIER PRINCIPLES 63
marked improvement in the amplifier response. This follows from the
fact that the stray and wiring capacitances external to the tube may allow
significant grid-plate coupling. It is necessary that care be exercised in
order that plate and grid circuits be shielded or widely separated from
each other in order to utilize the inherent possibilities of the tube.
4-7. Input Admittance of a Pentode. The discussion in Sec. 2-6
showed that, even though the tetrode had a significantly smaller grid-
plate capacitance than the triode, the presence of the screen grid was
accompanied by the effects of secondary emission from the plate when
the instantaneous plate potential fell below the screen potential. As

FIG. 4-8. The equivalent circuit of a FIG. 4-9. The current source equivalent
pentode amplifier. circuit of the pentode amplifier.

discussed, the effect of this is overcome by the insertion of a suppressor

grid between the screen grid and the plate. ~,
When used in a circuit as a voltage amplifier, the pentode is connected
in the circuit exactly like the tetrode with the addition that the suppressor
grid is connected to the cathode. By drawing the complete equivalent
circuit of the pentode amplifier, by appropriately extending the rules of
Sec. 3-3, and by including all tube capacitances, it is easy to show that the
equivalent circuit reduces to that shown in Fig. 4-8. In this diagram

Cl = Cgk + CUB }
(4-20)
C2 = Cpk +C +C
P8 p3

where Cp3 is the plate-grid No.3 capacitance.

The plate load impedance Zz is frequently much smaller than the plate
resistance of the tube, and it is convenient to use the current-source
equivalent circuit representation of the tube, as shown. For the range
of frequencies over which the input and output capacitances C l and C 2
are negligible, and with rp» Zz, the total generator current passes
through Zz. Under these circumstances the output potential is
E pk = -gmEgZZ
and the gain is gi ven by the simple form
K = -gmZz (4-21)
64 ELECTRON - TUBE CIRCUITS [CHAP. 4

If the assumed conditions are not valid, then the gain becomes
(4-22)

where Z is the combined parallel impedance in the output circuit.

4-8. Voltage Sources for Amplifiers. A number of different potential
sources are required in an amplifier. These are the following: the
filament, or A, supply; the plate, or B, supply Ebb; the grid-bias, or C,
supply Eco; the screen supply E cc2 . These potentials are supplied in
different ways.
The Filament, or A, Supply. The most common method of heating the
cathodes of indirectly heated tubes is from a low-voltage winding on a
transformer which operates from the a-c supply lines. Storage batteries
may be used if d-c heating is necessary, but this is ordinarily not necessary
except in special applications. Special low-drain tubes are available for
use in portable radio sets and are fed from dry batteries.
The Plate, or B, Supply Ebb. Most equipments involving the use of
electron tubes are operated from the a-c supply mains, and the d-c plate
supply is then secured by means of a rectifier and filter unit (see Chaps.
13 and 14 for details). For applications with severe requirements on
regulation or low ripple, the power supply must be electronically regu-
lated. For low-drain requirements, dry batteries may
be used.
The Grid, or C, Supply Ecc. The grid circuit of
most amplifiers ordinarily requires very little current,
C.k and hence low-power dry batteries may be used.
In most cases, however, self-bias is used (although this
FIG. 4-10. Ob- is restricted to class A and class AB amplifiers). Self-
taining self-bias by
means of a cath-
bias is achieved by including a resistor Rk in the cath-
ode resistor. ode of the amplifier tube and shunting this resistor
with a capacitor Ck, the reactance of which is small
compared with Rk over the operating frequency range. The quiescent
current lb flows through this resistor, and the potential difference provides
the grid bias. The correct self-biasing resistor Rk = Ecc/ lb.
The capacitor Ck serves to by-pass any a-c components in the plate
current, so that no a-c component appears across the resistor R k • If such
an a-c component, or varying bias, does exist, then clearly there is a
reaction between the plate circuit and the input circuit. Such a "feed-
back" effect will receive detailed consideration in Sec. 5-7. If this effect
is to be avoided, large-capacitance condensers may be required, particu-
larly if the frequency is low. High-capacitance low-voltage electrolytic
capacitors are available for this specific service and are quite small
physically.
SEC. 4-8] BASIC AMPLIFIER PRINCIPLES 65

The Screen Supply E cc2 . The screen supply is ordinarily obtained from
the plate-supply source. In most cases the screen potential is lower
than the plate supply, and it is usual practice to connect the screen to the
plate supply through a resistor. The resistor is chosen of such a size
that the potential drop across it due to the screen current will set the
screen at the desired potential. A capacitor is then connected from the
screen to the cathode so as to maintain this potential constant and inde-
pendent of B-supply variations or variations in the screen current.

Decoop/;ng
filfer -

FIG. 4-11. Resistance-capacitance coupled amplifier, with self-bias, decoupling

filters, and screen dropping resistors.

It is customary to use a common B supply for all tubes of a given

amplifier circuit. Because of this, the possibility for interactions among
the stages through this common plate supply does exist and might be
troublesome unless the effective output impedance of the power-supply
unit is very small. It is necessary in some applications to include RC
combinations known as decoupling filters so as to avoid this interaction.
A typical resistance-capacitance coupled-amplifier circuit which is
provided with self-bias, decoupling filters, and screen dropping resistors
is illustrated in Fig. 4-11.
PROBLEMS
4-1. Two waves, one of amplitude 10 volts and frequency 1,000 cps, the second
of amplitude 5 volts and frequency 3,000 cps, are applied to the input of a certain
network. The two waves are so phased that they both pass through zero in the
positive-going direction together.
a. Sketch the resulting input voltage.
b. Suppose that the fundamental component suffers a phase delay of 10 deg
on the fundamental scale and that the third-harmonic component suffers a
50-deg delay on the third-harmonic scale, although neither amplitude is
effected. Sketch the output wave.
y 4-2. a. The output potential of a given amplifier is 18 volts, when the input
voltage is 0.2 volt at 5,000 cps. What is the decibel voltage gain of the amplifier?
 66 ELECTRON - TUBE CIRCUITS [CHAP. 4

b. The output voltage is 7 volts when the input voltage is 0.2 volt at 18,000 cps.
By how many decibels is the response of the amplifier at 18 kc below that at 5 kc?
4-3. Prepare a table giving the power sensitivity of the following tubes (assume
that the output power and the grid excitation are those specified in the tube
manual): 6A3, 6F6, 6V6, 6L6, 6AG7.
,,/ 4-4. An a-c excitation potential of 5 volts rms at a frequency of 2,000 cps is
applied to a 6J5 tube for which /L = 20, rp = 7,700 ohms. The load is a pure
resistance of 15,000 ohms. Calculate the following:

a. The a-c current in the plate circuit.

b. The a-c output voltage.
c. The gain of the amplifier.
d. The a-c power in the load resistor.

4-5. Repeat Prob. 4-4 if the load is an inductive reactance of 15,000 ohms.
,4-6: A type 6SF5 high-mu triode is operated as a simple amplifier under speci-
fied-conditions at 30,000 cps. The important factors are

/L = 100 rp = 66,000 ohms COP = 2.4 /L/Lf COk = 4.0/L/Lf

Cpk = 3.6 p/Lf

a. Calculate the input capacitance and the input resistance of the tube alone
when the load is a resistor Rz = lOOk.
b. Repeat when the load impedance is of the form 60,000 +
j60,000 ohms.

4-7. A type 6J5 triode is operated as a simple amplifier under specified condi-
tions at 22,000 cps. The important factors are

/L = 20 rp = 7,700 ohms COP = 3.4 /L/Lf

Cpk = 3.6/LJ.l.f

a. Calculate the input capacitance and the input resistance of the tube when
the load is a resistor Rz = 20k.
b. Repeat when the load is an impedance of the form 10,000 jlO,OOO. +
4-8. A type 6J5 tube is operated in the circuit of the accompanying diagram.
Calculate the output voltage. (See Prob. 4-7 for the important factors of the
tube.)

4-9. Show that Fig. 4-8 does represent the complete equivalent circuit of the
pentode.
4-10. A type 6SJ7 pentode is operated as a simple amplifier under specified
conditions.
 BASIC AMPLIFIER PRINCIPLES 67
a. When connected as a pentode, with Rl = 25k , the important factors are
gn. = 1,575,umhos rp = 0.7 megohms Cup = 0.005,u!lf Cinvut = 6.0 J.!,uf
Coutvut = 7.0 ,u,uf
Calculate the input capacitance of the amplifier.
b. When' this tube is reconnected as a triode, the factors become
J.! = 19 rp = 8,000 ohms

Calculate the input capacitance ~th Rl = 25 k and compare with the results
of part b.
/4-11. A6AC7 pentode is to be used as a class A amplifier with Eb = 250 volts,
Determine the value of the self-biasing cathode resistor to set Eee = -2 volts;
the screen dropping resistor to set Ee2 at 150 volts if Ebb = 350.
 CHAPTER 5

UNTUNED VOLTAGE AMPLIFIERS

I T IS frequently necessary to achieve a higher gain in an amplifier than

is possible with a single stage. Amplifier stages may be cascaded to
achieve this higher gain, the output voltage from one stage serving as the
input voltage to the next stage. Ordinarily, however, two or three stages
in cascade serving a common function is about the limit for stable opera-

ECC1 Eccz
FIG. 5-1. Schematic diagram of a two-stage RC coupled amplifier.

tion, although six- and nine-stage amplifiers are common in radar-receiver

practice. Extreme caution is required in the design of such multistage
amplifiers.
To calculate the over-all gain and frequency response of such a multi-
stage amplified, the equivalent circuit of the amplifier must be drawn.
The rules for accomplishing this are given in Sec. 3-3. The resultant
equivalent network is then analyzed as a conventional problem in a-c
circuit analysis.
A variety of coupling networks between the cascaded stages are
possible,and a few have become very common, either by virtue of their
simplicity or because of some especially desirable characteristic. A
number of the more common types will be considered in some detail.
5-1. Resistance-Capacitance eRC) Coupled Amplifier. The resist-
ance-capacitance eRe) coupled amplifier, illustrated in Fig. 5-1, is one
of the more common and more important amplifier circuits. This ampli-
fier circuit is used when a sensibly constant amplification over a wide
range of frequencies is desired. By the use of tubes with high amplifica-
tion factors, it is possible to achieve a gain of 50 or more per stage. It
will be found that high-gain triodes possess certain inherent disadvan-
tages, and it is frequently desirable to use pent odes instead. If pentodes
68
SEC. 5-1] UNTUNED VOLT AGE AMPLIFIERS 69

are used, the screen potential must remain constant; otherwise the follow-
ing analysis will no longer be valid.
The capacitors C l , C2 , and Cs in this schematic diagram are known as
coupling or blocking capacitors and serve to prevent any d-c potentials
that are present in one stage from appearing in another stage. That is,
capacitor Cl serves to prevent any d-c potential in the input from
appearing across the grid resistor R gl and thus changing the d-c operating
level of the amplifier. Capacitor C2 serves a similar function in coupling
stage 1 to stage 2. The value of the coupling capacitors is determined
primarily by the I-f amplification. They ordinarily range from about
'0.001 to 0.1 fJ.f for cO!lventionaJ.~::! stages. The resistor R g , which is

K
FIG. 5-2. The equivalent circuit of the RC amplifier of Fig. 5-1.

known as the grid resistor, furnishes a path by which the grid-bias supply
is applied to the grid. It also serves as a leak path through which any
electrons that may be collected by the grid from the electron stream
within the tube may be returned to the cathode. If such a leak path
were not provided, the grid would acquire a negative potential with the
collection of the electrons, thus influencing the operation of the tube. A
negative bias-supply potential is ordinarily used, and the grid current is
usually very small. This permits the use of relatively large resistors for
Rq, say from 50 k to 2M. Large values of Rg are desirable in achieving a
wide frequency response. The load resistor Rz is determined principally
by the gain and th~dI!29J.lency Q!!!HL!Yidth_tha,UadesiJed.
The equivalent circuit of the amplifier of Fig. 5-1 is shown in Fig. 5-2.
This circuit is valid for triodes, tetrodes, or pentodes provided that the
screen potential of the latter two is maintained constant. In this circuit
Eg denotes the a-c input voltage applied to the grid of the first stage.
This potential appears across the parallel combination con,sisting of the
resistor Rg in parallel with the input impedance to the amplifier. The
input impedance of the stage is considered to be a resistor (and assumed
positive) in parallel with the input capacitance. However, it is supposed
that the impedance of the driving source is low and that the loading
by the total input impedance of the first stage does not affect the input
potential. The output circuit of the first stage consists of the load
resistor, the coupling capacitor C 2 , and the total input impedance of the
70 ELECTRON- TUBE CIRCUITS [CHAP. 5

second stage. This is denoted as Rg and Cg for the total resistive and
capacitive components. The output of the amplifier is the potential
difference across the output impedance, which is denoted by the symbol
Z. This impedance cannot be specified more completely until the nature
of the output circuit is known.
The coupling between the grid and the plate of the tubes through the
interelectrode cap"acitances can be neglected over a wide frequency range
with pent odes and over the a-f range with triodes. Consequently each
-stage may be considered as independent of the following stage, but the
output of one stage is the input to the next stage. As a result, it follows
that since
Kl = E g2k = output voltage of 1st stage
E g1k input voltage to 1st stage
and
K = Eg3k = output voltage of 2d stage = output voltage of 2d stage
2 E g2k input voltage to 2d stage output voltage of 1st stage
then the resultant over-all gain is
K = E g3k = output voltage of 2d stage
E g1k input voltage to 1st stage
It follows from these expressions that
K = KIK2 (5-1)
By taking twenty times the logarithm of the magnitude of this expres-
sion,
20 loglo K = 20 loglo Kl + 20 loglo K2 (5-2)
It follows from this that the total decibel voltage gain of the multistage

(l
c
Gz
amplifier is the sum of the decibel
voltage gains of the separate stages.

R[ Rg e.g
1 This fact is independent of the type
of interstage coupling.
5-2. Analysis of RC Coupled
Amplifier. A typical stage of the
RC coupled amplifier is considered
K
in detail. This stage might repre-
FIG. 5-3. A typical stage of an RC sen t any 0 f th e amp l'fi h'
1 er cam, ex-
coupled amplifier.
cept perhaps the output stage.
Representative subscripts have been omitted. The circuit is given in
Fig. 5-3.
This circuit may be analyzed with the aid of the Millman theorem. A
direct application of the results of this theorem between the points G 2
SEC. 5-2] UNTUNEb VOLTAGE AMpLIFiERS 71

and K yields the expression

Ep1kY c
E g2k = Y c Y R. + Y c. (5-3)
+
where Y c = jwC, Y Ra = l/R g , and Y Ca = jwC g • An application of this
theorem between the points PI and K yields the following expression,
E _ -}LEgYp+Eg2kYc
plk - Yp + Y l + Y c (5-4)

where Yp = l/r p and Y l = l/R l • By combining Eqs. (5-3) with (5-4)

and solving for the gain K = E g2K /E g , there results

(5-5)

This is the complete expression for the voltage gain of such an amplifier
stage. If the constants of the circuit
are known, the gain and phase-shift
characteristics as a function of fre-
quency may be calculated. However,
by making suitable approximations, it
is possible to study the behavior of the
FIG. 5-4. The mid-frequency equiv-
circuit for vario:us limiting conditions alent circuit of the RC amplifier.
of frequency.
Intermediate Frequencies. T9-JLinte~1p.ediate frequencies, or mid-fre-
qU!IDcies, are those for whicl(Y c is large ).ng:
Y C a is smalL SUOJectto
these conditions, the equivaleltt circuit of Fig. 5-3 reduces to that shown
in Fig. 5-4. For the range of frequencies over which this equivalent cir-
cuit is valid, the expression for the gain reduces to
K = Ko = -}LYp (5-6)
Yp + Y + YR.
l

This expression for the gain is independent of the frequency, since no

reactive elements appear in the circuit.
c Since each of the elements is resistive,
the phase angle between input and out-
Rg put voltages is constant and equal to
180 deg.
L-F Region. At the lowJ1:illLllEi-nQiw
FIG. 5-5. The I-f equivalent circuit the effect of Cg is negligible, and Y C a
of the RC amplifier. may be made zero. The- effect of the
coupling capacitor C becomes very iIll:"
12ortant. The equivalent circuit under these conditions has the form
shown in Fig. 5-5. The general expression for the gain [(5-5)] reduces to
72 ELECTRON- TUBE CIRCUITS [CHAP. 5

It is found convenient to examine the l-f gain relative to the mid-frequency

gain. The ratio KdKo becomes
1
(5-8)
Ko - 1 + YRg(Y p + Y 1) {
Y c( Y p + Y 1 + Y Rg)
i

This may be written in the simple form, for any frequency f,

Kl_ 1
(5-9)
-Ko- -_1.- j(fdf)
where
j (5-10)

If the load is a pure resistance, then fl is a real number and the magni-
tude of the relative gain becomes

Kll 1 (5-11)
IKo = VI + (ft/fF
This shows that the parameter fl represents the frequency at which the
gain falls to 1/0, or 70.7 per cent of its mid-frequency value. This
frequency is usually referred to as the l-f cutoff frequency of the amplifier.
The relative phase angle (h is given by

tan (h = it
1 (5-12)

This approaches 90 deg as the frequency approaches zero.

It should be noted that the I-f cutoff value [(5-10)] depends, among
other terms, on the size of the coupling capacitor C. Since the value of
C appears in the denominator of the expression for fl' then for a decreased
I-f cutoff, larger values of C must be chosen. Of course, the gain must
ultimately fall to zero at zero frequency.
There are several practical limitations to the size of the coupling capaci-
tor that may be used. The capacitor must be of high quality so that
any leakage current will be small. Otherwise a conduction path from
the plate of one stage to the grid of the next stage may exist. But good-
quality capacitors in sizes larger than 0.1 p.f are physically large and are
relatively expensive. Also, if the coupling capacitor is large, a phenome-
non known as blocking may result. This ari§~~~he_n_.th,e time constant
CRg is much larger than the period of the highest frequency t~J~~pa"Ssed
by the amplifier. Thus if an appreciable charge flows into the capacitor
 SEC. 5-2] UNTUNED VOLTAGE AMPLIFIERS 73

with the application of the input signal and if this cannot leak off quickly
enough, a charge will build up. This may bias the tube highly negatively,
perhaps even beyond cutoff. The amplifier then becomes inoperative
until .the capacitor discharges. This condition is sometimes desirable
in special electron-tube circuits, and will be the subject of a detailed dis-
cussion in Chap. 7. However, it is a condition that must be avoided in
an amplifier that is to reproduce the input signal in an amplified form.
The grid resistor Rg must be made high to keep the gain high, since
Rg of one stage represents a loading across the plate resistor Rz of the
previous stage. The upper limit to this
value is set by the grid current. Ordi-
narily the grid current is small, par-
ticularly when the grid bias is negative.
But if the grid resistor is made too hiSh,
and several megohms is the usual limit,
i th·· t ·11 t FIG. 5-6. The h-f equivalent circuit
t h e vo tage across IS resIs or WI ac of the RC amplifier.
~puriou!,})i_~s on the tube. While
special low-grid-current tubes are available, these are designed for
special operations and would not ordinarily be used in conventional
I. circuits.
H -F Region. At the high frequencies, the admittance of C is very
large, and the admittance of Cg becomes important. The equivalent
circuit corresponding to these conditions becomes that shown in Fig. 5-6.
The general expression for the gain reduces to
K - K - -}LYp (5-13)
- 2 - Yp + Y z + Y Rg + Y c.
The gain ratio KdKo becomes
K2 1
(5-14)
Ko = 1 +' Y e•
Yp + Y z + YR.
This expression may be written in a form similar to Eq. (5-9) for the 1-f
case. It becomes
K2 1
(5-15)
Ko 1 + j(f/!2)
where
/ (5··J 6)

In this expression Cg denotes the total capacitance from grid to cathode

and comprises the iii.illii:)apacitance of the following_.stage, the_ o'!!put
,Tiring, ~~_!ll~__output tu_be capacitance.
 74 ELECTRON- TUBE CIRCUITS [CHAP. 5

If the load is a pure resistance, then J2 is a real number and the magni-
tude of the relative gain becomes

(5-17)

It follows from this that f2 represents that frequency at which the h-f
gain falls to 1/O, or 70.7 per cent, of its mid-frequency value. This
frequency is usually referred to as the h-f cutoff of the amplifier. The
relative phase angle (J2 is given by

(5-18)

This angle approaches - 90 deg as the frequency becomes very large

compared with}2.
Note from Eq. (5-16) that the h-f cutoff value depends on the value of
Cg, among other factors. Sin~e the value of Cg appears in the denomina-
tor of t1!EL~I!ression, t1!~nclearly a hig~E::-f cutoff reql!!!:.es _~ sll!l:!:ll yalue
of Cfl. ___ Moreover, since the input capacitance of a pent ode is appreciably
less than that of a triode, the pent ode possesses inherently better possi-
bilities for a broad frequency response than does the triode. It will be
found in Sec. 6-5, in the discussion of the cathode-follower amplifier,
that triodes with cathode-follower coupling stages also possess broad-band
capabilities, although this is accomplished at the expense of a tube.
_N_o.1e__ll:~~~Elthat _tlt~ h-f cutoff is imp!..():ved by t~()_llse__ of lar~LYll and
Y R , which implies the use of small values of resistance and a tube with a
_~Ill~ll_plate re~istance.-- ----------- -----------

210 ~ 1.0
~ ~
-£i _t;:;o
., 180 :s,0.5
I'
I ,----------- I
Vl
IS.,,= '" I1 '\I
if ~ 1 'k
90 0 L----,+-I- - - - - - - _ + 1-'-~f
c::
fi £2
FIG. 5-7. A typical frequency response and phase characteristic of an RC coupled
amplifier.

5-3. Universal Amplification Curves for RC Amplifiers.! The fore-

going analysis shows that the gain of an RC coupled amplifier is sub-
stantially constant over a range of frequencies and falls off at both the
high and the low frequencies. A typical frequency-response curve has
the form sketched in Fig. 5-7.
Since the relative gain and the relative phase-shift characteristics
depend only upon the two parameters f1 and h it is possible to construct
SEC. 5-4] UNTUNED VOLTAGE AMPLIFIERS 75
curves which are applicable to any such amplifier. Such universal curves
are given in Fig. 5-8.
The frequency-response characteristics of any RC coupled amplifier
can -easily be obtained with the aid of these curves. The first step in the
analysis is to calculate the values of the parameters 11 and 12 from Eqs.

90
1.0
-r--. ~
t-. Phase
80 0.8
./
70 .!:: V
-!60 &0.6 1"'-
~ 50 ~
J5,.
~ 40 ]0.4
.c
CL 30
OJ
0::: "-
"
--
/'
20 0.2
10 V 6ain
~ I--
o o0.1. 0.2 0.3 0.5 1.0 2 3 5 10 20 30 50 100
Frequency r(;ltio ~/f or flEz
FIG. 5-8. Universal gain and phase-shift curves for an RC coupled amplifier.

(5-1:0) and (5-16). Then the values of the relative gain and the relative
phase angle are obtained from the curves for a number of values of the
ratio hi! and !1!2. These are plotted as a function of f. It must be
remembered in using Fig. 5-8 that the ordinate is KIIKo or fh when the
abscissa ishlf. Also, the ordinate is KdKo or (h when the abscissa is!I!2.

Ecc~ Bcc;!
FIG. 5-9. Schematic diagram of an inductance-capacitance coupled amplifier.

6-4. Impedance-Capacitance (LC) Coupled Amplifier. The circuit of

the impedance-capacitance coupled amplifier differs from that of the
resistance-capacitance coupled amplifier only in the use of an inductance
plate load instead of a plate resistor. The schematic diagram of the
amplifier is given in Fig. 5-9. The use of an inductance instead of a
resistance in the plate circuit makes possible the use of a smaller plate-
supply voltage for a particular tube operating condition, since the d-c
resistance of the inductor is small and the d-c voltage drop across this
inductor is also small.
76 ELECTRON - TUBE CIRCUITS [CHAP. 5

The equivalent circuit of one stage of this amplifier is given in Fig.

5-10. Observe that this circuit differs from the corresponding equivalent
circuit of Fig. 5-3 only in the plate-circuit impedance. Consequently
the analysis leading to Eq. (5-5).is
c valid in the present case provided
that the plate-circuit admittance is
interpreted to be
Cg
Yl = jwC L + RL ~ jwL (5-19)

Here CL is the distributed winding

FIG. 5-10. The equivalent circuit of acapacitance, which has been assumed
typical stage of an LC coupled
to shunt the inductance, and RL is
amplifier.
the resistance of the inductor.
The frequency-response characteristic of this amplifier may be examined
in the same way as that for the RC system. The analysis shows that the
response is generally similar to that for the RC amplifier, except that the
sensibly flat region is narrower, although the mid-frequency gain is
somewhat higher. T~Kai!l__drops off more rapidly than for the RC

FIG. 5-11. Schematic diagram of a transformer-coupled amplifier.

amplifier at the low frequencies owing to the low reactance of L in addition

to the high reactance of 2. The gain drops off more rapidly than for the
RC amplifier at the high fregQ~!l:~i~l' because of the shunting effects of
bQth CL and C g • The mid-frequency gain is higher than for the RC
amplifie-;" p~~Vided that the impedance of the inductor at these frequencies
is higher than the resistance of the plate load of the RC amplifier.
The LC coupled amplifier is seldom used as a voltage amplifier owing
to the narrow frequency band and the cost of the inductor. This type
of coupling, which is also referred to as shunt- or parallel-feed coupling, is
frequeptly used in power amplifiers. ----.........--.._..._..
5-5. Transformer-coupled Amplifiers. The circuit of a transformer-
coupled amplifier is given in Fig. 5-11. The transformer as a coupling
device possesses several desirable features. Owing to the step-up
character of the transformer, a total amplification per stage greater
than the jJ. of the tube can be achieved. Also, the d-c isolation provided
by the transformer automatically removes the requirement for a blocking
 SEC. 5-5] UNTUNED VOLTAGE AMPLIFIERS 77

capacitor. High-quality interstage transformers generally have voltage

ratios of 1: 3 or less. Higher transformation ratios usually are aCCOl1l-
, panied by distributed winding capacitances and by interwinding capaci-
tances that are excessive. The effects of these capacitances will be
considered below.
The equivalent circuit of a typical transformer-coupled stage is given
in Fig. 5-12 .
. Mid-frequency Response. An approximate expression for the gain per
stage may readily be found if it is assumed that the transformer is ideal.
eM

(aJ (bJ
FIG. 5-12. The equivalent circuit of a typical transformer-coupled stage.

An ideal transformer is one for which unity coupling exists between

primary and secondary windings and in which the !9sSei:l and stray
ca2acitances are negligible. If it is assumed that the secondary of
the transformer, which feeds the grid of the following stage, is essen-
tially open-circuited, then the plate-circuit impedance is infinite.
Consequently the full effective voltage j.l.E g of the stage appears across the
transformer primary. The corresponding output voltage at the second-
ary terminals is nj.l.E g , where n is the transformation ratio of the trans-
former. The gain of the stage is then simply

K = Ko = nj.l. (5-20)

a constant, independent of the frequency.

L-F Region. The foregoing considerations are only approximate, owing
to the character of the assumptions
a br---JV'VVv----+ t1J.a-t were made. To examine t he
,-----'WVv---..

1",

P f;;\
R1
LJ L2
Rz f
response at the. lower f.requencies,
E2 the effective d!s.!:.ri~uteg _'''.inding
" \ -I. ) capacitances and l~a~~ge induc-
+ 'pEg tance's may be neglected. The
K -'.-.
FIG. 5-13. The I-f equivalent circuit of a corresponding equiv~lent. circuit
transformer-coupled amplifier. has the form shown III FIg. 5-13.
The resistance of the primary
winding is denoted as R 1 , and its inductance is L 1 ., Similarly, R2 and L2
represent .the secondary resistance and inductance 7 respectively.
 78 ELECTRON-TUBE CIRCUITS [CHAP. 5

It iB evident from thiB diagram that the primary current is

11 = p,Eg
Tp + Rl +jWLl
The potential difference across the primary inductance is
E - -p,Eg (5-21)
ak - 1 _ j (TP + Rl)
wL 1
But the potential that appears across the secondary winding Ebk is n
times as large as E ak. The secondary voltage may be in phase or 180 deg
out of phase with Eak and depends upon the relative winding direction
of the primary and secondary windings. The gain of the transformer-
coupled stage is then
Ebk_
K 1-- - - (5-22)
Eg

The ratio of I-f to mid-frequency gain is then

K1 + 1 (5-23 )
Ko = - 1 _ j (TP + R1)
wL 1
which may be written in the form
K1_ 1
Ko - ± 1 - j(jdf) (5-24)
where
(5-25)

Observe that the gain ratio has the same general form as for the RC
coupled amplifier. N ow the gain drops off at the lower frequencies
because the reactance of the primary winding decreases and no longer
will be large compared with the total resistance of the primary circuit
Tp + R 1•
Equation (5-25) contains. an explanation why high-gain tubes, which
inherently possess high plate resistances, are not employed in transformer-
coupled amplifiers. If such high-gain tubes were used, then the I-f
.; response would be very poor. If an attempt were made to improve the
I-f response by designing the transformer to have a very high primary
inductance, the associated distributed capacitances would become exces-
sive. The alternative method of obtaining high primary inductance
through the use of core material of very high permeability has led to the
development of high-permeability alloys such as permalloy and hiperm.
H-F Region. At the high frequencies the shunt(ng effects of the
primary and secondary windings may be neglected. However; the effects
SEC. 5-5] UNTUNED VOLTAGE AMPLIFIERS 79
of the leakage inductances, and also the interwinding and distributed
capacitances, are important. The circuit of Fig. 5-12 appears in Fig.
5-14a with the transformer replaced by its T-equivalent, and referred to
the primary. At the high frequencies the circuit reduces to that of
Fig. 5-14b, in which the winding, interwinding, interelectrode, and stray
wiring capacitances are lumped into an equivalent capacitance C across
the primary of the transformer.

(a)
FIG. 5-14. The C'omplete and approximate h-f equivalent circuit of a transformer-
coupled stage.

The total effective shunting capacitance is related to the several com-

ponents by the approximate expression
(5-26)
The ± sign depends on the relative disposition and connection of the
primary and secondary coils. When properly connected, the minus
sign usually applies. To justify this expression, examine Fig. 5-14a. If
the voltage across the input terminals is E 1, and that across the output
terminals is E 2, the difference of potential across Cm is
ECm = El ± E2 ~ E 1 (1 ± n)
where n = N2/N 1• The energy stored in this capacitance per cycle is

W= Cm:~m = ;m (1 ± n)2E~
The equivalent capacitance across the primary terminals which will store
the same energy per cycle will be (1 ± n)2Cm • The other capacitances in
the secondary are reflected into the primary as n2(C 2 +
Cg). The total
shunting capacitance is that given in Eq. (5-26).
An analysis of the approximate equivalent circuit yields for the gain
of the amplifier the expression
jnp.Xc
K2 = ± R +
j(XL _ Xc) (5-27)

where R = rp +
Rl +
R2/n2; L = L~ +
LUn2; XL = wL; Xc = 1/wC.
The primary leakage inductance is L~, and the secondary leakage induct-
ance is L~. The gain ratio may be written in the form
K2 + . jX c (5-28)
Ko - - R + J(X L - Xc)
80 ELECTRON- TUBE CIRCUITS [CHAP. S

The magnitude of the gain ratio is

11: I = yR2 + ~L - X C)2

(5-29)

At the lower end of the region of frequencies where this analysis it;
valid, XL is small and Xc is large, so that the gain ratio approachet; unity,
as it should. At the higher frequencies, Xc is small, XL is large, and the
gain falls to zero. Notice, however, that the secondary circuit may pass
through a maximum, owing to a resonance condition that exist:,;. The
maximum is found to occur when
_ 2Xl + R2 \.
Xc - 2X j ,

Usually R will be much smaller than XL, and the maximum occurs when
Xc = XL, the condition for series resonance. But as the frequency at
which this resonance occurs is
]
Wo =--
VW
then the corresponding value of the gain is

(1:)=x = wo~R = ~ ~ (5-30)

A typical frequency-response curve, the dependence on the primary

inductance L 1, and the total leakage inductance L are illustrated III
Fig. 5-15.
Max

~----------------------L-------~f
FIG. 5-15. A typieal frequcncy-rcsponse characteristie of a transformer-couph d
amplifier.

The peaking of the transformer-coupled-amplifier frequency-response

curve may be suppressed to a considerable extent by several methods.
The simplest way consists in shunting the secondary of the transformer
with an appropriately chosen resistance. This reduces the height of the
resonant peak, although it also causes the mid-frequency gain to be less
than np.. Another method for improving the response characteristic is
in the design of the transformer. In modern transformers the t;hunt
 SEC. 5-6] UNTUNED VOLT AGE AMPLIFIERS 81

capacitance is decreased by decreasing the contributing factors. The use

of a grounded shield between windings reduces the interwinding capaci-
tance. The design of the windings reduces the distributing capacitances.
Further, the leakage inductance is reduced, so that the peaking will occur
at frequencies beyond the normal operating range of the amplifier. This
reduction has been accomplished both by proper design of the windings
and by use of suitable high-permeability-core materials. Also, the use
of a high-resistance secondary winding tends to suppress the resonant
peak. A high-resistance primary winding is to be avoided owing to the
adverse effect on the I-f response. High-quality transformers are avail-
able that are flat with 1 db over a range of frequencies from approximately
20 to 10,000 cps. The peaking may also be reduced materially through
the use of inverse feedback in the amplifier, as will be shown below.
Transformers are seldom used as interstage coupling devices merely to
obtain higher gain. This follows because the use of a pent ode in an RC
coupled amplifier will ordinarily provide a higher gain than is possible
with a triode with a step-up transformer. Furthermore, the RC coupled
stage requires less space and makes use of relatively inexpensive equip-
ment. The principal uses of transformers are as the coupling stage
between the driver and a push-pull amplifier, so as to provide the required
two voltages that are 180 deg apart in phase (although this application is
largely being supplanted by tube circuits), and also as the output trans-
former in a power amplifier. Here the principal function is one of imped-
ance matching and permits the matching of low impedance loads to
high-internal-resistance tubes. These applications will be discussed in
some detail in Chap. 9.
5-6. Feedback in Amplifiers. 2 When a part of the output signal is
combined with the input signal, feedback is said to exist. If the net
effect of the feedback is to increase the effective input signal, the feedback
is called positive, direct, or regenerative. If the resultant input signal is
reduced by the feed-back voltage, the feedback is called negative, inverse,
or degenerative.
The application of negative feedback to an amplifier results in a number
of characteristics that are highly desirable in the amplifier. It tends to
U flatten theJre.quellii.Y=r~s.QQ!!Sf? ch~r~gteristic and t6'-Bxtend the range of
;\uniform r~§12@tle. -'!tIt materially reduces nonlinear and phase distortion.
Jlt impI.QYE!s ~h~stabilityof the am~lifi.er, ma~i.ggjop.e gl1i!L~sS dependent
on the_r3Perating. voltages or on variations of the tUl:>e characteristics.
Also, it~tends to make. the gain less dependent on thJLlQad., so that load
variations do not seriously influence the operating characteristics of the
amplifier. The use of feed-back networks of special design will provide
selective attenuation, thus permitting a frequency response of desired
,,-,characteristics.
82 ELECTRON - TUBE CIRCUITS [CHAP. 5

The application of positive feedback has the opposite effects. Thus

positive feedback tends to sharpen the frequency-response curve and to
decrease the range of uniform response. This permits an increased gain
and selectivity. Positive feedback in an amplifier is critical of adjust-
ment. Too much regenerative feedback in any system may result in
oscillation. Ordinarily, negative feedback is more common than positive
feedback in amplifiers.
The action of a feed-back path depends upon the frequency of opera-
tion. That is, the feedback may remain regenerative or degenerative
G p
+ Amplif/er
E.9,t Egler nominal gain t Eple (oulpvf)
K .j,
7i IK

'r- feedback nefworR -

t
jlE.'pk~ Feedback fraction
fJ
FIG. 5-16. The principle of feedback in amplifiers.

throughout the range of operation of the circuit, although the magnitude

and phase angle of the feed-back signal may vary with frequency. It is
also possible for the feedback to be positive over a certain range of fre-
quencies and negative over another range.
The principle of feedback is illustrated in the schematic diagram of
Fig. 5-16. For simplicity, series injection is shown at the input, but
other forms of network coupling may be employed. In the diagram
shown, a voltage Eg is applied to the input terminals of the amplifier,
with the polarity as shown. Suppose that the resultant voltage at the
output terminals is E pk • Suppose that a fraction {3 of this output is fed
back in series with the input signal in such a way that the resultant signal
that appears between the grid-cathode terminals has the form
E gk = Eg + {3E pk (5-31)
But since the nominal gain of the amplifier is, by definition,
K = output potential _ Epk
- input potential between grid and cathode - E gk
then
E pk = KE gk (5-32)
By combining this with Eq. (5-31), there results
E pk = KEg + K{3Epk
from which it follows that
KEg
E pk = 1 - K{3 (5-33)
SEC. 5-6] UNTUNED VOLT AGE AMPLIFIERS 83

But the resultant gain of the amplifier including the effects of feedback is
defined as
K = output potential Epk
r - input-signal potential = Eg
Therefore it follows that
K
Kr = 1 -\K{3
(5-34)

This equation expresses the resultant gain of the amplifier with feedback
in terms of the nominal gain K of the amplifier without feedback, and the
feed-back fraction {3. It should be noted that the three quantities K r , K,
and {3, which appear in this equation, may be complex quantities.
Suppose that the feedback is negative and that the feed-back factor
K{3 is made large compared with unity. The resultant gain equation
(5-34) becomes

(5-35)

This means that when I K{3 I » 1, the actual amplification with negative
feedback is a function of the characteristics of the feed-back network
only. In particular, if {3 is independent of frequency, then the over-all
gain will be independent of the frequency. This permits a substantial
reduction of the frequency and phase distortion of the amplifier. In
fact, by the proper choice of feed-back network, it is possible to achieve
almost any desired frequency characteristic.
Note that if K{3» 1, then Kr == K/K{3« K, so that the over-all gain
of the amplifier with inverse feedback is less than the nominal gain
without feedback. This is the price that must be paid to secure the
advantages of negative feedback. This is not a serious price to pay,
since the loss in gain can be overcome by the use of additional tubes.
Clearly, if K{3 is greater than unity, then Eq. (5-35) shows that the
over-all gain will not change with tube replacements or with variations
inbattery potentials, since {3 is independent of thetube:~Even if Eq.
(5-35) is' not completely valid, a substantial Improvement results in
general stability. This follows from the fact that a change in the nominal
gain dK for whatever reason results in a change dK r in the resultant gain
by an amount
(5-36)

where I 1 - K{3 I represents the magnitude of the quantity 1 - K(3.

This equation is the logarithmic derivative of Eq. (5-34). In this expres-
sion, dKr/Kr gives the fractional change in K r, and dK/K gives the
84 ELECTRON- TUBE CIRCUITS [CHAP. 5

fractional change in K. If, for example, the quantity [I - K6 [ = 5 in

a particular feed-back amplifier, then the variation in any parameter that
might cause a 5 per cent change in the nominal gain will result in a change
of only 1 per cent in the resultant gain of the amplifier.
One effect was omitted in the above considerations. It was implicitly
assumed that the dynamic curve was linear and that the output voltage
was of the same wave shape as the input. If an appreciable nonlinear
distortion exists, then the output contains harmonic components in
addition to the signal of fundamental frequency. Suppose, for simplicity,
that only a second-harmonic component B2 is generated within the tube
when a large signal voltage is impressed on the input. Because of the
feedback, the second-harmonic component B~ that appears in the output
is different from that generated within the tube. To find the relationship
that exists between B~ and B 2, the procedure parallels that for the gain
considerations. Thus, for a second harmonic B~ in the output, a fraction
{3B~ is supplied to the input. As a result, the output actually must con-
tain two components of second-harmonic frequency, the component B2
that is generated within the tube and the component K(3B~ that arises
from the signal that is fed back to the input. This requires that
K{3B~ + B2 = B~
or
(5-37)

Note that since both K and (3 are functions of the frequency, in general,
the appropriate values that appear in this equation must be evaluated at
the second-harmonic frequency.
It should be pointed out that this derivation has assumed that the
harmonic distortion generated within the tube depends only upon the
grid swing of the fundamental signal voltage. The small amount of
additional distortion that might arise because a fraction of the second-
harmonic component is returned to the input has been neglected. Ordi-
narily this procedure will lead to little error, although a more exact calcula-
tion taking these successive effects into account is readily possible. 3
Another feature of Eq. (5-37) should be noted. According to this
expression, if I 1 - K(3 I = 10, then the second-harmonic distortion with
feedback is only one-tenth its value without feedback. This is the situa-
tion when the total output-voltage swing is the same in each case; other-
wise the harmonic generation within the tube could not be directly
compared. This requires that the signal, when feedback is applied,
must be 11 - K(3 I times that in the absence of feedback. As a practical
consideration, since appreciable nonlinear distortion is generated only
when the signal voltage is large, then the full benefit of the feed-back
•
SEC. 5-7] UNTUNED VOLT AGE AMPLIFIERS 85

amplifier in reducing nonlinear distortion is obtained by applying nega-

tive feedback to the large-signal stages.
5-7. Feed-back Circuits. The voltage fed back from the output of the
amplifier may be proportional either to the voltage across the load or to
the current through the load. In the first case the feedback is called

EI
i

FIG. 5-17. Circuits employing current feedback.

voltage feedback; in the second case, current feedback. In either case, the
feedback may be positive or negative, depending upon the connection.
An amplifier may possess any combination of voltage and current
feedback.
Two examples of amplifiers with current feedback are shown in Fig.
5-17. In the circuit of Fig. 5-17 a, the feedback is effected through the
resistor in the cathode of the amplifier and is frequently referred to as
.cathode degeneration. In the circuit of Fig. 5-17b, the feedback is
effected through a resistor in the load circuit, the total feed-back potential
depending on the feed-back resistor R,.
The feed-back fraction is of the same form in both circuits and is
Rk/Z in the first circuit and R,/Z in the second circuit. To find the
expressions for the gain, it is noted that the equi~alent circuit for both
circuits have the same form, as shown in Fig. 5-18.

FIG. 5-18. The equivalent circuits corresponding to the current-feed-back circuits of

Figs. 5-17.

Two methods of analysis exist. One method effects the solution by

direct application of the general principles of electron-tube circuit analy-
sis. The second method makes use of the feed-back considerations
implied by the general equation of the feed-back amplifier [Eq. (5-34)].
The solution by both methods will be given.
According to the general principles of electron-tube circuit analysis,
the output potential is readily found. Thus the current is given by
I = P.Euk
(5-38)
rp + Rk + Zz
 •
86 ELECTRON - TUBE CIRCUITS [CHAP. 5

But the potential E gk is given by

ty
E gk = -IRk + Eg (5-39)
Combining these two expressions gives
I = J1.Eg (5-40)
rp + (J1. + l)Rk + Zl
The output potential E pk is then
E pk = -IZI = -J1.ZzE g (5-41 )
rp + (J1. + l)Rk T Z/
and the gain becomes
(5-42)

To apply the feed-back method, it is noted that the feed-back frac-

tion is
(5-43)
Also, the current is
(5-44)

so that the nominal gain K is

K = E pk = -J1.Z1 (5-45)
E gk rp + Rk + Zl
The resultant gain is then

(5-46)
which is the same as above.
Refer to Eq. (5-40) for the current in the plate circuit. This may be
interpreted to show that the effect
of the negative feedback is to in-
crease the effective internal resist-
R1 ance of the tube from rp without
feedback to the value
FIG. 5-19. Circuit employing voltage rp + (J1. + l)R k•

feedback. This means, of course, that, if the

effective internal impedance of the tube is high compared with the imped~
ance of the external load, the current is effectively independent of variation
of load impedance.
A circuit for achieving voltage feedback is given in Fig. 5-19. In this
 SEC. 5-7] UNTUNED VOLTAGE AMPLIFIERS 87

circuit, the resistor combination Rl + R2 which shunts the output is

made large compared with the load impedance Zz. The capacitor C
has a reactance that is negligible comparEJd with Rl±_R2 at the fre-
[email protected] to be employed. Its sole purpose is to block the d-c potential
from the plate circuit from appearing in the grid circuit.
To obtain an expression for the resultant gain of the amplifier, the feed-
back method will be employed. By neg!~ciiIlgt1:J.e_ s1!uIliil!geffE.lct~f the
- fee.d:cb~QlLresistor O_Il. the lOlJ,.!i!mp~da-.n.Qe_,.jt follows that the nominal gain
of the amplifier is given by
(5-47)
Also •
{3 = Rl (5-48)
Rl + R2
Then the resultant gain with feedback is
K. = K = -J.LZz (5-49)
1 - K{3 rp + Zz + {3J.LZI
This expression may be transformed to the form
-J.L'ZZ
K.= rpI + Z'.II (5-50)
where
I J.L I rp
(5-51 )
J.L = 1+J.L{3 rp =l+J.L{3

This is the gain that would be obtained from a tube whose amplification
factor is l, whose plate resistance is r~, and which feeds an external load
Zz. These results show that the effective amplification factor of the tube
is reduced in the same ratio as the
plate resistance of the tube. This
feature will permit an impedance
match between a tube with a high
plate resistance and a low imped-
ance load. This is accomplished,
of course, at the expense of effec-
. 1 . h . FIG. 5-20. Circuit employing compound
tlve y convertmg t e tube mto a feedback.
triode, with low J.L and low rp.
The combination of current and voltage feedback in an amplifier is
frequently called compound or bridge feedback. The circuit of such an
amplifier is given in Fig. 5-20. The feed-back fraction is readily found
to be
(5-52)
88 ELECTRON - TUBE CIRCUITS [CHAP. 5

As in the analysis of Fig. 5-19, it is assumed that the resistance combina-

tion Rl +R 2 » Zz and that the reactance of the capacitor is negligible
over the frequency range of operation. The resultant gain of the ampli-
fier has the form
-J.l.ZZ
Kr = -rp-+-:--(~J.I.-+"'--:;l""')R~k-:'+'--o-::(l-+-;--J.l.---:{J~l)C-;=;Zz (5-53)

This may be written in the form

K" = -J.I."Z z (5-54)
r r~ + Zz

l
where

= {J.I.~l+ +
•
(5-55)
J.I.:: !p l)Rk
rp - 1 J.I.{Jl

The effect of the feedback is seen to reflect itself as a change in effective J.I.
and rp of the tube. Moreover, owing to the form of the expression for
r~, this quantity may be made greater than, equal to, or less than its
value without feedback.

FIG. 5-21. A two-stage RC coupled amplifier with current feedback in the first
stage and voltage feedback between stages.

Feedback can be effected over several stages and need not be confined
to a stage-by-stage practice. A two-stage RC coupled amplifier employ-
ing current degeneration in the first stage through resistor Rl and voltage
feedback between stages is illustrated in Fig. 5-21. A careful considera-
tion of the application of the potentials will show that both types of
feedback are negative.
6-8. Effect of Feedback on the Output Impedance of an Amplifier.
The discussion in the foregoing sections has shown that the application of
negative feedback in an amplifier is accompanied by an equivalent tube
resistance which varies according to the type of feedback that is employed.
As a consequence of this, the output terminal impedance of the amplifier
iikewise depends on the type of feedback that is employed.
 SEC. 5-9] UNTUNED VOLT AGE AMPLIFIERS 89
A general relationship between the output terminal impedance without
feedback and that with feedback is possible. To find this, suppose that
the input terminals are shorted, and that a generator is applied to the
output terminals. The ratio of the applied potential to the current is
the output impedance. Consider Fig. 5-22, which is Fig. 5-16, a circuit
with voltage feedback, which has been modified for the determination of
output impedance. Note that a voltage appears in the input through the
feedback network even though the input terminals are shorted.
if -the applied voltage to the output terminals is Eo, a voltage j3Eo
will appear at the input terminals to the tube. The equivalent voltage
that appears in the output or plate circuit is Kj3E o. The nominal output
impedance of the amplifier is
designated Z~. The current in
the output circuit is seen to be

I = Eo - ·-;Cj3E o
Zo
The effective output impedance is
FIG. 5-22. The modifications of the gen-
Z Eo Z~ (5-56)
eral feed-back circuit for calculating the
o = To = 1 - Kj3 output impedance of the amplifier.

This expression shows that the output impedance is reduced by the same
factor as the voltage gain, when voltage feedback is applied.
A calculation for a circuit provided with current feedback yields the
expression (see Prob. 5-17)
Zo = Z~ (1 - KZf/Z~~ (5-57)
(1 - KZf/Z L)
where Zf is the feedback impedance, and ZL is the load impedance. In
general, for current feedback, Zo is greater than Z~. For compound
feedback, the output impedance will depend on the relative fraction of
voltage and current feedback, and may be higher or lower than that
without. feedback. .
5-9. Feedback and Stability. A great deal of information about
the stability of an amplifier can be obtained from an analysis of the
factor 1 - K(3 that appears in the general gain expression [Eq. (5-33)].
This is best analyzed through the use of the polar plot of the expres-
sion Kf3. Since Kj3 is a function of the frequency, points in the
complex plane are obtained for the values of Kj3 corresponding to all
values of f from 0 to 00. The locus of all of these points forms a closed
. curve.
As a particular example, suppose that the locus of Kf3 in the complex
90 ELECTRON-TUBE CIRCUITS [CHAP. 5

plane is drawn for the amplifier illustrated in Fig. 5-19. To do this, the
complete expression for the nominal gain, including the effect of the feed-
back circuit, must be written, rather than the simple form given in Eq.
(5-46). Also, the value of /3 must include the effects of the blocking

Mt&ld/e
ranqeoT ~1~------------~j----=~~~---4--
frequencies I

"f.z
FIG. 5-23. The locus in the complex plane of K{3 for the circuit of Fig. 5-19.

capacitor C. Certain of the features of the response are known. At the

mid-frequencies, the gain is substantially constant and has a phase of
180 deg. For the low and high frequencies, the gain falls to zero, and
the phase approaches ± 90 deg. At the I-f and h-f
cutoff values, the phase is ± 135 deg. It may be
shown that the general locus of K/3 for all frequencies
is a circle. The result is shown in Fig. 5-23. A vector
drawn from the polar locus to the point (+ I,jO) is
1 - K/3. For this particular circuit its magnitude is
FIG. 5-24. A plot greater than unity for all frequencies, and it has its
of 11 - KJ31 = 1 in
the complex plane. maximum magnitude at the middle range of frequen-
cies. Moreover, since the resultant gain varies
inversely with 1 - K/3, then the effect of the feedback is to cause a
general flattening of the frequency-response characteristic.
The criterion for positive and negative feedback is evident on the com-
plex plane. First note that the quantity 11 - K/3 I = 1 represents a
circle of unit radius with its center at the point (I,jO), as illustrated.
Clearly, if for a given amplifier 11 - K(3 I > 1, then the feedback is
negative, with- an over-all reduction of gain. Likewise; if] T - K/3 1<:: 1,
there is an over-all increase in gain, and the feedback is positive. These
considerations show that, if K(3 extends outside of the unit circle for any
frequency, then the feedback is positive at that frequency. If K(3 lies
within the unit circle, then the feedback is negative. If K(3 passes
through the point (I,jO) then 1 - K/3 = 0, and, as will later be shown, the
amplifier becomes an oscillator. A more general analysis by Nyquist 4 • 2
has shown that the amplifier will oscillate if the curve of K(3 encloses the
point (I,jO) and is stable if the curve does not enclose this point.
SEC. 5-9] UNTUNED VOLT AGE AMPLIFIERS 91

Clearly, therefore, if the magnitude of K(3 is less than unity when its
phase angle is zero, no oscillations are possible.
As a specific example for discussion, suppose that the plot of a given
amplifier is that illustrated in Fig.
5-25. The feedback is negative for
this amplifier in the frequency range
from 0 to h. Positive feedback ex-
ists in the frequency range from h to
00 • Note, however, that, since the
locus of K(3 does not enclose the
point (l,jO), then, according to the
Nyquist criterion, oscillations will F!G. 5-25. The pol arI
p 0t 0 f an ampl·fi
1 er.
not occur.
REFERENCES
1. Terman, F. E., Electronics, 10, 34 (June, 1937).
2. Black, R. S., Elec. Eng., 53, 114 (1934).
Peterson, E., J. G. Kreer, L. A. Ware, Bell System Tech. J., 13,680 (1934).
3. Fe1dkeller, R., Teleg. fernsp. Tech., 25, 217 (1936).
4. Nyquist, R., Bell System Tech. J., 11, 126 (1932).
PROBLEMS
5-1. The important constants of one of a chain of RC-coupled-amplifier stages
employing pentodes (see Fig. 5-3) are
rp = 10 6 gm = 1,600 pmhos C = 0.01 Jlf
Cpk = 8 JlJlf Ro = 500k
a. Calculate the mid-frequency gain and the upper and lower cutoff frequencies.
b. Between what frequencies is the amplifier-stage phase 180 ± 15 deg?
5-2. The frequency response of a three-stage cascaded RC amplifier employing
pentodes is to be constant within 0.5 db up to 18 kc. Calculate the h-f cutoff of
each stage. ,
5-3. The LC coupled amplifier of Fig. 5-9 uses a triode. The important factors
are
rp = 10k gm = 2,000 Jlmhos Ro = 10 6 C = 0.01 Jlf L = 40 k
Distributed capacitance = 200 JlJlf
Determine the upper and lower cutoff frequencies and the maximum gain.
5-4. A transformer-coupled amplifier is to be constant within 3 db over the
frequency range from 100 to 8,400 cps.
a. Specify the required values 6f primary inductance, leakage inductance
(reduced to unity-turns ratio), and frequency of secondary resonance. The
tube is a 6J5 with rp = 7,700 ohms. Neglect the winding resistance in the
calculations.
b. If the turns,ratio is 3 and the total input -and wiring capacitance of the next
92 ELECTRON -TUBE CIRCUITS [CHAP. 5

tube is 25 p.p.f, what is the permissible equivalent capacitance across the

secondary of the transformer?

5-5. An amplifier has a gain 3,000/0. When negative feedback is applied, the
gain is reduced to 2,00019. Determine the feed-back network.
5-6. An amplifier without feedback gives an output of 46 volts with 8 per cent
second-harmonic distortion when the input is 0.16 volt.

a. If 1 per cent of the output is fed back into the input in a degenerative circuit,
what is the output voltage?
b. If an output of 46 volts with 1 per cent second-harmonic distortion is per-
missible, what is the input voltage?

5-7. Given the amplifier stage with cathode degeneration shown in the accom-
panying diagram,

Ebb = 250 volts Rz = lOOk

g.. = 1,200 Itmhos It = 70
Ru = 1M

a. What should be the value of Rk to give an over-all gain of 8?

b. What is the value of Eee, and the largest value of eg to yield an output without
distortion?
5-8. Plot the gain as a function of frequency of the simple amplifier shown in
the accompanying figure. Also plot on the same sheet the gain of the stage when
fixed bias is used.

Ym=2000,umhos
{ rp=IOk

fO,uf

5-9. \The first stage of the circuit of Fig. 5-21 uses a 6SJ7 pentode with
Eb = 250 volts E,,2 = 100
The second stage is a 6C5, with Eb = 250, E,,! = -8, Ib = 8 rna.
The other factors are
 UNTUNED VOLT AGE AMPLIFIERS 93
Rll = lOOk Rg = 250k RZ2 = 25k C = 0.04 p.f Ck1 = 10 p.f
Ck2 = 2.5 p.f Cd = 0.1 p.f Rl = 200L! R2 = 150 k
a. Specify the values of R k1 , R k2 , Ebb.
b. Draw the complete mid-frequency equivalent circuit.
c. The total shunting capacitance across R g2 is 80 p.p.f. Calculate and plot a
gain-frequency-response curve over the range from 20 to 50,000 cps.
d. Repeat (c) if Rl = 0, R2 = 150k •
5-10. Given the two-stage circuit which is provided with negative voltage
feedback. The tubes have rp = 10 6 ohms, gm = 1,200 JLmhos.

a. Calculate the output impedance.

b. Calculate the impedance between points AA.

5-11. Given a simple pentode amplifier stage as illustrated, the screen by-paslil
being omitted. Derive an expression for the gain of the amplifier stage. Assume

that Ib is independent of E b , and that JL,g of the screen grid relative to control
grid is the same relative to plate and to screen currents.
5-12. Calculate the gain of the inverse feed-back pair. * Assume that the tubes
are identical and that R g » R z•

* Mezger, G. R., Electronics, 17, 126 (April, 1944).

94 ELECTR ON - TUBE CIRCUITS [CHAP. 5

5-13. Given a three-stage RC coupled amplifier, each stage of whicn has an 1-f
cutoff of 20 cps, an h-f cutoff of 84 kc, and a gain of 220. Plot the locus of the
complex voltage amplification.
5-14. The locus of the complex voltage amplification of a certain amplifier is
illustrated. If 1 per cent negative feedback is applied, determine the value of the
K=I"SOO~o
f=S700 ---~

gain K at the frequencies 100, 10 kc, 40 kc. Assume that B is independent of fre-
quency and that the voltage fed back is in phase with the output voltage.
5-15. Calculate the effective input impedance at 10 kc of the circuit in the
accompanying figure. Each tube has the value gm = 2,000 JLmhos, fp = 10k •
Neglect all tube and wiring capacitances.

5-16. Calculate the effective input impedance of the circuit in the accompany-
ing diagram.

gm=ISOOpmhos
zp=66k

5-17. The general current feedback circuit, as viewed from the output ter-
minals for output impedance measurements, is illustrated. Analyze this circuit,
and verify Eq. (5-57).

5-18. Apply Eq. (5-57) to Fig. 5-17. Show that the results so obtained yield
the expected results.
 CHAPTER 6

UNTUNED VOLTAGE AMPLIFIERS-Continued

T HE untuned voltage amplifiers that are discussed in the foregoing

chapter possess flat frequency-response characteristics over a range of
frequencies. Frequently, however, the region of uniform amplification
must be wider than is possible with the simple circuits. For example,
radar receivers may require a uniform response from 2 to 8 megacycles,

FIG. 6-1. A shunt-peak video amplifier stage.

depending upon the service, although the I-f response in these is not too
critical. Television receivers require a sensibly uniform amplification
"over the range from about 30 cps to 4.5 megacycles. These broad-band
amplifiers may be accomplished by compensating the simple amplifier
at both the l-f and the h-f ends of the frequency scale; by the use of tubes
as coupling devices, these being connected ordinarily as cathode followers;
or by the use of circuits operating with certain of the elements which
normally cause the droop in the frequency-response characteristic elim-
inated from the circuit.
6-1. Compensated Broad-band Amplifiers. It is possible to compen-
sate for the drooping of the frequency-response characteristic of a resist-
ance-capacitance coupled amplifier at both the h-f and the l-f ends of the
curve. Several methods exist for accomplishing this result, and these
will be considered below in some detail. However, it is advisable to
examine roughly what occurs in these several methods of compensation
before undertaking a complete analysis.
_In. the shunt-12eaked m(.J~4Qcl~Lb_·:f cpmpensation, an inductallce is
i!!:~erted in series with the plate resistance. The circuit has the form
illustrated in Fig. 6-1. The inductance Lc is chosen of such a value that 1
it· resonates with the effective input capacitance of the following tube in~
the neighborhood of the frequency at which the response would otherwise
95
96 ELECTRON - TUBE CIRCUITS [CHAP. 6

begin to fall appreciably. In this way, the h-f end of the response curve
can be appreciably extended. The choice of the value of Le is critical;
otherwise a peak in the response curve may occur. Such overcompensa-
tion must be avoided in most applications.
The_l:Lenslof the respol!s~9tlr'::"~l!l~Y be illlpr~wed by the use of a
cauaQit.Qr~cr.9§s _a_ p()I."tjQ!l: 5>f tlJ.e load resistor, as illustrated in Fig.- 6-2.

c
Rl

FIG. 6-2. L-F compensation by use of a capacitor across a portion of the load.

In this method, the load resistance is effectively Rl at the high frequencies,

owing to the shunting action of the capacitance Cc across Re. At the low
frequencies, the shunting effect of Ce is negligible, and the load resistance
is effectively Rl + Re. The increased gain of the stage resulting from
the increased effective load impedance thus compensates for the loss of
gain resulting from the voltage-di-
vider action of the coupling ca-
pacitor and the grid resistor. In
this case, as for the h-f compensa:
Rg tion, care must be exercised in the
choice of circuit constants.
6-2. H-FCompensation. To
FIG. 6-3. The equivalent circuit of an study the gain characteristics of
RC amplifier with an inductance in series an amplifier that is provided with
with the plate resistor for h-f compensa- a shunt compensating circuit, the
tion. equivalent circuit of Fig. 6-1 is
drawn.
As the series inductance Le is small (20 to 50 Mh), its presence does not
in any way affect the I-f or mid-frequency gains of the amplifier. Con-
sequently for this amplifier, Eq. (5-6) for the mid-frequency gain and
Eqs. (5-11) and (5-12) for the I-f gain are still valid. These expressions
are rewritten here for convenience.

K - -MY p \
o- Yp +Y +Yl Rg (
(6-1)
Kl 1 It -1 f If (
Ko = VI + (ftifF an 1 )
SEC. 6-2] UNTUNED VOLT AGE AMPLIFIERS 97

Broad-band amplifiers usually employ pent odes with relatively small

pl~=toad resistances. Because of this, The-dlscussronhere wIn be con-
fined to amplifiers of this type. For
pentodes
Tp» Rz } (6-2)
R g2 » Rz
and the equivalent circuit of Fig. 6-3
reduces to the form of Fig. 6-4. It
follows directly from this that the FIG_ 6-4_ The approximate equivalent
mid-frequency gain is circuit of Fig. 6-3 at the mid- and high-
Ko = -gmRI (6-3) frequencies.
and the h-f gain becomes
K - -MY p (6-4)
2 - Yp + Y z + Y Rg + Y c" -

The h-f- to mid-frequency-gain ratio is

K2_ 1
(6-5)
Ko - RI(YZ + Yeo)
But the half-power frequency without compensation is, from Eq. (5-16),

fz = Yp + Y z + YR. == _1_ (6-6)

'hC g 'hRzC g
Also, it is convenient to define the quantity Q2 as
Q2 = ~ = w2Lc (6-7)
RiCg Rl
Equation (6-5) may then be written in the form

(6-8)

This expression is expanded, thus,

K2 1 +
j(W/W2)Q2
Ko = [1 - (w/w2)2Q21 + j(W/W2)
which may be written as

The significance of this equation is best understood by examining curves

98 ELECTRON - TUBE CIRCUITS [CHAP. 6

of gain and phase shift for various values of Q2. A set of gain curves is
given in Fig. 6-5.
It is ordinarily desired that the voltage gain be practically constant up
to a certain designated high frequency. An inspection of Fig. 6-5 shows
that this is best achieved by choosing Q2 to have a value approximately

1.4
V
~\Y r\
1.2
0.'1./ 15 . . . \
../
~J 1\\
~I~
/.0
I--::::
- ~

r--.
~
oW- ~~
\\
.~
<S
cr>
...
0.8
"'' ' '\\
~
}. 0.6
.!:!
OJ
a:
0.4
"" ~ '-.
~" ~
0.2 "" "" ~ -..;;::
~i::::
o
0.1 O. '1. 0.3 0.4 0.5 0.6 08 1.0 '1.:> 4 5 6 8 10
W/UJ2
FIG. 6-5. Dimensionless relative gain curves for the shunt-compensated amplifier.
(Adapted from Bedford and Fredendall.1)

0.45. Actually the optimum value for Q2 is 0.414, under which conditions
dK/df = d 2K/df2 = Oat 12, which yields the maximum flatness. -From
Eqs. (6-6) and (6-7), this requires apprQximately that

(6-10)

That is, approximately constant gain is achieved when the load resistance
is approximately equal to the reactance of the effective shunt capacitance
and when the inductive reactance is equal approximately to one-half the
load resistance at that frequency.
In order to preserve the wave form of the signal in an amplifier, not
only must the relative amplitudes of the various frequency components
be maintained, but also their phase relations must be held constant.
If this is not so, then phase distortion results. The phase shift through
the amplifier is contained in Eq. (6-9), which is
SEc.6-2}

or
ip

ip = 180 -
UNTUNED VOLT AGE AMPLIFIERS

= 18O + [ tan-1 W2
W Q2

tan-1:2 [1 -
-

Q2
tan -1 1 _ W/W2
(W/W2)2Q2 ]

+ (W~2y] = 180 - ()
l (6-11)
99

Evidently, for the wave form to be preserved, either the phase shift of
the various components through the amplifier must be zero, or else the
phase of all the frequency components must be changed by the same
amount in time and the relative phase relations among the harmonics in

0.16

0.14
-- t--.
~
1

r-- r-.C?) i' i,\

l'- J \
0.12
dJ) j. ~
0.10
'" \
(J Wz
2ffw
0.08

0.06
- 10.S ,.Ar
_r""'"

/
/
~ l\.\
~\
'\.."'

0.04
\'VI
J

----
0.02 /
.... v
o
0.1 0.2 0.3 0.4 0.50.6 0.8 1.0 2 3 4 5 6 8 10
w/wz
FIG. 6-6. Dimensionless curves of Ojw as a function of W/W2 for shunt-compensated
amplifiers.

the wave form must be preserved. Clearly, a criterion for zero phase
distortion is that O/w be a constant. A plot of ()wz/27rw vs. W/W2 is con-
tained in Fig. 6-6.
Figure 6-6 shows a family of dimensionless curves which apply to any
amplifier. With W/W2 and Q2 given, the phase shift at any frequency is
readily determined. It will be observed that the curve for Q2 = 0.34
shows the least variation of 0/ wand so introduces the least phase distor-
tion or time delay. On the other hand, the curve for Q2 = 0.414 shows
the least variation in gain. In general, therefore, it would appear that
the value of Q2 should lie in the range from 0.34 to 0.41. Frequently a
value of Q2 of 0.5 is used, for the following reasons.
The curve for Qz = 0.5 yields a relatively uniform amplification over
 100 ELECTRON- TUBE CIRCUITS [CHAP. 6

as wide a frequenoy range as for any value of Q2. For this case, the
relative gain increases by about 3 per cent (=0.3 db) at W/W2 = 0.70 .
. This is frequently a tolerable increase if only a few stages are used.
\. Moreover, the time-delay errors of the individual stages are additive and
would be tolerable in many applications. If many stages are to be used,
then the situation changes. For example, a lO-stage amplifier designed
with Q2 = 0.50 would have a 3-db (41 per cent) bump in the gain curve,

(a)-Shunt (d)-Series m·cJerived '

(b)- Series (e)-Shunt m-derived

(c)-Series-shunt (f)·2sectiOn const~nt k

FIG. 6-7. Various h-f-compensating coupling circuits. (D. L. Jaffee, Electronics,
April, 1942.)

with a corresponding serious time-delay error. Such characteristics

might not be tolerable, and a value of Q2 of 0.41 would then be used.
Clearly, therefore, the acceptable value of Q2 will be determined by
the maximum allowable variation in gain and the tolerable time-delay
error. If one can tolerate the resulting variation in gain, then a given
number of stages with Q2' = 0.50 will provide a given gain with a broader
band width than is possible with Q2 = 0.41. This means that the require-
ment for maximum flatness sacrifices band width for a more uniform
amplification curve and less time-delay or phase distortion.
It should perhaps be emphasized that the inclusion of an inductance
as a compensating network has resulted in a rather remarkable improve-
ment in response. That such a simple correcting network does not
provide both optimum gain and constant-phase characteristics for the
SEC. 6-3] UNTUNED VOLT AGE AMPLIFIERS 101
same conditions should not be surprising or unexpected. More elaborate
h-f compensating networks that provide better gain characteristics have
been devised. A number of these are illustrated in Fig. 6-7.
The transient response of these more elaborate networks is generally
inferior to that of the simple shunt-peaking compensation. 1 This
results from the fact that the amplification and phase-shift characteristics
do not deteriorate so rapidly in the shunt-peaked circuit as in other h-f
compensating systems having the C G2
same h-f limit. Consequently the
apparent advantage of these cir-
cuits is largely eliminated.
6-3. L-F Compensation. A
circuit that compensates for the
droop in the low-frequency re-
sponse characteristics of an RC FIG. 6-8. The equivalent circuit of an I-f-
amplifier is given in Fig. 6-2. A compensated RC amplifier.
physical explanation of the operation of the l-f compensating network is
given on page 96. An analytical study is to be made of the performance
of this circuit, in order to obtain a more complete description of its
behavior.
The equivalent circuit, given in Fig. 6-8, is to be analyzed. Use is
again made of the fact that pent odes are generally used in video amplifiers,
whence rp is greater than the combined output load impedance Z. Also
in general, the grid resistor of the following stage Rg is large compared
with the load resistor R z•
The mid-frequency gain assumes that the reactances of Cc and Care
small compared with Rz and R g , respectively. The gain expression is
then simply
(6-12)
The low-frequency gain is readily obtained from an examination of
Fig. 6-8. The expression is
Rg
K 1 = -gm Z 1 (6-13)
Rg+~C
JW
where Z is the total effective impedance of the output network at the
low frequencies. By inserting the known value of Z in Eq. (6-13), and
combining with Eq. (6-12), there results

(6-14)
+
Rg
1)
+ JW
1
·;--C
Rg 1
+ jwC
102 ELECTRON-TUBE CIRCUITS [CHAP. 6

or
Kl 1 1
Ko= 1 + jwCR , 1 + _1_
1 + Re 1 + jwCRg jwCRg
Rz + jwCcRcRz
or
Kl 1 1
Ko= Rz + jW~eRcRz + jw?R , 1 + jwCRg
Rl + Re + JwCcRcRl 1 + JwCR g jwCRg
which is
1
Kl
Ko = HI
Rg
+( + jwCeRcRl )
Rl
Rz + Re + jwCeReRI
C+ jwCR g)
jwCRg
(6-15)

But ordinarily the ratio R,/Rg « 1, and Eq. (6-15) becomes

Kl 1
(6-16)
Ko = Rl + jwCeRe 1 1 + jwCRg
Rl + Re 1 + . C ReRI jwCRg
JW eRe + R,
Three time constants appear in this expression, viz.,
RC C ReRI RC
e c e R e + Rl 9

By choosing the parameters such that

"x CRg = Ce R~+Rkl (6-17)
then Eq. (6-16) becomes
Kl 1
Ko = Rl 1 + jwCeRe 1 + jwCR g

Rz + Re 1 + jwCRg jwCRg
which is
Kl 1 1
(6-18)
Ko = (1 + jWCcRll) JW. C~ R
eel
C
. 1
1 -J--
wCeRe
This may be written in the form
Kl 1 /tan-1 X /R (6-19)
Ko = V1 + (Xc./Re)2 c, e

It is noted that this choice of parameters yields exactly the same form
for the l-f response for the compensated case as that of the uncompensated
amplifier, except that the l-f response is now controlled by the time con-
stant ReCe rather than by the output time constant RgC.
6-4. The Cathode Follower.2 The foregoing sections show that, to
achieve a uniform frequency response of an RC amplifier over a wide
SEC. 6-4] UNTUNED VOLT AGE AMPLIFIERS 103
frequency band, it is necessary that the effective output impedance of
each stage be low and the effective input capacitance be small. The small
effective input capacitance can be attained by the use of pentodes.
However, the output impedance of a pentode is high unless the load
resistance is small, in which case distortion may result. Considerable

la)
FIG. 6-9. The basic circuit (a) and the equivalent plate circuit (b) of the cathode
follower.

improvement results from the use of two or more high-gain stages, prefer-
ably using pent odes, which are coupled by~jr)2.Q~_\¥Eich is connected to
p!:gvide a low input capacitance and a low output impedance. Although
such a coupling tube may contribute nothing to the over-all gain of the
system, it does make possible a uniform amplification over a wider fre-
quency range than with the conventional coupled amplifier. The use
of a cathode follower as the coupling stage is particularly suitable because
it possesses a low effective input capacitance, high input impedance, low
output impedance, and low nonlinear distortion. Such a stage does
possess a gain that is less than unity, although this is not a serious
limitation.
The basic circuit and the corresponding equivalent circuit of the
cathode follower are given in Figs. 6-9. To find an expression for the
gain of the amplifier, it is noted that

But it follows that

(6-21)

Solving for the gain K, which is given by

K !.E E kp (6-22)
E~'P
104 ELECTRON - TUBE CIRCUITS [CHAP. 6

there results
K = (jwrpC yk +
P.)Zk (6-23)
jwrpZk(Cgk + Cpk +Clk) + rp + (p. + l)Zk
For those values of Zk which are normally used, the effect of the inter-
electrode and wiring capacitances on the voltage amplification is negligible
for frequencies below about 1 megacycle. Equation (6-23) reduces, for
these lower frequencies, to
K ==
rp + (p.P.Zk+ l)Zk (6 24)
-
Examination of this expression shows that the gain K approaches the
limiting value p./(p. + 1) as the ratio of Zk/rpapproachesinfinity. For the
values of Zk and rp found in normal cases, K is of the order of 0.9.
The limits of linear operation of the cathode follower are determined by
the grid current and cutoff characteristics of the tube. These limits can
be obtained readily. It is assumed -that the voltage /Egk = Egp - EkI1
has the limits zero (for zero grid current) and Eo (at cutoff). If the
region of a-c operation is so chosen that the operating point is midway
between these limits and also that the grid-cathode signal reaches these
limits at the peaks, then the maximum output is given by
Ekp,max _ K
Egk,max - 1 - K
But
Eo - 0
Egk,max = 2 V2
and

then
Eop.
(6-25)

The value of Eb to be used in this expression will be dependent on the

character of the load. For example, if
the load is inductive, then Eb will be
a value less than E/ib, owing to the
elliptical load curve.
6-5. The Cathode-follower Ampli-
fier. When used as a coupling ampli-
Cg
fier between stages, the cathode-fol-
FIG. 6-10. An RC coupled cathode- lower circuit becomes essentially that
follower stage, illustrated in Fig. 6-10. In this cir-
SEC. 6-5] UNTUNED VOLTAGE AMPLIFIERS 105
cuit, Co is the sum of the effective output capacitance of the cathode fol-
lower and the effective input capacitance of the next stage.·
The gain of this amplifier will be examined for the various frequency
ranges. These follow directly from
Eq. (6-24) with the proper interpreta-
tion of Zk.
Mid-frequency Gain. The in t e r-
electrode capacitances are negligible
over the mid-frequency band, whence
Cg may be neglected. Also the cou-
· . C' d' ffi FIG. 6-11. The mid-frequency cath-
p1mg capacItor IS assume su - ode-follower circuit.
ciently large so that its reactance is
negligible. The resulting circuit becomes that of Fig. 6-11. The mid-
frequency gain becomes

(6-26)
(}Jo + 1) + (rp/Z~)
where
1 1 1
(6-27)
Z~= Rk + Ro
H-R Gain. At the h-f end of the response curve, the coupling capacitor
may be omitted, although the effect of Co becomes important. The
equivalent circuit has the form shown
in Fig. 6-12. The gain equation [Eq.
(6-24)] now become!;!
}Jo
K2 = (}Jo + 1) + (rp/Z£') (~-28)
where
FIG. 6-12. The h-f circuit of the 1 1 + 1 . C (6-2)
cathode follower. Z~' = Rk Ro + JW 0 9
The gain ratio KdKo becomes
K2_ 1
Ko - 1 + jwCorp
(}Jo + 1) + (rp/Z~)
which is
Ke 1 1
(6-30)
Ko = 1 + jwr~Co = 1 + j(!/!2)
where
r'p (6-31 )

It should be noted that this expression has substantially the same form
as Eq. (5-15) for the conventional RC circuit. However, since the prod-
106 ELECTRON- TUBE CIRCUITS [CHAP. 6

uct r~Co for the cathode follower is much smaller than TpC a of the RC
amplifier, then the upper frequency limit of uniform response is much
greater for the cathode follower than for the conventional RC stage.
Because of this, it is possible to achieve a high h-f limit even when the
cathode follower is followed by a stage
having a high input capacitance.
This means that a rather wide fre-
quency response is possible under
these conditions even with a following
Rg
triode stage.
L-F Gain. At the low frequencies,
FIG. 6-13. The l-f circuit of the cath- C bid d h ff f
ode-follower amplifier. g may e neg ecte ,an tee ect 0
the coupling capacitor becomes very
important. Equation (6-24) appropriately modified becomes

Kl = (p. + 1) ~ (rp/Z~") Rg + 1;JJwC) (6-32)

This expression may be written in the form

_ __________~p.~__________ jwCRg
Kl = (6-33)
(p. + 1) + rp _ rp 1 + jwCR g
Z~ Rg(1 + jwCRg)
where use has been made of the fact that

(6-34)

The gain ratio becomes

Kl 1 jwCR g [( + 1) +!2]
Ko = (p. + 1) + rp _ rp 1 + jwCR g p. Z£
Z~ Rg(1 + jwCRg)
which reduces to the form
Kl 1 1
Ko = 1 - j (1/wCR 1) 1 - j(/tl/)
where
1 (6-35)
It = 27rCR l

This expression has substantially the same form as Eq. (5-9) for the I-f
gain of the RC amplifier except that it may be shown that under typical
operating conditions the value of II is much lower for the cathode follower.
SEC. 6-6] UNTUNED VOLTAGE AMPLIFIERS 107
X 6-6. Input Capacitance and Output Impedance of the Cathode Fol-
lower. To find an expression for the input capacitance of the cathode
follower, refer to Fig. 6-9b. It is seen that the current flowing through
the source comprises two components. One of these is the current
through the capacitance Cgp and is
II = jwCgpEgp (6-36)

The second is the current through the capacitance Cgk. This is

12 = jwCgkEgk (6-37)

But as Egk = Egp + E pk and K = Ekp/E gp , then

(6-38)
The total current is

and the effective input capacitance is

(6-39)
Since in many circuits K is approximately 0.9, then Ci has the approxi-
mate value
(6-40)
A comparison of this expression with the corresponding form given by
Eq. (4-13) shows a roughly similar dependence on the tube capacitances,
alt~ough the numerical viiLlue for the cathode follower is considerably
smaller than for the conventional
amplifier stage.
The effective output imped-
ance can be determined by find-
ing the current that flows as a +
Fesult of the application of an a-c
potential to the output terminals
of Fig. 6-9. The grid exciting FIG. 6-14. The equivalent output cir-
potential is made zero, and the cuit of the cathode follower.
grid generator is replaced by its
internal impedance-according to the superposition theorem. The
equivalent circuit may than be drawn in the form of Fig. 6-14. The
effective output admittance of the tube alone is
Yo = _1 (EPk + JLE gk + E pk ) (6-41)
Epk rp _1_ + 1
• Y a•• Yg + Y a..
108 ELECTRON -TUBE CIRCUITS [CHAP. 6

But from the diagram it is seen that

Z C" E _____1--=_ _ E (6-42)
~.= 1 ~- Y ~
1 + y +c,.y
-=--.,---,,",y:--
y Za + Cgp + Z Ca'
Zg Cap

Ordinarily Zk is a resistor which is 10k or less, and the reactances of Co.

and Cyp are approximately 20 k at 1 megacycle (for capacitances of 10 /-L/-Lf).
Under most conditions Eq. (6-42) becomes
Eo. == Ep.
and the output admittance reduces to the approximate form

Yo == 1 + /-L (6-43)
rp
This expression shows that the output terminal admittance of the tube
is somewhat greater than the transconductance of the tube. But for the
actual values of transconductance of available tubes, the output terminal

I
t:
I I
ep
l
-1
FIG. 6-15. A general feed-back circuit, and its equivalent circuit.

impedance Zo may be less than 150 ohms and would seldom exceed
1,000 ohms with single tubes, under the assumptions made in the deriva-
tion of Eq. (6-43). In many applications the impedance of the coupled
load is much larger than the output impedance of the tube and so will
have very little effect on the resultant output impedance. A double-
cathode-follower circuit has been devised which has a greatly reduced
output impedance. 3
6-7. The Anode Follower. A circuit which possesses roughly the same
properties as the cathode follower but which also provides for phase
inversion is called an anode follower. This may be a desirable property
in certain applications. The circuit of such an anode follower is given
in Fig. 6-15. Actually, the circuit of Fig. 6-15 is that of a general feed-
back circuit and will be shown in Chap. 8 to have many applications.
It is a simple amplifier which is provided with voltage feedback through
the impedance Zj.
An analysis of the equivalent circuit is readily carried out. An appli-
cation of the Millman network theorem between the points K and G yields
SEC. 6-7] UNTUNED VOLTAGE AMPLIFIERS 109
- EkIY 1 + EkpYJ
E ko - (6-44)
Y 1 + Y J + Yo
When applied between K and P, the network theorem yields
E
kp
= EkoYJ + p,EgkY p = Eko(Y' - gm) (6-45)
YJ + Y p + Yz Y f + Y p + Yz
By combining these expressions, there results
E - Y f - gm EkIY 1 EkpYf + (6-46)
kp - Y J + Yp + Y z Y 1 + Yo + Yj

By solving for the gain, the result is

which may be written in the form

K = Y1(Y f - gm) (647)
r (Y 1 + Yf + Yo)(Yp + Y z) + Yj(Y l + YI/ + gm) -
To obtain an expression for the output impedance of this general feed-
back circuit, the procedure followed
is substantially that used in Sec. 6-6. ~ ,
In the present case, the equivalent ,,,
circuit for this calculation becomes ep
I
as shown in Fig. 6-16. The output I
I

admittance is given by ~--~~~--~----~~

Yo =~ Ip + 10 (6-48) FIG. 6-16. The equivalent circuit for
E pk E pk calculating the output impedance of the
feed-back circuit of Fig. 6-15.
This may be written in the form

Yo = ~
E
[EPk +rp P,Eok + Eok(Yo + Yl)]
pk
which is
(6-49)

But from the diagram

Y
Eok = Y 1 + Yfr + Yo Epk (6-50)

and the expression for the output admittance becomes

Y
o
= Y
p
+ Y r (Yl + Yg + gm)
Y + Y + Yo
(6-51)
1 f
 110 ELECTRON-TUBE CIRCUITS [CHAP, 6

An expression for the input impedance is readily obtained. It follows

directly from the circuit that
E Ig + E gk + Ekl = 0

The input admittance is given by

Yi = ~
Elk
= Y1(1 - Egk)',
Elk
(6-52)

But by Eqs. (6-44) and (6-47)

(6-53)
Then

which may be written in the form

(6-54)

If it is supposed that a pentode is used in the circuit, and with the fol-
'-' lowing choice of circuit elements,
Yp r-.J 10-6
gm 2,000 X 10-6
r-.J

Y 1 = Y f = 2 X 10-6 for R1 = Rf = 0.5 M

Yg = 0.2 X 10- 6 for Rg = 5M
the several important results are as follows:
1. The expression for the gain [Eq. (6-47)]

(6-55)

2. The output impedance, given by Eq. (6-51), reduces to

Zo ==.!
gm
(6-56)

3. The input impedance becomes, approximately,

Yi -
-'- 1 - 2 Kr Y 1..:..- Y 1 (6-57)

A comparison of these results with those for a cathode follower is

interesting. This comparison is contained in Table 6-1. It is observed
that the gains of the two circuits are both approximately unity; the
ca,thode-follower output impedance is approximately one-half that of the .
SEC. 6-8] UNTUNED VOLT AGE AMPLIFIERS III
anode follower; the input impedance of the cathode follower may be
considerably higher than that of the anode follower.
TABLE 6-1
COMPARISON OF THE CATHODE AND ANODE FOLLOWER

Output Input
Gain
impedance impedance

1 2
Anode follower ........................ - - Rl
1 +_2_ gm
gmRl

Cathode follower ....................... +

1
-gm1 R.
1 +_1_
gmRk

6-8. Direct-coupled Amplifier. 4 It is possible to build a type of

cascaded amplifier without reactive elements and, in principle at least,
secure a very broad band amplifier. The voltage gain of such an amplifier
does not depend on the frequency, at least to a first approximation, except
for the effect of tube and wiring capacitances at the higher frequencies.
It might appear that such amplifiers would find very widespread use
because of these desirable charac-
tfieristidcS. However, ~ucdh. admpli-
ers 0 possess certam Isa van-
tages, and their use is limited,
11-: 1= IY Il1]--- - "=.

though they find extensive use in G Po G P

applications as d-c amplifiers and

as amplifiers for very slowly vary-
+~g'.l :~e.lr~ ,Btl ,2 []:p
_+ Rl~
ing inputs. ~
A battery-coupled cascade-am- FIG. 6-17. Basic battery-coupled ampli-
plifier circuit of basic design, to- fier circuit.
gether with the equivalent plate
circuit for small changes in voltage and current, is shown in Fig. 6-17.
The gain of such an amplifier stage is readily found to be •
K = -p.Rz (6-58)
rp + Rz
It will be observed that the circuits are quite like the RC coupled
amplifier except that the coupling (blocking) capacitors are absent.
Because of the fact that the grid of one stage is directly connected to the
plate circuit of the previous stage, it is necessary to include doc sources
in the various critical points in the circuit in order that the quiescent condi-
tions be those of class A operation.
112 ELECTR.ON-TUBE CIRCUITS [CHAP, (,

The battery-coupled amplifier has the outstanding feature that it will

amplify a steady component in the input voltage, but it suffers from three
main disadvantages. The first is the cost of the relatively high voltage
grid-bias batteries. These are required when a common plate and a
common filament supply are used. In an alternative arrangement,
indirectly heated cathodes having different potentials are used, thus
obviating the necessity for large grid-bias voltages. However, separate
plate supplies are required in this case.

FIG. 6-18. A resistance-coupled amplifier,

The second disadvantage of the direct-coupled amplifier is the inherent

instability associated with the direct coupling. The characteristics of
the tubes in the circuit change slightly with time; the battery voltages,
or the a-c line-operated rectified power supplies, likewise change with time.
Since such changes are amplified, the d--c amplifier is not feasible unless
precautions can be taken which tend to overcome this instability. For
p
this reason, balanced circuits and cir-
G
cuits with degenerative feedback are
used, since they tend to minimize this
difficulty.
The third disadvantage arises from
the capacitance between the grid-bias
batteries and the cathodes. This, plus
FlO. equivalent circuit the interelectrode capacitances, stray
of a typical stage of the resistance-
{'oupIed amp l1'fi er 0 f F'Ig. 6- 18. wiring capacitance, and stray induct-
• ance, influences the transient-response
time and materially affects the rapidity with which the amplifier output
responds to rapid changes of input voltage. In consequence, even though
the amplifier is direct-coupled, precautions must be taken to ensure a
broad h-f response in order to provide a short response time.
It is possible to build a direct-coupled amplifier that uses a positive
plate supply, a negative bias supply, and resistance coupling networks.
This overcomes disadvantage 1. The circuit of such an amplifier is
illustrated in Fig. 6-18. The equivalent circuit of a typical stage of this
~mplifier is given in Fig. 6-19. The gain of such an amplifier is re~dily
 SEC. 6-9] UNTUNED VOLTAGE AMPLIFIERS 113
found to be
R g2 Rll(R ei + R g2 )

K= f.I R g2 + ReI Rll + Rcl + R g2

(6-59)
+ Rll (R c1 + R
g2 )

rp Rll + Rcl + R g2

For an appreciable voltage gain, the parallel combination of Rll and

.Rei +
R g2 should be large compared with r p , and R g2 should be large
compared with ReI. This will necessitate the use of a large bias voltage.
Direct-coupled amplifiers are used extensively as the amplifier in a cir-
cuit, the grid exciting source of which has a very high internal resistance
or which is capable of supplying only a very small current. In this case,
the grid current must be very small. In particular, the grid current is
significant when the grid-cathode resistance of the tube, though high,

(bJ-EquivGlient piate circuit

FIG. 6-20. Schematic and equivalent plate circuit of the differenee amplifier.

might not be large in comparison with the resistance of the circuit that
supplies the grid signal voltage. Special electrometer tubes in which the
grid current is of the order of 10-15 amp are available for such applica-
tions. The grid current of the typical negative-grid tube is of the order
of 10-8 amp with normal rated voltages applied to the tube electrodes.
With the electrode voltages at very low values, the grid current may be
reduced as low as 10- 12 amp. More will be said about the applications
of such amplifiers in Chap. 22.
f / 6-9. The Difference Amplifier, A two-tube amplifier which is used
~ extensively as a d-c amplifier is illustrated in Fig. 6-20. This circuit
114 ELECTRON -TUBE CIRCUITS [CHAP. 6

overcomes disadvantage 1 of the previous section and permits the use of

common battery supplies for all stages. Also, the amplifier possesses
some very desirable features and as a result finds applications in various
services. The analysis below will indicate the reason for the descriptive
name of this amplifier circuit.
To analyze the operation of the circuit, Kirchhoff's law is applied
around each circuit. This yields the equations

Il(Rn + Tp• + R k) + 12Rk = P.Eglk } (6-60)

IIRk + 12(R 12 + Tp + R k ) = P.Eg2k
But the potential difference between grid and cathode of each tube is
related'to the circuit parameters by the equations

E g1k = el - (II + 12)Rk } (6-61)

E g2k = e2 - (II + 1 2)R k
By combining these expressions with Eqs. (6-60) there result
I 1(R n + Tp + R k) + 12Rk = p.[e1 - (II + 1 2)R k]

IIRk + 12(R 12 + Tp + R k ) = p.[e2 - (II + 1 2)R k]

which may be written in the form

I 1[Rn + Tp + (p. + I)Rk] + 12(p. + I)Rk = P.el }

(6-62)
Il(p. + l)Rk + 12[R 12 + Tp + (p. + I)R k ] = P.e2
A solution of this set of equations for the currents yields, for II,

(6-63)

and, for the current 12,

Rn + Tp
12 = - p. [ Rkel - ( p. + 1 + Rk)e2] (6-64)
(Rn
+ 1 + Tp) + R
+ p.Tp)(RI2 k
(R
11
+ R 12 + 2Tp )
It follows from these that the output potentials are, respectively,
SEC. 6-9] UNTUNED VOLT AGE AMPLIFIERS 115

and
_ +p.R Z2 [ Rkel - (R~l : ;P+ Rk) e2]
- 12Rz2 - ~-:----;-:~--;-~--'---'------'---'''--- (6-66)
(Rn + p.rp)(R12
+ 1 + rp) + R k(R 11 + R Z2 + 2r )
p

Under ordinary conditions of operation of tubes, the ratio (R z + rp)/

(p. + 1) «R k • Under these conditions Eqs. (6-65) and (6-66) reduce to

(6-67)

(6-68)

It will be observed from these equations that the output voltages are
given in terms of an amplified difference between the two input potentials.
Special cases of these equations are extremely important. Four such
cases will be examined: (a) when the plate resistors are equal; (b) when
one plate resistor is zero; (c) when one grid is grounded and one plate
resistor is zero; (d) when the plate resistors are equal, with one grid
grounded.
Equal Plate Resistors. In this case, it is supposed that the plate resis-
tors are the same, Ru = RZ2 = R z• Equations (6-67) and (6-68) reduce to
-p.Rz
eOl = 2(R
z
+ rp) (el - e2) (6-69)
+p.Rz
e02 = 2(RI + rp) (el - e2) (6-70)

These expressions show that the two output potentials are of equal mag-
nitude but of opposite polarity. Also, appreciable amplification is pro-
vided by the circuit. Note also that the difference between the two
output voltages is
(6-71)

Although this expression was obtained from expressions that are only
approximate, this result ls actually exact. That this is so may be verified
by using Eqs. (6-65) and (6-66) directly, and not the approximate forms,
Eqs. (6-67) and (6-68).
Evidently, the difference amplifier may be used to indicate the exact
point of balance between two d-c potentials. In fact, such a balancing
circuit, with a center-reading voltmeter connected between the plates of
the two tubes, is frequently drawn as a bridge circuit in the manner
illustrated in Fig. 6-21. In this circuit, when there is no input signal and
when the tubes are properly matched, the indicating center-reading volt-
116 ELECTRON-TUBE CIRCUITS [CHAP. 6

meter is at zero. When a signal is applied, an amplified signal results, a

measure of the signal strength being indicated on the voltmeter.

FIG. 6-21. A difference amplifier in conventional form, and also drawn in a manner to
indicate the equivalent bridge character of the circuit.

One Plate Resistor Zero. Suppose that the plate resistor Ru = o. It

follows from Eqs. (6-67) and (6-68) that

eOl = 0 JLRz }
(6-72)
e02 = R, + 2rp (el - e2)

This circuit yields an amplified

output that is the difference be-
tween the signals to the two grids.
If only a single output signal is
required, the present connection
provides slightly higher gain than
the connection of Fig. 6-21, with-
out the output from a single
FIG. 6-22. The difference amplifier with plate.
one plate resistor at zero.
One Grid Grounded, One Plate
Resistor Zero. In this case, it will be supposed that e2 = 0, and Rll = O.
Under these conditions, Eqs. (6-67)
and (6-68) reduce to

(6-73)

With this arrangement, an ampli-

fied output voltage e02 results for an
input signal el. In effect, therefore, FIG. 6-23. The difference amplifier con:
two tubes are used to perform the nected to provide a single output from a
single input.
same function as a single tube in a
conventional type of circuit. As discussed above, however, it is now
SEC. 6-9] UNTUNED VOLTAGE AMPLIFIERS 117

possible to use a common set of voltage supplies. A typical circuit show-

ing a cascade amplifier of difference amplifiers is given in Fig. 6-24.
100 200 tOO 200

-200 -200
FIG. 6-24. A multistage d-c amplifier employing difference amplifiers.

Equal Plate Resistors, One Grid Grounded. In this case, the two plate
resistors are equal, and one grid is grounded. Consequently, this circuit
produces two potentials from a single input voltage, and from Eqs. (6-fii7)
and (6-68) the results are

(6-74)
e02 = 2(RI + rp) el
This circuit is known as a cathode-coupled paraphase amplifier and is used
extensively to provide push-pull signals from a single source of potential.
It is used to provide a virtually bal-
anced set of push-pull potentials for
the deflection plates of a cathode-ray
oscilloscope and is also used as the
driver of a push-pull power amplifier.
An examination of the complete ex-
pressions, Eqs. (6-67) and (6-68), will
show that the output potentials are
exactly balanced only if Rk = 00.
Ordinarily the amount of unbalance FIG. 6-25. A cathode-coupled para-
phase amplifier.
is not serious, but if completely bal-
anced voltages are required, the self-balancing paraphase inverter (see
Sec. 9-11) may be used.
REFERENCES
1. For a detailed discussion of the transient characteristics of video amplifiers, see
Bedford, A. V., and G. L. Fredendall, Proc. IRE, 27, 277 (1939).
Kallman, H. E., R. E. Spencer, and C. P. Singer, Proc. IRE, 33, 169,482 (1945).
118 ELECTRON-TUBE CIRCUITS (CHAP. 6

Arguimbau, L. B., "Vacuum Tube Circuits," Chap. IV, John Wiley & Sons,
Inc., New York, 1948.
2. Reich, H. J., "Theory and Application of Electron Tubes," 2d ed., Sec. 6-11,
McGraw-Hill Book Company, Inc., New York, 1944.
Reich, H. J., Proc. IRE, 36, 573 (1947).
Kraus, H. L., Electronics, 20, 116 (January, 1947).
Schlesinger, K., Electronics, 21, 103 (February, 1948).
3. Hammock, C., M.I.T. Radiation Lab. Rept. 469,1943.
4. Artzt, M., Electronics, 18, 112 (August, 1945).

PROBLEMS
6-1. A 6AC7 tube is used as one of a chain in a video amplifier that is shunt-
compensated at the high frequencies and is also I-f-compensated. The circuit
of this amplifier is shown in the accompanying diagram. Calculate the value of

,,
,,
o'SM eo
I

:,

Lc that will yield the same gain at the upper frequency f2 as at the mid-band.
Assuming the stray wiring capacitances to be 20 J.l.fJ.f, the tube capacitances of the
6AC7 are Cup = 0.015 J.l.fJ.f, Cpk = 5 J.l.fJ.f, Cok = 11 J.l.fJ.f. Calculate also f1 under the
optimum conditions for the I-f compensation.
c~ 6-2. A video amplifier stage is constructed using a 6AC7. An experimental
test shows that the voltage amplification drops to 0.707 of the mid-frequency gain
at f2 = 1 megacycle with R, = 6k •

a. For what value of R, will the h-f cutoff value be 3 megacycles?

b. What value of Lc is required for flattest gain under conditions of (a)?
c. What value of Lc is required for" compromise" gain?
d. What is the mid-frequency gain?
e. What is the phase·angle for flattest gain whenf = 0.8f2?

6-3. A video amplifier using 6AC7 tubes is to provide an over-all gain of

20,000 with a 3-megacycle over-all band width. If the total stage shunt capaci-
tance is 25 J.l.J.I.f and with gm = 9,000 J.l.mhos
a. Calculate the number of stages required, for Q2 = 0.414.
b. If 6SJ7 tubes are used (gm = 1,600.umhos), how many stages are required?
6-4. A five-stage shunt-peaked video amplifier is to be built..
 UNTUNED VOLTAGE AMPLIFIERS 119
a. If the amplification must not rise more than 1 db above the mid-frequency
gain, estimate the value of Q2 to be used. All stages are identical.
b. If the time delay at the h-f cutoff point is to be within 10 per cent of the
midfrequency value, what range of Q2 is permissible?

6-5. Calculate and plot as a function of frequency on semilog paper the gain,
output impedance, and input impedance of the cathode-follower amplifier of
Fig. 6-10 at the following values of w: 250, 2,500, 25,000, 250,000, 2.5 X 10 6,
5 X 10 6 rad/sec. Choose a 6J5 tube for which

Tp = 7,700 J.1 = 20 Cap = 3.4 J.1J.1f

Also choose

Ra = 200k C = 0.01 J.1f

6-6. What must be the value of Rk in Fig. 6-9 if Zo = 300 ohms at 1,000 cps?
A 6J5 tube is used.
6-7. Given the cathode-follower circuit with the grid resistor Ra tied from grid
to cathode, as shown in the accompanying figure. Derive an expression for the

input impedance and the output impedance of this circuit, neglecting the tube
capacitances.
6-8. In the circuit shown in the accompanying diagram, determine

a. Positive signal which will drive ec to zero.

b. Negative signal to drive the tube to cutoff.
c. Mid-frequency gain.
d. Input admittance when Cap = 3.4 J.1J.1f, Cak = 4 J.1J.1f.

~rOOk
6-9. Repeat (c) and (d) of Prob. 6-8 when the tube is changed to a 6AC7,
assuming that the tube operates in its linear region.
 120 ELECTRON -TUBE CIRCUITS [CHAP. 6

/ 6-10. A video amplifier is coupled to a cathode follower, as shown in the figure.

The frequency-response curve of this amplifier is also sketched. Assume that

the transconductance gm = 9,000 JLmhos.

30
r
20 /
/ 1\
c
" 10
<.!) L \
1.0 10
Frequency,cps x 103
100
l\
1000

a. Calculate the approximate value of the coupling capacitor between the two
stages.
b. Calculate the approximate value of the total shunt capacitance.

6-11. The essentials of a double cathode follower 3 are illustrated.

a. Show that the expression for the gain is

K _ JL2 + JLrp/Rz
- (JL" + JL + 1) + (JL + 2)r p/R z
 UNTUNED VOLTAGE AMPLIFIERS 121
b. Show that the output admittance is

1 + J1.(J1. + 1)
Yo = J1. +1 + 1 + rp/R!
Tp + R! rp
= )J.gm

6-12. Repeat Prob. 6-11 for the double cathode follower in the diagram.

6-13. Show that the h-f cutoff of the anode-follower amplifier is approximately
one-half that of the cathode follower. Show also that the I-f cutoff is approxi-
mately the same as that of the cathode-follower amplifier, for Zl = Rg in Circuit
Fig. 6-10.
6-14. Refer to Fig. 6-18 showing a resistance-coupled amplifier. The circuit
constarits are

Rn = 250k R g2 = 500 k ReI = 50().l:

If fbI = 0.5 rna, Ebb = 300 volts, what must be the value of E" if Ee of T2 is to
be -8 volts?
6-15. Calculate the gain of the series balanced d-c amplifier shown in the
diagram.

6-16. Calculate the input admittance and the output impedance of the differ-
ence amplifier when connected as shown in Fig. 6-23. Neglect tube and wiring
capacitances.
 122 ELECTRON-TUBE CIRCUITS [CHAP. 6

6-17. Obtain an expression for the output voltages of the cathode-coupled

two-tube circuit shown in the figure, and compare with Eqs. (6-67) and (6-68).

) 6-18. The circuit of a grounded-grid amplifier is shown. (Batteries have been

omitted for convenience.) Determine the following:
a. The gain.
b. Input impedance.
c. Ratio of output to input power if Zg and ZI are resistors.

,I
~----------~----~--±
6-19. Repeat the calculations of Prob. 6-18 for the inverted amplifier of the
accompanying figure.
 CHAPTER 7

SPECIAL AMPLIFIER CIRCUITS

A WIDE variety of special amplifier circuits have been devised. Among

those to be studied are limiting and clipping amplifiers, peaking circuits,
and clamping circuits.
7-1. Limiting and Clipping Amplifiers. In addition to the various
amplifiers that are intended to reproduce a given wave form with a
minimum of distortion, there are other forms of circuits which are designed
to alter the wave shape of an input wave in some predetermined manner.

\
\ I
,
\ I
\.,/
FIG. 7-1. A series diode used to limit negative signals.

Limiting or clipping circuits are designed to remove by electronic means

one or the other extremity of an input wave.
A limiter is used when it is desired to square off the extremities of an
applied signal. For example, it is used as one stage in a chain to obtain a
substantially rectangular wave form from a sine-wave signal. A limiter
may also be used to eliminate either the positive or the negative portion
of a wave. Such circuits also find application in f-m receivers to limit
to a constant value the amplitude of the signal that is applied to the detec-
tion system. This application is discussed in Sec. 17-14.
Limiting and clipping may be accomplished with the aid of diodes,
triodes, or multielectrode tubes. For triodes and multielectrode tubes,
the limiting may be accomplished either in the grid or in the plate circuits.
The general features of a number of such limiting circuits will be examined
below.
Series-diode Limiting. The circuit of the series-diode limiter is given
in Fig. 7-1. It will be observed that this circuit is precisely that of a
diode as used in a rectifier circuit. Since the tube conducts only when
the plate is at a positive potential with respect to the cathode, then only
the positive portion of the applied wave will pass through the tube, the
negative portion of the wave being eliminated. By neglecting the rela-
tively small drop across the tube during conduction, the output wave
123
124 ELECTRON-TUBE CIRCUITS [CHAP. 7

form is simply the positive portion of the applied wave form. If the
diode connections are reversed, then the positive portion of the wave
will be eliminated and only the negative portion of the wave will pass
through to the output.
r The series-diode circuits possess the feature that they tend to isolate
the driving circuit from the following circuit and thus prevent feedback.
Shunt-diode Limiting. A diode may be connected in shunt across the
load for limiting action. In such cases, the diode may be looked upon
as an infinite impedance for polarities opposite to that necessary for
conduction and as a virtual short circuit for the polarity in the conducting
direction. The ciiode is then acting as a switch which will short-circuit
a given load for a certain polarity and amplitude of th~ ap~lied potential.

FIG. 7-2. A shunt-diode positive limiting circuit.

The connections of a shunt diode for limiting the positive signab at

approximately ground potential are shown in Fig. 7-2: With the cathode
maintained at ground potential, the diode conducts throughout the entire
positive half cycle. During the portion of the cycle when the diode is
conducting, the current passes through the series resistor R. With R
large compared with the.drop across the tube, practically the entire input
voltage is developed across R, and the output voltage is only the small
drop across the diode. On the negative portion of the input voltage, the
diode does not conduct, and the voltage that appears across the output
is then determined by the resistor R and the resistance of the load.
If the connections to the diode are reversed, with the anode held at
ground potential, the tube will conduct only when the input potential is
negative with respect to ground. As a result, the negative potential will
appear across the series resistor, except for the small tube drop, which
may ordinarily be neglected.
An input voltage can be limited to any desired positive or negative
value by maintaining the proper diode electrode at the desired potential.
Two circuits for limiting about a desired potential are shown in Figs. 7-3.
In the circuit of Fig. 7-3a for positive limiting, it is observed that the
cathode is maintained at a fixed potential E above the input. As a
result, the diode does not conduct until the positive potential on the
~node exceeds E, when the action becomes exactly like that of Fig. 7-2.
(The input voltage during the conducting portion of the cycle is lost in
~he series resistor R.
SEC. 7-1] SPECIAL AMPLIFIER CIRCUITS 125
The operation of the circuit of Fig. 7-3b is essentially like that of Fig.
7-3a except that conduction of the diode occurs when the cathode falls
below the value - E. The input portion of the voltage during conduc-
tion appears across the resistor R.

(a)- Positive limitin9

f\~
L.JE
\,-., I

(6)- Negative limiting

'"
FIG. 7-3. Positive and negative limiting above and below ground.

Series-Shunt Limiting. Somewhat improved limiting action is possible

by combining both the series and shunt limiting circuits into a single cir-
cuit. The possible combinations are illustrated in Fig. 7-4. In the
form shown, the circuits limit completely all signals of a given polarity.
Of course, if biasing potentials are applied, as in Fig. 7-3, the limiting
point would be placed at the discrete level determined by the bjas volt-
age E.

FIG. 7-4. Double-diode limiters for removing (a) all positive-going potentials; (b)
all negative-going potentials.

This circuit, as for the simple series-diode limiting circuits, isolates the
driving circuit from the following circuit and thus prevents feedback
through the circuit.
Double-diode Limiting. It is possible to limit both amplitude extremi-
ties of a wave form at any desired levels by placing two diodes in the
circuit, one of which acts to limit the positive peaks, and the other of
which acts to limit the negative peaks. The circuit for such double-
diode operation is given in Fig. 7-5. In this circuit the diode Tl conducts
whenever the input voltage exceeds the positive value E t • The diode T2
conducts when the input voltages falls more negative than the potential
-E 2•
126 ELECTRON -TUBE CIRCUITS [CHAP. 7

Grid-circuit Limitiny. Limiting in the grid circuit of a triode, tetrode,

or pentode is possible in exactly the same way as in the plate-cathode
circuit of the diode circuit of Fig. 7-2. The series resistor R in the grid

-tv- -~kA
:-p
FIG. 7-5.
~
~~ -=ifi'--''-----+-U---,-----,k-
-= -=
L- J
£2
T----'-
Double-diode limiting circuit.
\j
----L 2

circuit is large compared with the grid-cathode resistance when the grid
is drawing current. The circuit is given in Fig. 7-6.

FIG. 7-6. Grid-circuit limiter.

v
If a bias voltage is used, as in Fig. 7-7, limiting occurs about the
bias level, precisely as in the manner of the 'circuit of Fig. 7-3a. It is

-Ecc
FIG. 7-7. Grid-circuit limiting about a fixed potential level.

possible to use self-bias instead of fixed bias for setting the potential level
about which the positive portion of the wave is limited. The basic cir-
cuit of such a limiter is given in Fig. 7-8.
Saturation Limiting. Limiting action
may also be obtained in the plate circuit
of an amplifier by employing a large load
resistor in conjunction with a low value
of plate voltage. Such limiting arises
from the fact that, with a highly positive
FIG. 7-8. A self-biased grid-cir-
signal on the grid, the maximum tube
cuit limiter. current is limited because of the low
plate potential Ebb. The mInImUm
output potential is Ebb - Eb. Since these limits, for small Ebb, are reached
SEC. 7-1] SPECIAL AMPLIFIER CIRCUITS 127
with small signals, clipping occurs for values greater than these limits.
These results are made evident by an examination of the plate charac-
teristics of the tube as illustrated in Fig. 7-9. The circuit and the results
using a pent ode for saturation limiting as employed in a receiver for f-m
signals are given in Sec. 17-14.

FIG. 7-9. Saturation limiting in a triode.

Cutoff Limiting. If a tube is operated in the class B region by setting

the bias so that the current in the tube is nominally near cutoff, then the
application of a sine-wave grid signal will give rise to an output current
that possesses features not unlike the current from a diode rectifier.
This results from the fact that a small negative potential will drive the

FIG. 7-10. Cutoff limiting in a strongly biased amplifier.

tube beyond cutoff and that no current will flow for any potential below
this value. The operation is made evident by an examination of the
dynamic characteristic of the tube as illustrated in Fig. 7-10.
Combination of Limiting Actions. A combination of grid limiting and
cutoff limiting may be employed in an amplifier to produce a substantially
128 ELECTRON-TUBE CIRCUITS [CHAP. 7

square wave from a sine wave or other comparable wave shape. The
action is illustrated in Fig. 7-11. During the positive portion of the
swing, the grid-circuit limiting is effective ; the plate current reaches its

FIG. 7-11. Formation of a square wave by grid-circuit and cutoff limiting.

maximum, and the output potential reaches its minimum. During the
negative portion of the grid swing, the tube is cut off, whence the plate
current falls to zero and the plate
6

k
io potential is at its maximum.
A combination of saturation lim-
ec iting and cutoff limiting in a tube
i~ I----'---'--~;t is called an overdriven amplifier.
The operation is substantially like
that of the circuit with grid-circuit
limiting, except that the grid cir-
FIG. 7-12. Formation of a square wave
by saturation and cutoff limiting. The
cuit limits by virtue of the high
overdriven amplifier. grid potential, which causes a high
grid current and hence a small grid-
cathode resistance. The operation is illustrated in Fig. 7-12.
,7-2. Peaking Circuits. Electronic control systems frequently require
the accurate synchronization of one event or a multiplicity of events rela-
tive to each other. Often this synchronization is accomplished by
generating triggering pulses in the proper time sequence and having each
triggering pulse initiate the operation of some portion of the total circuit
in the proper order. As a rule these triggering pulses should be of short
duration and should have an extremely sharp leading edge. Such pulses
can be generated in specially designed pulse-forming circuits, and a
number of these will be examined in Chap. 19. It is also possible to
produce them in a peaking circuit. Such a circuit is capable of distorting
an input signal wave in such a way as to produce an output wave form
in which the time duration is shortened and in which the leading edge is
made as nearly vertical as possible.
The choice of peaking circuit used will depend primarily upon the input
wave shape. One of the common methods is to use sufficient limiting
 SEC. 7-2] SPECIAL AMPLIFIER. CIRCUITS 129
and amplification so ths.t a substantially rectangular wave is available and
then apply this square wave to an RC differentiating circuit. Although
a number of such differentiating circuits exist, the requirements imposed
by this application are not very stringent in general, and simple forms of
so-called differentiating circuits may be used.

r
FIG. 7-13. Several differentiating and integrating networks.

The forms of circuit in common use for differentiation and integration

are shown in Fig. 7-13. In the RC network, the instantaneous behavior
for an applied potential e is governed by the equations

eR = ~.iR = RC dec

ec = bJ
dt

i dt = zic JeR dt
l (7-1)

The corresponding equations for the LR circuits are

(7-2)

But the sum of the potentials appearing across the capacitance and the
resistance in series must be equal, at every instant, to the applied poten-
tial. If eo is very small compared with er , then er must be approximately
\.equal to the applied potential. It therefore follows that (1) eo is approxi-
\ mately proportional to the time integral of the applied potential when
ec/er is small and (2) er is approximately proportional to the time deriva-
\ _tive of the applied potential when er / eo is small.
If a perfect differentiating and integrating network were used, then
the output potential may be specified directly. Thus it follows that

Input Differentiation Integration

Square .......... Vertical lines above and below base line Triangular
Saw-tooth ....... Horizontal line on one side of base line spike Parabolic
Triangular ...... Square Parabolic half cycles
130 ELECTRON-TUBE CIRCUITS [CHAP. 7

ThB bBhll.vior of thB circuits for an applied d·c potential is governed

respectively by the equations
a. For RC Circuit b. For RL Circuit
t
. Eo -RC . Eo _!i!
t = lIe t = R (1 - e L)
t Rt
(7-3a) (7-3b)
eR = Eoe -RC eL = Eoe -7:
t Rt
ec = Eo(1 - e -RC) eR = Eo(1 - e -7:)
e = eR + ec (7-4a) e = eL + eR (7-4b)
A plot of the general characteristics of these equations is given in Fig. 7-14.
An examination of these equations reveals the following very important
facts: In the RC circuit, the current through the capacitor may change
instantaneously, although the voltage across it cannot change suddenly.

--- -
1.0 1-
0.8 I"'" ~ ~
r-
0.6 ......... r-
0.4 ....-
k
-r---r--
0.2

00
V
L
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
-1.8
x
FIG. 7-14. The exponential characteristics of the transient solutions of Eqs. (7-3a)
and (7-3b).

That is, a capacitor acts like a short circuit for sudden changes
in potential. In the RL circuit, the current through the inductance
~annot change suddenly, although the voltage across it may change
instantaneously.
If, instead of a steady d-c potential, a recurring square wave is applied
to the circuit, the potential across each element will begin to rise according
to the time constant of the circuit and will begin to decay when the applied
potential reverses. Curves showing the character of the increase, with
the ratio 1/10 as a parameter, where 1 is the frequency of the applied
square wave and 10 = 1/ (2'Tr-RC) is the quantity defined by the time con-
stant of the circuit, ~re given in Fig, 7-15. These curves are essentially
a plot of the function
t
Cc = Eo(1 - c- RC )

or
Cc
-=
1 -e -2.. (~).!..
! T (7-5)
Eo
SEC. 7-2] SPECIAL AMPLIFIER CIRCUITS 131
For example, if the quantity fo of the circuit is 3,000 cps and a 1,000-cps
square wave is applied, fifo is Ys and the capacitor voltage will rise in
accordance with the curve marked Ys and will reach 90 per cent of its
full value in about 0.13 cycle.
1 "
, io~
1.0 .700.·io
~L /'"
/'
/ ' f...-
f...-
~ -
0.8 /1
j/
I I IV -----
1/
/
I
/'
V fifo-paramefer

V
t-- ....- -
/ / ~
/ /
0.'2 / /
1// .,/"
o V~ --:0.1 -
/"'"
/' V

I---
-- -I-"
/0

30
-
IOC
-
o 0.2 0.3 0.4 0.5
tiT
FIG. 7-15. The potential across the capacitor as a function of time, with ///0 as a
parameter.

fi£o=OJ f/£o=o.l

70 fO

FIG. 7-16. The voltage forms ec and er fot an applied square wave on a differentiating
circuit, with/f/o as parameter.

To obtain the complete response for a given applied square wave, a

careful application of the curves of Fig. 7-15 must be made. This gives
the form for ce • The set of curves for Cr is obtained from the fact that
Cr = C - ce • The results have the form illustrated in Fig. 7-16.
132 ELECTRON-TUBE CIRCUITS lCHAP. 7

Theoretically it is of no consequence whether the circuit is composed of

inductance and resistance or resistance and capacitance. Apart from
the higher cost, it is ordinarily inadvisable to use an inductive differentiat-
ing circuit if it can be avoided, because the effective resistance of the coil
is usually large enough to change the output character of the results.
When an RC differentiating circuit is used, the series capacitance should

FIG. 7-17. Differentiation circuits with similar response characteristics.

~
T

FIG. 7-18. Integrating circuits with similar response characteristics.

ep2

,
~-+--~--~~--~----~~~~--~~
FIG. 7-19. A typical application of an RC differentiator as a peaker.

be large compared with any distributed capacitances that may exist in the
circuit. In particular, if the output of such a differentiating circuit is to
be used to feed the grid of a tube, the series capacitance should be large
compared with the input capacitance of the tube in order that the time
constant will not be altered by an indefinite amount.
A number of circuits exist which give transient-response curves that are
similar to those represented by Eqs. (7-3a) and (7-3b). The circuits of
Fig. 7-17 give transient-response curves similar to those represented by
Eqs. (7-3a), and the circuits of Fig. 7-18 give results similar to those of
Eqs. (7-3b). .
The circuit of Fig. 7-19 illustrates a practical application of an RC
SEC. 7-3] SPECIAL AMPLIFIER CIRCUITS 133

differentiator as a peaker circuit. The general character of the wave

shapes at several points in the circuit are illustrated in Fig. 7-20.
Care must be exercised in analyzing the actions in such a circuit, par-
ticularly if the grid of any tube is
driven positive. In particular, the
grid-cathode resistance rc when the
grid is conducting is low and is of the
order of 1,000 ohms. When a tube
is conducting, the average tube
resistance, or beam resistance, Ib
(the ratio eb/ib) is low. When the
tube is cut off, the grid-cathode
resistance may be taken as infinite;
and the beam resistance is also
infinite. Thus, for the circuit of
Fig. 7-19, the equivalent circuit
during the time that T1 is cut off FIG. 7-20. The wave shapes at several
points in the circuit of Fig. 7-19.
has the form shown in Fig. 7-21.
The equivalent circuit during the time that T1 is conducting is as shown
in Fig. 7-22.
2 2
6 C
PoO 6
l

_ = Rl~
R IC-
=-Ebb
rc-
L -_ _ _ -+-_.....i
FIG. 7-21. The equivalent circuit of Fig. 7-19 during the time that Tl is cut off.

7-3. Clamping Circuits. Because no conductive path exists whenever

any asymmetrical wave form is passed through a capacitor or a trans-
former, the average ordinate of the
wave form must assume a zero po-
tential. This is another way of say-
Ring that the d-c component of a wave
form that is passed through a capaci-
tor or a transformer must necessarily
be zero since neither device will pass a
FIG. 7-22. The equivalent circuit of
Fig. 7-19 during the time that T1 is d-c current. This fact is a source of
conducting. difficulty whenever the average ordi-
nate of a wave form must assume a
potential other than zero. Thus the passage of a saw-tooth or other
asymmetrical voltage through a capacitor or the passage of a saw-tooth
current through a transformer will cause the zero of the wave to be at some
level other than its base. The effect of the average ordinate not being at
134 ELECTRON-TUBE CIRCUITS [CF.AP. 7

the base in a cathode-ray oscilloscope is to displace the sweep to one

side. The results are illustrated graphically in Fig. 7-23.
It is possible to bring the sweep back to its proper position by applying
a bias voltage which is equal to the average ordinate displacement from
the base. This is equivalent to restoring to the wave the d-c component

FIG. 7-23. (a) A saw-tooth deflecting voltage. (b) The saw-tooth voltage after
passing through an element without a conductive path.

which was eliminated by the capacitor. This method, referred to as

d-c restoration, is acceptable if an invariable wave form is employed. If,
however, some dimension of the wave is changed, such as its amplitude
or the duration of the voltage rise, the average value will change and
the amount of d-c restoration required will be different.

nuCG D DI
-+-+-H+-+-U-+---'--'--'~I~D""'-----:'LJ
ReFerence
level

No clamp PosdlVe clamp Negativec/amp

FIG: 7-24. A voltage variation with respect to a definite reference potential.

Simple electronic circuits exist which will hold either amplitude extreme
of a wave form to a given level of potential. Such circuits are known as
clamping circuits and are divided roughly into two classes, continuously
acting and synchronized types. The continuously acting diode and grid
clamping circuits clamp either amplitude extreme and allow the wave
form to extend in only one direction from the reference potential. Figure
7-24 illustrates the effect of such clamps. One type of synchronized

Referen~ce~ _ _--\--_ _ _ -+__+ ___

potential _____ J
FIG. 7-25. A synchronized clamp that introduces a d-c reference level.

clamp maintains the output potential at a fixed invariable level until a

synchronizing pulse is applied, when the output potential is allowed to
follow the input. At the end of the synchronizing pulse, the output volt-
age is returned immediately to the reference level. The general action
is illustrated in Fig. 7-25. In a type of synchronized clamp that is
extensively used in television receivers, the output potential is reset
SEC. 7-4] SPECIAL AMPLIFIER CIRCUITS 135
during the synchronizing period to a fixed reference level, and the clamp
is then opened to allow the output to follow the input for a fixed period
of time before the level is again reset.
7-4. Continuously Acting Clamps. Diode Clamp. The simplest type
of clamping circuit utilizes a diode in conjunction with the ordinary RC
circuit, as illustrated in Fig. 7-26. The action of this circuit depends on

JO[
FIG. 7-26. A positive clamping circuit.

the fact that, when the cathode of the diode is made negative with respect
to the anode, current will flow and the circuit acts as if a low resistance
has been connected across the terminals. When the cathode is positive
with respect to the anode, no current will flow and there is, in effect, a
thigh resistance across the terminals. The" closed" impedance is usually
i between 300 and 3,000 ohms for most diode and triode clamps, and the
'" open" impedance is several megohms.
To examine the operation of the positive clamping circuit in greater
detail, suppose that the wave form of Fig. 7-27a is applied to the input of
the circuit of Fig. 7-26. The resulting output wave form is shown in
Fig. 7-27b. During the time in-
terval from A to B, the input volt- O~--~Br-1E~__~Fr-.-____~
age is zero, and the output voltage
is likewise zero. At the point B, (a)
~IOO c G
the input voltage drops suddenly
to -100 volts, to the point C.
Since the capacitor cannot change (6)
A
its charge instantaneously, the
potential across R also drops sud-
denly to -100 volts. But the C
cathode of the diode is now 100 FIG. 7-27. The applied negative-going
volts negative with respect to the voltage applied to a diode clamp, and the
anode, and the tube will conduct corresponding clamped output.
heavily, charging the capacitor very rapidly through the short time-con-
stant RC path (the R being the beam resistance of the diode), until the
capacitor voltage becomes equal to the applied voltage. At this time
the output voltage has returned to zero, and the diode becomes noncon-
ducting. Moreover, during the interval until D is reached, the output
voltage remains at zero potential.
At point D the input voltage changes back to zero, a lOO-volt change
 ELECTRON-TUBE CIRCUITS [CHAP. 7
136
in the positive-going direction. This rise appears across R. Now, how-
ever, the capacitor discharges slowly through R, as the diode is noncon-
ducting (and R is now a high resistance). The voltage across R decays

-H~.Rr_~
D D ~JOL
FIG. 7-28. A negative clamping circuit.

slowly/in accordance with the time constant RC of the circuit, until the
point F is reached, when the input voltage drops to -100 volts. Instan-
taneously the output across R falls to a value that is 100 volts below
its value at the instant that point
c F is reached. The diode con-
ducts, quickly reduces the output
\1) G
to zero, and returns the charge on
A B the capacitor. Note that no por-
tion of the wave form is lost after
F
the first cycle.
If the situation is as illustrated
FIG. 7-29. The output wave form from
in Fig. 7-28 for a negative clamp-
a negative clamping circuit for an ap-
plied positive gate pulse. ing circuit, the corresponding out-
put wave form for an applied
positive gate voltage will be of the form shown in Fig. 7-29.
Grid Clamping. The function of clamping may be performed at the
grid of a triode, since if the grid is made positive with respect to the

..!!:!!..E
I
FIG. 7-30. A grid clamping circuit. FIG. 7-31. Clamping circuit which es-
tablishes the reference voltage E volts.

cathode, an electron current flows and the cathode-grid resistance is very

low, whereas when the grid is made negative with respect to the cathode,
no grid current flows and the circuit is essentially open. Hence the grid
of a triode or a multielectrode tube when connected according to the
circuit of Fig. 7-30 will act as the plate of a diode and produce the same
clamping action. In fact, the open and closed impedances of the grid
clamp are roughly comparable with those of the diode.
 SEC. 7-41 SPECIAL AMPLIFIER CIRCUITS 137
Clamping Above or Below Ground Potential. It is not necessary that a
clamping circuit tie one extremity of the input signal to zero potential.
The reference potential can be made almost any desired value by intro-
ducing the necessary fixed reference potential. In the circuit of Fig. 7-31,
clamping is established with reference to the fixed potential E.
The operation of a simple clamp and also the effects of the clamp
impedance are made more evident by a specific example.
Example 1: The circuit and the applied voltage wave form are illustrated in the
sketch. Assume thatihe effective grid-cathode resistance during grid conduction
\s 1,000 ohms.
+100 allL f A

o D D
-~/Oh----40----~
C ---If---=t=--+

D-----±---'
B
a. Estimate the reading of a d-c voltmeter which draws negligible current
when connected between points C and A in the circuit in the sketch.
Charging b. What would the voltmeter
read when connected between C
___Hat Time const and D?
tOk 11k T = 91 p.sec c. What would the voltmeter
I J read when connected between A
and B?
Discharging Solution: The equivalent cir-
cuits during the charging and
Time const
discharging periods are shown in
T = 1,000 p.sec
the accompanying diagrams.

The potential differences at various points in the circuit are illustrated in the

~oh
o 10
n
50 60
1 ,
ec aOJ9E.' 1
I I
E Q96/E'
I
I
,,
I
eR 7.6 '24.6

7£4
138 ELECTRON -TUBE CIRCUITS [CHAP. 7

figure. These sketches show the potential variations across the capacitor
and across the grid-cathode resistor after steady-state conditions have been
reached.
The potential across the capacitor at the end of the charging cycle (10/-tsec)
is denoted as E. During the discharge portion of the cycle, the capacitor poten-
40
tial falls from the value E to the value E(e -1,000) = 0.961E. During the
charge portion of the cycle, the potential across the capacitor increases from
0.961E to E according to the charging curve (100 - 0.961E)(1 - e-' %'). For
an equilibrium condition to be established, it is required that
(100 - 0.961E)0.104 = 0.039E
or
E = 75.4 volts
(a) ECA = ~ = 0.98 X 75.4 = 73.9 volts
10
+ 10 = 20 volts
(b) ECD = 100 X 40

- e - ~~(27.6 + 24.6) - ~~(75.4 + 72.4)

(c) E AB-R- -54.1 volts
50 =

Owing to the approximations made, ECD = ECA + EAB is not exactly verified.

Example 2: Repeat the foregoing example for the case where the circuit con-
stants are changed to read C = 0.001 /-tf, R = 10 6 ohms.
Solution: The charging and discharging circuits, with the corresponding time
constants, are
Charging: T = 1,000 X 0.001 X 10- 6 = 1 J.lsec
Discharging: T = 10 6 X 10- 9 = 1,000/-tsec
The important wave forms are illustrated in the accompanying figure.

D
I I
E a961B

eRr L.l~9 .~9___. . ;-.....~~ I

4 al r •

A comparison of the above examples shows that, the lower the clamp
impedance relative to the circuit impedance, the "tighter" the clamp
SEC. 7-5] SPECIAL AMPLIFIER CIRCUITS 139
becomes. In particular, if a zero-impedance clamp were possible, then
perfect clamping would result.
7 -6. Switched Clamps. It is sometimes necessary to open or to close
a clamp at regular intervals which mayor may not be directly related to
the signals. If the clamp can conduct in only one direction when closed,
it is called a one-way, or single-sided, clamp. If the clamp can conduct
in both directions when closed, it is a two-way, or double-sided, clamp.
Switched Clamps-Single-sided. The clamping circuits discussed in
Sec. 7-4 are essentially single-sided, and if provision is made to cause thp,
diodes or triodes that cause the
clamping action to turn off during ---ff-~--~---;--­
a certain prescribed period, the
clamp then becomes a switched
single-sided clamp. The modifi-
cation that is necessary in order
-=-E
to convert the biased clamp of T
Fig. 7-31 into a switched clamp is FIG. 7-32. A switched diode clamp that is
illustrated in Fig. 7-32. Here the open for a time T.
diode is switched out of the circuit
by the application of a square pulse for a time T to the grid of the cathode
follower. Clearly, for the switching to operate, the amplitude of the sig-
nal at the cathode of the cathode follower :must exceed the bias potential
E. The other clamping circuits can be modified in generally similar ways.

-1t-.,-....,....----

FIG. 7-33. A one-way clamp using a single triode.

A somewhat comparable circuit that utilizes a triode is given in Fig.

7-33. In this circuit the clamping is done through the beam resistance
of the tube, but since the clamp impedance is rather large, the clamp is
correspondingly quite "loose."
Switched Clamps-Double-sided. Several different types of switched
double-sided clamps exist. One type of synchronized clamp is so
arranged that the output may follow the input during the time that the
synchronizing pulse is applied but is then returned to the reference poten-
tial when the synchronizing pulse is removed. The elements of such a
circuit are illustrated in Fig. 7-34.
140 ELECTRON -TUBE CIRCUITS [CHAP. 7

Suppose that the synchronizing pulse is applied as a negative rectangu-

lar gate which drives Tl and T2 beyond cutoff for the desired length of
time. With tubes .Tl and T2 cut off, the grid of T3 is free to follow any
changes in the input voltage amplitude. Since capacitor C1 has no dis-
charge path, the voltage is transferred to the grid of T3, which then fol-
lows this voltage variation. At the end of the synchronizing period,

S/gna/ /l /l --4~C~1____~~~______~

--4t-C.=..2_-+-+
Synch. 0 0

FIG. 7-34. A switched double-sided clamping circuit.

tubes Tl and T2 again conduct, and the grid potential of T3 is returned

to the reference potential determined by Tl and T2 in series.
If for any reason a signal appears on the grid of T3 without a synchroniz-
ing potential on the clamping triodes, the effect of Tl and T2 is to prevent
the grid potential on T3 to change. This self-compensating effect arises
because a slight increase in the potential at the point A causes T2 to con-
duct more current, which reflects itself as an increase in the bias of Tl.
This tends to reduce the current through Tl, which thus counteracts the
impressed potential. In a similar manner, if the signal tends to decrease
the current through T2, the grid bias of Tl decreases, with a resulting
increase in current. Hence, so long as Tl and T2 conduct, the voltage
at the grid of T3 is held constant and no input signal will reflect itself as a
variation in the output of T3.
This clamping circuit operates quite satisfactorily and is s;:ttisfactory
for many applications. However, owing to the existence of interelectrode
and distributed capacitances to the signal line, a slight dimple might appear
on the signal, somewhat as illus-
11 11
No coupling
11I
11 I
with interOiction
trated in Fig. 7-35, owing to the
interaction of the synchronizing
FIG. 7-35. The character of the effect pulse on the signal line. This
on the output signal by the synchronizing effect is quite tolerable in some ap-
pulse acting through the interelectrode plications but is objectionable in
and distributed capacitances to the signal
line.
television circuits, where such an
extraneous signal might affect the
actual signal pattern. This effect is overcome by using two synchroniz-
ing pulses of opposite polarity in appropriate circuits.
A clamping circuit of importance in television practice,l provides for
SEC. 7-5] SPECIAL AMPLIFIER CIRCUITS 141
establishing an arbitrary reference potential at fixed and regular intervals
on some chosen circuit element in the picture amplifier. The clamping is
applied during the period that the television picture is blanked during the
retrace period. A diagram of such a clamping circuit is given in Fig. 7-36.

~G
~~r0__~~ww__~~

FIG. 7-36. A double-sided switched clamp.

The reference potential is that which exists at the mid-point of R 1•

This may be deduced as follows: During the keying-pulse intervals, both
diodes conduct, and both terminals of Rl are at the same potential.
Because of this conduction, the equal coupling capacitors C2 and C 3
are oppositely charged. During the intervals when the diodes are not
conducting, a current flows in R 1, thus discharging these capacitors.
Since both the circuit and the keying signals are balanced, the diodes
always reach the same potential during the pulse intervals, this potential
being that at the mid-point of Rl during the intervals between pulses.
The time constant C2R 1 = C3 Rl is made large compared with the period
of the pulses, so that the current in Rl is small and the charges on C2
and C 3 remain substantially constant.
If R 1 is connected as shown, the reference potential may be shifted with
respect to ground. Hmvever, if . C.
the circuit is seriously unbalanced, Tl ~f--"-1_ _ _t-:-_ _- - - ,
some difficulty may be experienced
in maintaining the pulse shape. ~B
To minimize this, the resistor R2 U--
may be inserted in the ground
connection. The current through FIG. 7-37. A self-balancing clamp circuit.
this resistor is very small, and
even with large R2 there is no serious disturbance of the reference
potential.
The coupling capacitor C 1 must charge through the clamp circuit, in
the absence of grid current in T2. During the open-circuit intervals in
the clamp circuit, the charge on C 1 cannot change. Consequently the
l-f response of the coupling circuit between Tl and T2 is not seriously
affected.
A clamping circuit that is essentially self-balancing is illustrated in
Fig. 7-37. If it is assumed that the four diodes are identical, then upon
 142 ELECTRON -TUBE CIRCUITS [CHAP. 7

the application of the keying pulses the points A and D are at the same
potential.
Both the foregoing clamping circuits require that the keying pulses be
larger than the amplitude of the picture signal at the point where the
clamp operates. Otherwise conduction from the signal line through the
clamp may occur and thus affect the level. Usually the amplitude of the
keying pulses is made about twice the amplitude of the picture signal.
REFERENCES
1. Roe, J. H., Proc. IRE, 12, 1532 (1947).
2. As a general reference, consult "Principles of Radar," 2d ed., M.I.T. Radar
School Staff, McGraw-Hill Book Company, Inc., N~w York, 1946.
PROBLEMS
General Note: The solution of many of the following problems may require direct
reference to a tube manual for additional tube information.

7-1. The double-diode limiting circuit of Fig. 7-5 is used to clip a sine-wave
input 100 sin 5,000t. In the circuit R = lOOk, and El = E2 = 5 volts. 6H6
tubes are used.
a. Plot the output voltage e2 as a function of time.
b. Suppose that the output e2 is amplified to a 200-volt peak-peak voltage and
is passed through a second identical clipper. Plot the output as a function
of time. Indicate the total rise time on the diagram.

7-2. Consider the circuit in the accompanying figures. Determine the wave

eel f - - - + - - - t - - -

e 6 1 f - - - - t - - + -....

e~~-~--~--
o
shapes at the several points indicated, and sketch these in the manner shown.
Neglect interelectrode and wiring capacitances. Assume Tc = 500 ohms when
grid current is d~awn.
J 7-3. Write an expression for the output voltage as a function of time after
closing the switch (see the figure for this problem).
 SPECIAL AMPLIFIER CIRCUITS 143

7~Dhe interelectrode and wiring capacitances will affect the shape of the
output wave form for a given input wave form. Refer to the circuit in the
accompanying diagram.

eg
o t-----.---r-~,t lube
--(-'.. -- ,,
eg
6JS
cu'rorf=-IS y ,
I

a. Draw the equivalent circuit of this amplifier when the tube is conducting;
when the tube is nonconducting.
b. Determine the output wave shape for the designated input wave.
7-5. Repeat Prob. 7-4 when the circuit is modified as shown in the accompany-
ing figure .

./ 7-6. A saw-tooth voltage is applied to the circuit shown in the diagram for this
problem. Calculate and sketch the output voltage during one complete period.

50~ ,/'

../7-7. The input voltage to the amplifier is as shown (see the figure).
E6b=250v

ISOO

'"
 144 ELECTRON-TUBE CIRCUITS [CHAP. 7

a. Calculate and sketch Ce during one complete period. Neglect tube and
wiring capacitances.
b. Calculate and sketch the output voltage during the period.

7-8. Given the circuit shown in the diagram. The generator has negligible
internal resistance and produces 100-volt rectangular pulses of 2 fJ.sec duration.

S/JO##f'

~
l, OSM

a. Draw a simplified equivalent circuit for the generator and its load during
the 2-fJ.sec pulse interval.
b. Draw the simplified equivalent circuit for the generator and its load during
the interval between the pulses.
c. From the equivalent circuits, determine the principal function of the 500-fJ.fJ.f
capacitor. Is any appreciable signal bias developed'? If so, how much?
d. Sketch the grid voltage ee approximately to scale.

/ 7-9. Refer to the circuit in the figure.

200v

a. What is the maximum value of ebl?

b. What is the approximate value of the voltage across the O.l-fJ.f capacitor.'
c. Sketch the wave form ebl.

7-10. Given the circuit shown in the figure. The input to the circuit is the
rectangular pulses shown, with a pulse-recurrence frequency of 82.6 per seconu.
Choose the following values:

Tc = +
500 ohms when the grid exceeds 10 volts
Tb = 1,000 ohms when the grid exceeds + 10 volts
eel at the beginning of the pulse = -27.5 volts
 SPECIAL AMPLIFIER CIRCUITS 145
+350

a. Draw the equivalent circuit for the charging of CI during each pulse, and
determine the voltage across CI at the end of the pulse.
b. Determine the grid voltage Col at the end of the pulse.
c. Sketch the voltage across C I and eel.
d. Calculate the average d-c value of eel.
e. Calculate and sketch the voltage e2.
f. Calculate and sketch the voltage ea.
 CHAPTER 8

ELECTRONIC COMPUTING. CIRCUITS

CIRCUITS of the type to be considered here are particularly important

in what are known as analogue computer8. In these computers, electrical
quantities, usually voltages, which can be varied and measured conven-
iently are made to obey differential equations which are identical in form
with those of the system under survey. This permits the use of purely
electrical principles and components to yield information concerning the
behavior of a wide variety of physical problems.
Several other basic methods are used in computer design. Some are
entirely mechanical in character and operation, some are electromechani-
cal, and others are essentially electrical. The electrical types are divided
into two general classes, the dig!~~~::L_Ilt:!.~~.~n::Llogu{lty:pes. The circuits
of importance in the analogue type are to be examined.
Among the circuits to be examined are those which perform such basic
mathematical operations as addition, subtraction, differentiation, and
integration. In some cases these circuits depend for their operation on
the'special shapes of the tube characteristics. In other cases, feedback
is applied in special ways to
achieve the desired results. Cer-
tain operations are performed by
combination of circuits.
8-1. D i ff ere nee Amplifiers.
Several circuits exist in which the
output voltage is the difference
FIG,8-1. The common cathode difference between two input signals. One
amplifier, of these was discussed in detail in
"/Sec. 6-9, and the circuit is redrawn
as Fig. 8-1. The results of an analysis of the operation of this circuit are
contained in the equations

eOl = 2(;/'!I~p) (el - e2) ~

+""R 12 ( ) (8-1)
e02 = 2(Rl + rp) el - e2

If a single output is desired, the circuit may be modified to the form shown
146
SEC. 8-1) ELECTRONIC COMPUTING CIRCUITS 147
in Fig. 8-2. The output from this circuit is given by Eq. (6-72) and is
p,R/
e02 = R/ + 2rp (el - e2) (8-2)

It should be noted that no restrictions have been placed on the input

voltages el and e2, and the results
are therefore independent of the
wave shapes and amplitudes, ex- eo2
cept that the amplitudes must not
be so large as to vitiate the linear
equivalent circuit of the tubes.
A second, though generally less
desirable, form of difference am- FIG. 8-2 . A mo d ifi cation of Fig. 8-1 to
plifier is illustrated in Fig. 8-3. yield a single output.
Note that one of the input poten- .
tials is isolated from ground. This may introduce a complication in
many cases.

......L._--4~--+ _ _~ ----r
t
t
e0 1
t
t

FIG. 8-3. A series or "cascode" type of FlG. 8-4. The operating circuit of tube
difference amplifier. Tl.

The operation of the circuit can be analyzed by the usual methods

involving the equivalent plate circuit theorem. For diversity of analysis,
use is made of the superposition theorem.
When the circuit of tube Tl is considered, e2
is set equal to zero. The equivalent circuit
for Tl then becomes that shown. The out-
1 tput voltage from the circuit is readily calcu-
t
: lated to be
~L
t
: (8-3)
FIG. 8-5. The operating cir-• I h
cuit of tube T2. SImi arly, w en the circuit of tube T2 is con-
sidered, the input to Tl is set equal to zero.
The equivalent circuit of T2 then becomes that illustrated in Fig. 8-5.
The output voltage from this circuit is found to be
148 ELECTRON-TUBE CIRCUITS [CHAP. 8

-fl(R k + rp)
(8-4)
e02 = (J.I. + 2)Rk + 2rp
The total output voltage is given by

eo = eOl + e02 = (J.I.

J.I.(R k + rp)
+ 2)Rk + 2rp
(
el - e2
) (8-5)

For the case where the following conditions are satisfied,

• Rk » 2rp
the output voltage becomes
eo == J.I. ~ 2 (el - e2) (8-6)

which shows that the output voltage appears without gain.

Owing to the need for an isolated input potential, a requirement that
can be met if the input is an a-c potential of nominal frequency by the
use of an isolating transformer,
and the fact that the output is
achieved without amplifier gain,
the circuit of Fig. 8-2 is generally
: to be preferred.
I
I
I

t;o, 8-2. Adding Circuits. Suppose

that it is desired to add a num-
I
: ber of voltages of arbitrary phase,
FIG. 8-6. A feed-back summing amplifier. amplitude, and f r e que n c y. A
number of circuits exist for effect-
ing this addition. The addition may be effected in the grid circuit, the
cathode circuit, or the plate circuit of appropriately connected tubes.
A simple feed-back summing amplifier is illustrated in Fig. 8-6. To
show that the output potential is very nearly the sum of the input volt-
ages, one may proceed as follows: A direct application of Kirchhoff's
current law yields the expression

(8-7)

from which

~
Rl
+~
R2
+ ... + !'Rn
.!!:... = eg (~
Rg
+~
R
+~
Rl
~) - ~
+ ... + Rn R
J J

If the resistors are chosen to be

Rl = R2 = ... = Rn = R J = R
and with
SEC. 8-2] ELECTRONIC COMPUTING CIRCUITS 149
then the above equation becomes
1 \' eo n +1 eo
R Lt en = - K Ir - R (8-8)
n

where K is the gain of the tube. Note that a distinction is being made
between the gain of the tube K and the resultant gain of the amplifier K r •
In fact, a comparison of Fig. 8-6 with Fig. 6-15 of the anode follower
would lead one to expect a resultant gain of the circuit of Fig. 8-6 to be

FIG. 8-7. A common cathode FIG. 8-8. The equivalent circuit of

summing chain. Fig. 8-7.

low ar;td of the order of unity. However, the gain of the stage as measured
between the grid and plate terminals will be high, particularly if a pentode
is used. For the case of a pentode, Eq. (8-8) reduces to the approximate
form

which is
(8-9)
A circuit for the addition of voltages in the cathode circuit of a chain
of stages is shown in Fig. 8-7. An analysis of this circuit is readily effected
if it is assumed that identical tubes are used. In this case the equivalent
circuit attains the form shown in Fig. 8-8. An application of the Millman
theorem yields
E + jJ.Eg2k Y p + ... + jJ.EgnkYp
_ jJ.EglkYp
kp - nY + Y p k

E kp = nY:; Y (Eg1lc + ... + E g2k ) (8-10)

lc
But the grid-cathode potentials are

~~" ~ ':' + E;, ~ ':' - E~'I (8-11)

Eonk ~ en +E pk - en - E kP
 150 ELECTRON-TUBE CIRCUITS [CHAP. 8

which leads to

or
JLYp \' (8-12)
E
kp = n(JL + I)Yp + ~k Lt en
which is given, with good approximation, by

E kp == n(JL ~ 1) 2: en (8-13)

This circuit allows for voltage summation with very small interaction
among the sources of voltage, owing to the large input impedance to each
stage. However, it does so at the
expense of one tube for each source
voltage. For this reason, the cir-
cuit of Fig. 8-6 with large series
resistors is to be preferred.
Addition is also possible by con-
necting a chain of tubes through a
common plate resistor, as illus-
FIG. 8-9. A common plate summmg trated in Fig. 8-9. The resistors r
.
chain. in the plate lead of each tllbe are
small suppressor resistors. It may
be shown that the output voltage from such a plate summing chain has
the form

'v (8-14)
n

Here, as in the case of the common cathode summing chain, a tube is

required for each voltage to be included in the adding group, although
some over-all gain is effected by the circuit.
8-3. Squaring Circuits. A squaring circuit is one in which the output
voltage is proportional to the square of the input voltage. Such a circuit
makes use of the curvature of the tube characteristics of two tubes to
accomplish the squaring operation. The schematic diagram of the cir-
cuit is given in Fig. 8-10. It will be observed that the circuit consists of
a single-tube paraphase amplifier (refer to Sec. 9-11 for a detailed discus-
J sion of this amplifier circuit) which provides two signals that differ by
180 deg in phase. l'hese signals are applied to the grids of two amplifier
stages which are so connected that the odd-harmonic terms cancel in
the output. This stage should be compared with a normal push-pull
ampHfier in which the even-harmonic terms are canceled in the output.
SEC. 8-3] ELECTRONIC COMPUTING CIRCUITS 151
As a result of its connection, the even-harmonic terms become the
important ones in the output.

FIG. 8-10. A squaring amplifier.

It is assumed that the plate current in tube T1 may be represented by a

series of the form
(8-15)

Likewise, the plate current in tube T2 will then be given by the series
i p2 = al( -egl) + a2( -egd + aa( -egl)3 +
2
(8-16)
The total plate current

becomes, if it is assumed that the first three terms of the series representa-
tion adequately represent the output,
(8-17)
and the output voltage is then of the form
eo = ipRz = 2a2Rle~1

An explicit expression for the amplitude factor a2 appearing in this

equation is possible in terms of the tube characteristics. If use is made
of the Taylor expansion of the current in terms of the grid potential, viz.,

(8':18)

then clearly the coefficient a2 is related to iJgm/iJe g and the output voltage
is given by the form
(8-19)

But· since the gain of the paraphase amplifier is approximately unity

[see Eq. (9-41)], then finally
(8-20)
152 ELECTRON -TUBE CIRCUITS [CHAP. 8

This expression shows that, for the output to be proportional to the

square of the input voltage, it is essential that the coefficient agm/ae g
remain constant over the range of operation of the tube. This condition
requires that the composite characteristic of the amplifier must be of the
square-law type.
A number of tubes have been examined in order to ascertain which yield
satisfactory characteristics for such squaring operations. It has been
found that the 6B8 and the 6SK7 tubes are satisfactory when grid-driven
and the 6U7G and the 6D6 are quite satisfactory when screen-driven.
A second circuit for squaring is included in the next section as a special
application of the multiplying circuit considered there.
8-4. Multiplying Circuits. A variety of multiplying circuits have been
devised for use in radio-circuit applications and have been extensively
used for modulation and demodulation. Thus it is the function of the
modulator to multiply the carrier and the modulating signals to yield a
modulated carrier wave. Likewise, in the demodulating circuits, the
modulated wave must be combined with the carrier wave to yield the
modulating signal. While such circuits are suitable for these applica-
tions, they are not suitable in general as simple multiplying circuits.
This is so because of the wide variety of other multiplication products
that appear in the output, owing to the nonlinear characteristics, although
these are suppressed in modulation and demodulation applications by the
use of tuned circuits in the output.
A solution is more nearly approached through the use of a tube circuit,
the voltage gain of which may be controlled by means of the potential
on one grid, while the signal output depends directly on the signal applied
to a second grid. That is, if the gain of the amplifier K is proportional
to a voltage ea, then
K = kea
If the output voltage eo is directly proportional to the input signal el,
according to the relation

then the application of the two signals el and e3 simultaneously will yield
an output
(8-21)
A number of tubes possess characteristics which would permit their
use as part of a multiplying circuit. In Fig. 8-11 are illustrated the
mutual-conductance curves of a 6L 7 multielectrode tube as it depends on
the voltage applied to grid 3. It will be noted that gm is a linear function
of the voltage applied to grid 3 over a wide range of potential variation.
Equation (8-21) is satisfied over this linear range. Of course, if grids 1
SEC. 8-5] ELECTRONIC COMPUTING CIRCUITS 153
and 3 are connected together and a single input voltage is applied, the
output voltage will be proportional to the square of the input voltage.

gna
16002

1200.~E
"""
800~
~
400 ~
~

FIG. 8-11. The connections and gm of a 6L7 tube for use in a multiplying circuit.

--,eco.1_ _~ Sum

O/ff'.

FIG. 8-12. The "quarter-square" method of performing the multiplication of two

quantities.

A second method for multiplying two quantities makes use of what

might be called the "quarter-square" method. This depends on the
fact that the product of two quantities may be expressed by an equation
of the form

The block diagram of Fig. 8-12 shows the elements required for yielding
the desired results.

!!!- Sq,!Cfr~nq
ClrcUlf
tXef O/fference
amplifier
ate: SquCfrlnq
circuif ~
~
Oufpul=.k(ef-e;J
FIG. 8-13. Circuit for obtaining the difference of squares.

8-5. Difference of Squares. It is possible to combine several of the

foregoing circuits to yield a number of circuits for performing other
mathematical operations. For example, the combination of two squaring
circuits and a difference amplifier, according to the block diagram of
Fig. 8-13, will yield an output that is the difference of the squares of the
154, ELECTRON-TUBE CIRCUITS [CHAP. 8

input signals. This circuit is a simple combination of the basic circuits

that are required to perform the separated indicated operations.
8-6. Square-root Circuit. A block diagram showing the elements for
yielding the square root of a given voltage is illustrated in Fig. 8-14.

D/frerence aeff Sq,!ar~nq

amp'/"/er clrcud

!Kj(ae~-el)
Amp//f/er KiI!2(ae:-el )

FIG. 8-14. A square-root circuit.

The input to the circuit is el, which is applied to one input of a difference
amplifier of the type illustrated in Fig. 8-2. Likewise, it is supposed that
a voltage e2 appears at the input of a squaring circuit, the output of which
is applied to the second input to the difference amplifier. The output
from the difference amplifier, which is the input to a simple amplifier of
gain K 2, is then Kl(ae~ - el). But the output from the amplifier is
actually the source of the voltage e2 which appears at the input to the
squaring circuit. This requires, therefore, that
K1K2(ae~ - el) = e2
But for large amplifier gain K 2 ,

~=O
K1K2
which requires therefore that

It follows from this that the output voltage e2 is then related to the input
voltage by the relation
(8-22)

8-7. Dividing Circuit. A block diagram showing the elements of a

circuit for yielding an output which is proportional to the ratio of two
given voltages is given in Fig. 8-15. In this circuit the two input voltages
are el and e2, one of which is applied to the difference amplifier, the second
of which is applied to a multiplying circuit. An examination of the
circuit shows that the output from the amplifier, which provides the
output and which also furnishes one input to the multiplying circuit, is
eg = KA(el - ke2ea)
SEc.8-8J ELECTRONIC COMPUTING CIRCUITS 155
If the amplifier provides a high gain, then it f()llows that

~=O
KA
from which

or finally
(8-23)

D/fference lce2 e3 Mulftplylnq

ampl/ft"er c/rcuif

!A(e.,-ke2 ea )
83

I1A(erlcl!28a) Ou/put
Ampl/f/er

FIG. 8-15. A circuit which divides two voltages.

~ 8-8. General Feed-back Circuit-Approximate Analysis. l A num-

ber of computing circuits utilize feedback in special ways in order to
achieve the desired results. A circuit that is used extensively in such
applications is given in Fig. 8-16, together with the equivalent circuit.

,,
,,,
,, t I
I
eo
, eo
,,, iI
,, ~ ____+-__~~ __~-L

FIG. 8-16. A general feed-back circuit and its plate-circuit equivalent.

This circuit was examined in Sec. 6-7 in connection with its application
as an anode follower. An exact analysis of this circuit is included in
Sec. 6-7. However, an approximate analysis yields important informa-
tion, and this will be examined before the exact analysis is discussed.
In this connection, an application of the Millman network theorem
yields directly
E - EkIY 1 + EkpYf (8-24)
ka - Y 1 + Yf + Yg
This is rewritten in the form

(8-25)
156 ELECTRON-TUBE CIRCUITS ICHAP.8

But the gain of the tube iti, by definition,

K == E pk (8-26)
E gk
and the amplifier resultant gain is
K = E pk (8-27)
T Elk

Then Eq. (8-25) becomes

K = Y1 + Y, + Yg E kg _ Y1 (8-28)
r Y, Ekl Y,
or

Solving for Kr yields

Kr (1 _ Y
1
+ ~~ + Yo ~) = _ ~; (8-29)
This becomes
(8-30)

or equivalently
Z, 1
Kr = - - (8-31)
Z11 _ ! _ Z, Zl + Zg!
K ZlZg K
Some important results can be obtained from this expression. As an
approximate analysis, suppose that the grid impedance Zg» Zl. Equa-
tion (8-31) becomes

If one defines, for convenience,

(8-33)
then
K r = - K 'r 1
1
K' (8-34)
1-----'=
K K
This expression shows that the resultant gain of the amplifier is slightly
less than the quantity K~ defined in Eq. (8-33). Moreover, if it is
assumed that the gain of the tube as defined in Eq. (8-26) is large, and this
is generally true for a pent ode, then Eq. (8-34) becomes approximately
Kr == -K~ (8-35)
 SEc.8-8J ELECTRONIC COMPUTING CIRCUITS 157
Consequently, a great deal of insight into the circuit operation is possible
by an examination of K~, the network ratio, without having to perform
a detailed calculation of a complex circuit. In fact, noting that

Kr = -K~ = -ZjY I == E
pk
Elk

then
(8-36)
Consequently, if one considers lZjY I ] as an operator, then Eq. (8-36)
may be interpreted as an operational expression. That is, one might
consider [ZjY I ] as an operator which, when acting on ElK, yields the
output voltage E pk •
It is sometimes found very convenient to draw an approximate equiva-
lent circuit of the above diagram. Such an
approximate equivalent circuit is shown in
Fig. 8-17. The presence of the virtual
ground in the diagram is used to emphasize
1 that, owing to the large gain of the tube,
FIG. 8-17. An approximate
the change in potential at the grid for a equiva 't 0 f F'Ig, 8- 16 •
' 1en tClrCUI
'

1 given output potential is so small that it can

~ be assumed as a first approximation that the grid potential does not
-vary,
A number of special applications of this circuit will be considered, and
the corresponding more detailed analyses of several of these applications
will be carried out below. Among the direct applications of this circuit
are circuits for integration and differentiation,
Integrating Circuit. If the impedance elements in the circuit of Fig.
~f--;­
l1=!/Rf 1
Zf=l/cP

8-17 are chosen as shown in the accompanying diagram, then Eq. (8-36)
becomes
(8-37)

which shows that the output potential is related to the integral of the
input voltage.
For the particular case in which the input voltage is sinusoidal and of
the form
Ene = Em sin wt
then
lf E
E pk = - RIC '
m sm wt
d t = RIC
Em cos wt +k
158 ELECTRON -TUBE CIRCUITS [CHAP, 8

If the constants are adjusted to be

1
RIC =-
W
then the output voltage is
E pk = Em cos i,;t = Em sin (wt + 90)
This permits an output v.oltage which is the same amplitude and shifted
90 deg with respect to the input voltage, Such circuits may be adjusted
to yield the 90-deg phase shift to within 10 min or less,
If the input voltage is a square wave, then the output is a linear func-
tion of time and has the form illustrated in the accompanying figure.

It

Actually it will be found that the variation follows an exponential varia-

tion, the start of which is a linear function of time. The output wave
form will be somewhat as illustrated.
Differentiating Circuits. If the circuit impedances are chosen and con-

---I~
1j=Cp I Zf~R

nected as shown in the accompanying diagram, then Eq. (8-36) becomes

dE
E pk = -CRpElk = -CR at 1k
(8-38)

This expression shows that the out-

put voltage is the time derivative of
the input potential.
I
I
I
8-9. The Integrating Circuit. The
I
I equations of Sec. 6-7 permit an exact
eo
I
I analysis to be made of the integrating
I
I
circuit. The circuit of such a com-
FIG. 8-18. A common integrating
mon integrating circuit is shown in
circuit. Fig. 8-18. The general expression
for the output 'is given by Eq. (6-46);
it becomes for this case, since
SEC. 8-9] ELECTRONIC COMPUTING CIRCUITS 159
Zl = Rl
Zg = Rg
Y, = Cp
where the operator p(= d/dt) is the usual time-derivative operator of
circuit analysis, and for a pent ode circuit,
K -!.. Y1(Cp - gm)
(8-39)
r - (Y 1 + Yg + Cp)Yz + Cpgm
This expression may be written in the form

(gm + Y)[
Z P
+(Yl+YO)YZ]
C(gm + Y z)
(8-40)
from which it follows that the output is related to the input voltage by
p-a
eo = A - - el (8-41)
p+b
gm
where a = C

b = (Yl + Yo) Yz (8-42)

gm + Yz C
Yl
A =
gm+ Y z
• For the application of a step function el, the results are expressed
analytically by the expression

eo = A [ e- bt - E(1 - e- bt ) ] (8-43)

which has the form illustrated in the accompanying figure. If the con-

e'l

~7
•t
stant b is small, a condition that is readily achieved, then Eq. (8-43)
reduces to the approximate expression
eo == A[1 - (a + b)tl (8-44)
160 ELECTRON-TUBE CIRCUITS

The output impedance of this circuit is of some interest. Equation

(6-51) becomes for the present case

Yo ~ (YP + Y 1 +
Cpgm
Yo + Cp
) ~
Y1 +
CgmP
Yo + Cp
or
Zo = Y 1 + Yo + Cp = .l + Y1 + Yg (8-45)
gmCp gm gmCp
This expression becomes, for an applied a-c voltage,
Zo = .l + Rl + Rg (8-46)
gm jwCgmRIRg
But the circuit that gives rise to an expression of this form is that illus-

trated in the accompanying figure. This circuit consists of a relatively

low series resistance 1/gm , which would ordinarily be of the order of several
thousand ohms or less, and a capacitor, the value of which can be made
very large. For example, suppose that the tube and circuit constants
are the following:
gm = 5,000 X 10-6 mho
C=lpf
Rl = Rg = 1M •
The equivalent circuit then becomes as shown, and the complete output

~
f;;;~#T .
circuit of the amplifier has the form sketched. Such a circuit as this

T 2500,uf
may be used as a voltage stabilizer, since, owing to the very large effective
capacitance across the output, a ripple of almost any frequency that
appears in Ebb is eliminated in the output of the circuit.
SEC. 8-10] ELECTRONIC COMPUTING CIRCUITS 161

Such a circuit as this finds extensive application as the reactance-tube

modulator of an f-m transmitter and
depends on the fact that the effective
output capacitance of the circuit
varies directly with gm of the tube, a R C
quantity that can be controlled by an ~f-+---+­
input signal. This application is dis- I
eJ. Rg
cussed in detail in Sec. 17-7.
8-10. Differentiating Circuit. The
FIG 8-19 A differentiating circuit.
basic differentiating circuit is iIIus- . .
trated in Fig. 8-19. This is substantially the circuit discussed in Sec.
8-8 (page 158), except for the presence of the series input resistor R. The
respective impedance elements contained in Eq. (6-46) are chosen to be

Zl = R +-1 YI = 1. _P_ a =
1
RC
Cp Rp+a
Zf = R f
Zg = Rg
There results, subject to the approximations of a pentode tube,
e . - YIgm
o = (Y I + Yf + Yg)Y z + Yr(Y I + gm) el
- YIg m
YI(Y Z + Y f ) + YAY + gm) + YgY z el
l
-gmp
p(Yl + Yf ) + R(p +a)[Yf(YZ + gm) + YgY I] el
gmp
p[Yz + Y f + R(YfYz+ Yrg m+ YgY z)] + aR[YtCYz+ gm) + YgYzl e]
(8-47) .
This may be written in the form
eo = A -P- el (8-48)
p+b
where

and
b= .! YfY + Yfg m + YgY z
Z
(8-49)
C (Y z + Yf ) + R(Y/Yz + Y/g m + YgYz)
lIC
R + Y z + Y/
Y/Yl + Y/gm + YgY z
The application of a step function to the input yields the expected
result, viz.,
eo = Ae-bt
162 ELECTRON-TUBE CIRCUITS [CHAP. 8

which has the form sketched.

,1:

~------------~f:

REFERENCES
1. Williams, F. C., F. J. U. Ritson, and T. Kilburn, J. lEE, 93, 1275 (1946).
2. As general references, consult I. A. Greenwood, Jr., J. V. Holdam, Jr., and
D. MacRae, Jr., "Electronic Instruments," M.I.T. Radiation Laboratory Series,
vol. 21, McGraw-Hill Book Company, Inc., New York, 1948.
Korn, G. A., Electronics, 21, 122 (April, 1948).
Ragazinni, J. R., R. H. Randall, and F. A. Russell, Proc. IRE, 35,444 (1947).
Mynall, D. J., Electronic Eng., June, 1947, p. 178; July, 1947, p. 214; August,
1947, p. 259; September, 1947, p. 283.

PROBLEMS
8-1. Derive Eq. (8-14) for the common plate summing chain.
8-2. Show that the output of the circuit ea in the diagram is proportional to the
ratio edel.

8-3. Show that the output of the circuit shown has the form

eo = -[Z2YI]e.
 ELECTRONIC COMPUTING CIRCUITS 163
8-4. Determine the character of the mathematical operation performed by
each of the feed-back circuits shown in the figure.

-iT C R

Cz
Rl 'it ri~
~F
8-5. Analyze the differentiating circuit shown in the figure. *

8-6. Analyze the differentiating circuit shown in the diagram.'

* Schmitt, O. H. and W. E. Tolles, Rev. Sci. Instruments, 13, 115 (1942.)

164 ELECTRON -TUBE CIRCUITS [CHAP. 8

8·7. Analyze the integrating circuit shown in the figure for this problem.

I
I
I
eg
I
I
I

./ 8-8. Analyze the integrating circuit in the diagram.

8-9. Analyze the integrating circuit shown in the accompanying figure.

8-10. What equation does the computer in the figure solve?

Oufputy

8-11. What equation does the computer in the diagram solve?

Amp//'fier
 ELECTRONIC COMPUTING CIRCUITS 165
8-12. Sketch the block diagrams of computers that will solve the following
equations:

(a) xi + X2 = Y
(b) x~ + xi = Y
X2
(c) XIX2 + Xl '7' Y

8-13. Show that in the indicated network the output voltage is related to the
input voltage by the equation

R I
I
eo
I I
Y t
8-14. Show that in the circuit shown the output voltage is related to the input
voltage by the expression

8-15. Show that the computer sketched yields a solution of the differential
equation indicated.

Hlnfeq~forl-l--+--~O:":U,".jPUf y
d y +ay : o
dx

8-16. Show that the computer sketched yields a solution of the equation
indicated.
Ou/puly
166 ELECTRON-TUBE CIRCUITS [CHAP. 8

8-17. Determine the differential equation that is solved by the computer

sketched.
 CHAPTER 9

UNTUNED POWER AMPLIFIERS

T HE voltage amplifiers discussed in Chaps. 5 and 6 are designed to

increase a voltage signal from a low level to one which is adequate for
operating some low-power circuit. Such amplifiers are generally operated
in class A since the amplification is to be accomplished without distortion.

R[

FIG.9-1. The schematic and equivalent circuits of a simple series-fed power amplifier.

A power amplifier serves to supply an appreciable amount of power to

some power-absorbing circuit, although in general it must be accomplished
under very low grid driving-power demands. Power amplifiers may be
operated as class A, B, or C or at any point between these limits, the
choice of operating conditions being determined by the ultimate purpose
of the amplifier. If the amplifier is to reproduce the audio spectrum with-
out distortion, then the amplifier must be operated in class A if a single
tube is used. If two tubes are used in a push-pull circuit, then the
amplifier may be operated in class AB or class B. If the amplifier is to
reproduce the input wave shape over a very narrow range of frequencies,
tuned class B or tuned class C amplifiers may be used. Only a-f ampli-
fiers will be considered in this chapter.
9-1. Class A Triode Power Amplifiers. The basic schematic diagram
of a typical series-fed power amplifier and its equivalent circuit are given
in Fig. 9-1. It is observed that this circuit is identical with that of Fig.
5-1 for the simple voltage amplifier.
If it is assumed that the dynamic curve is linear over the entire range
of operation, then the plate current is given by
I _ J.l.Eg ( )
P - Rl + rp 9-1

and the power supplied to the load is

167
168 ELECTRON -TUBE CIRCUITS [CHAP. 9

P = I;Jl1 = fL2E~ 4 R!
4rp (1 + ~ly rp
(9-2)

A sketch showing the variation of the output power as a function of Rz/rp

is given in Fig. 9-2. I t can be seen

(undsof'~)

p
2 2

1.0~rp I
~ O: 4~2
==:
fi
..-=---
that the power curve reaches a maxi-
mum at the point at which Rl/rp = 1,
although this maximum is quite
broad. P is a slowly varying func-
tion of Rl in the neighborhood of the
maximum, and the power is at least
o I 2 3 88 per cent of its maximum value for
Rdrp values of Rl/rp ranging from 0.5 to
FIG. 9-2. Variation of power output 2.0. This condition shows that the
as a function of resistance ratio power loss is less than 2.25 db for all
Rz/rp. values of R between 0.5rp and 2rp.
Since the maximum power transfer occurs when the load resistor equals
the internal plate resistance of the tuqe, it is necessary to use tubes with
low values of rp in order to obtain reasonable amounts of power with
nominal values of plate-supply voltage. Since the Ym of a tube cannot be
designed over very wide limits, then tubes with low rp also possess low
values of}J,. As a result, large grid excitation voltages are required for
appreciable amounts of power output. Note from Eq. (9-2) for a given
value of Ey that tubes which possess large values of

possess high output power capacity. In fact, the power sensitivity which
was defined by Eq. (4-4) becomes, under the conditions of maximum
power transfer, simplY}J,gm/4.

FIG. 9-3. The output current and voltage wave forms in a triode power amplifier.

To determine the power output directly from the static plate character-
istic of the tube, it is necessary only to draw the appropriate load line on
SEC. 9-2] UNTUNED POWER AMPLIFIERS 169
these characteristics and read the significant information from the dia-
gram. Thus, by referring to Fig. 9-3, it follows that

and (9-3)

The power output is

p _ Emlm _ I;"Rz
- -2- - -2-

which may be written in the form

(9-4)

If distortion is not negligible, the harmonic components must be evalu-

ated according to Eqs. (3-22). The total power output is then

P = (Bi + B~ + m+ .) Rz
2
This may be written as

which becomes, by Eq. (3-21),

P = (1 + D~ + Dl + .. ')Pl
or
P = (1 + D2)P 1 (9-5)
where D is the total distortion. Notice, however, that if the total dis-
tortion is high, say 10 per cent, then P = 1.01P 1• That is, a 10 per cent
distortion represents a power of only 1 per cent of the fundamental.
Thus, with little error, the output power is approximately that, of the
fundamental frequency component only.
9-2. Output Circuits. It is not always feasible, nor is it generally
desirable, to connect the load directly in the plate lead, as shown in Fig.
9-1. Among the reasons for this are that the quiescent current through
the load resistor represents a considerable waste of power, as it does not
contribute to the a-c component of power, and that the quiescent current
may cause a serious polarization of the output. For example, it is inad-
visable to pass a large d-c current through the voice coil of a loud-speaker.
For these reasons, the transformer-coupled load is used extensively,
although the parallel-feed system may be used.
170 ELECTRON-TUBE CIRCUITS [CHAP. 9

The circuit of the parallel-feed system is illustrated in Fig. 9-4. It is

clear that this is just the impedance-capacitance coupled system dis-
cussed in Chap. 5. In this system the plate supply is connected to the
plate of the tube through a high inductance L, the load resistor being
connected across the output through
c a blocking capacitor C. The inductor
Lmust be so chosen that wL» R l , and
Rl C must be so chosen that l/wC «Rz
over the operating range of frequen-
cies.
Eee The voltage and current relations
FIG. 9-4. A parallel or shunt-fed for the parallel-feed system are illus-
power amplifier.
trated in Fig. 9-5. Observe that the
quiescent current through the tube is determined by Rr" the resistance of
the inductor, although the "dynamic" resistance into which the tube is
working is R l • Since the static resistance of the choke or the transformer
winding is usually small, the static load line is almost vertical.

FIG. 9-5. The dynamic and static load lines of a shunt-feed or transformer-coupled
amplifier.

Suppose that the load resistance into which the tube works is small; for
example, the resistance of the
voice coil of a dynamic speaker 2

0
usually ranges from about 5 to 15
ohms. If such a low resistance : I Rl
Rl
I
load were used in either the series- II _

or the shunt-feed circuits, only a I-

I
I
very small power output would be
possible, most of the power being FIG. 9-6. A simple transformer-coupled
lost within the tube resistance. load in a power amplifier.
In this case, and in fact in any
case in which the load resistance does not properly match the tube resist-
ance, the use of a transformer as an impedance matching device will per-
mit optimum power transfer. Such a system is illustrated in Fig. 9-6.
The impedance-transforming property of an ideal transformer follows
SEC. 9-3] UNTUNED POWER AMPLIFIE·RS 171
from the simple transformer relations
N1
E1 = - E2
N2
and (9-6)
[1 = -N2 12
N1
where E1 and E2 are the primary and secondary potentials, respectively,
and [1 and [2 are the primary and secondary currents. The ratio of the
above yields

which may be written as

Rf = (NN21)2 Rl = 1..
n2
Rl (9-7)

since Ell [1 and Ed [2 represent the effective input and output impedances.
When the turns ratio NdN2 is greater than unity, the transformer is
called a step-down transformer; with the ratio N11N2 less than unity, it
is a step-up transformer.
Equation (9-7) is true for an ideal transformer. In general, however,
the coupling is not perfect, the primary and secondary resistances are not
negligible, and the core losses cannot be neglected. By taking these fac-
tors into account, the input impedance is given by

(9-8)

where R1 and R2 are the primary and secondary winding resistances, L~

and L~ are the primary and secondary leakage inductances, and Zl is the
load impedance.
The same distinction between the static and dynamic load lines must
be made for transformer-coupled loads as for the shunt-feed circuit, and,
as noted, Fig. 9-5 applies for both circuits. But for the reasons discussed
in Sec. 5-5 the frequency response of the transformer is not flat for all
frequencies. However, the effects are less severe than for a transformer
interstage coupling, since over the audio range the transformer capac-
itances, tube capacitances, and stray capacitances appear across a rela-
tively low plate-resistance tube or across the low-output load resistance.
9-3. Maximum Undistorted Power. The foregoing analyses, which
are based on the linear equivalent plate circuit, are not completely valid
owing to the curvature of the dynamic characteristic, particularly in the
region of small plate currents. In order to obtain the maximum possible
power output without making the instantaneous plate current too small
172 ELECTRON-TUBE CIRCUITS [CHAP. 9

during the most negative part of the applied signal, and without driving
the grid positive at the positive peak of the applied signal, it is necessary
to maintain a careful balance among the grid bias, load impedance, plate
supply voltage, and plate resistance.
To find the expression for the output power under these conditions, and
also to determine the appropriate conditions in order to achieve the
present results, refer to the graphical construction of Fig. 9-7. Since the
distortion that results at small plate currents arises from the curvature
of the static characteristics, this
region is eliminated by setting Irrrin
at an appropriate value. This is
the crosshatched area of the dia-
gram. Thus the grid is allowed
Il77i" ~1*~~~,ss::~;.§:~_ _• to swing from any point Q, corre-
Dt:: Emin Eo E'mtlx e6 sponding to the pot e n t i a 1 Eo
2-?n-p-+-Em'+-£'m->l (which would be Ebb if an ideal
FIG. 9-7. Graphical construction for shunt-fed or transformer-coupled
determining the operating conditions and
maximum undistorted power output from load were used) between zero grid
an amplifier. bias and to that bias which corre-
sponds to I rrrin. It will be assumed
that the distortion is negligible in this region. The characteristics are
essentially linear in the operating region, with a slope corresponding to rp.
To find the value of load resistance for which the power will be a maxi·
mum, refer to Fig. 9-7. It is noted that

But since

then it follows that

Eb = ED + 2Imrp + ImRz
= ED + 1m(Rz + 2rp)
Solving this expression for 1m , there results
I _ Eb - ED
m - Rl + 2rp (9-9)

The power to the load is then given by the expression

p =- I~RI = (E~ - E D )2 4R/2rp
(9-10)
2 16rp (1 + R/2rp)2
A sketch showing the variation of the output power as a function of R/rp
is given in Fig. 9-8. This curve, like that of Fig. 9-2, reaches a maximum,
but in this case at the point at which Rz = 2rp, although the variation is
not rapid in the region of the maximum. The power remains at least 88
SEC. 9-31 UNTUNED POWER AMPLIFIERS l73
per cent of its maximum value for load resistances Rl ranging from rp
to 4rp.
To find the appropriate bias for these conditions, combine Eq. (9-9)
with the fact that the current changes lunifso/Eb-ED);
from 1b to 1m when the signal voltage (' 1.0 /6rp
Egis equal to Eo. Thus
1m =
Rl
ILEe
+ rp
0.8

(9-11) P 0.6
0.4
r--

0.2 II/
I.
I
v -- -

The result, by equating Eq. (9-9) to

00 2 3 4 5
(9-11) yields
. Rdrp
FIG. 9-8. Variation of output power
(9-12) as a function of load Rz.

in which Rl has been set equal to 2rp. The value of En is obtained directly
from the curves.
The maximum undistorted power output becomes from Eqs. (9-10)
and (9-12)

Further, since Ee = E gm , this becomes

IL2E~m _ 2 E2
P max -_ ~ - "9 ILgm g (9-13)

and the power sensitivity at optimum power output is

. . . P 2 \ "" """"
Power senSItIvIty = E2 = "9 ILgm i mhos (9-14 )
g

which is slightly less than that for the condItions of Sec. 9-l.
The results showing the power out-
put and second-harmonic distortion as
+- 15 '" 3 t---+--j'l---+-~ a function of Rl of a type 5A3 triode
§ ~ are illustrated in Fig. 9-9. Optimum
~IO ~2 output is obtained at about Rl = 2,500
.~ ~ ohms, which is approximately three
"t 5 0.. I times the plate resistance rp of the tube.
-t; Although the second-harmonic distor-
a 0 0 o!:-.-_-c-l--:!:2-""""~--4:---5;!c---;!6 tion is not negligible at this point, a 5
LOCJ!a resishmce,kilohms per cent distortion is usually tolerable.
FIG. 9-9. Power output and second- The above analysis is based on the
harmonic distortion of a type 6A3 use of a plate-supply source of so-
triode as a function of load resist-
ance. called nominal value (about 300 volts).
If it is assumed that a platp source of
any voltage is available, then with increases in the value of Fi>!' tlw lllti-
174 ELECTRON-TUBE CIRCUITS [CHAP. 9

mate limitation will be imposed by the allowable plate dissipation.

Under these circumstances, it is found that the circuit should be operated
with a value of Rz that greatly exceeds 2rp.l
9-4. Plate-circuit Efficiency. The foregoing discussion indicates the
methods for calculating the output power of a power amplifier. The a-c
power so obtained is converted from the d-c plate supply by the vacuum
tube. The ratio of these quantities is called the plate-circuit efficiency of
the amplifier. Thus
7]p = a-c power output to the load X 100o/c (9-15)
d-c power input to the plate circuit 0

Suppose that P p denotes the average power dissipated by the plate.

Then by the principle of the conservation of energy
Ebdb = IgR 1 + Eplp + Pp
where R 1 is the static load resistance. By solving for P p, there obtains
Pp = Ebblb - IgR 1 - Eplp
But as

then P p has the form

(9-16)
This equation expresses the amount of power that must be dissipated by
the plate and represents the kinetic energy of the electrons that is con-
verted into heat at the plate. Notice in particular that the heating of the
anode is reduced by the amount of the a-c power that the tube supplies
to the load. Hence, a tube is cooler when delivering power to a load
than when there is no such a-c power transfer. This is a very important
factor in the operation of high-power r-f transmitting tubes, since such
tubes are ordinarily operated close to the rated allowed plate dissipation.
If for any reason the output circuit becomes slightly detuned, with a
consequent decrease in output power, the plate power may become
dangerously high.
The plate-circuit efficiency may be written in several different forms.
From Eq. (9-15),
7]p = EEpIIP X 100% (9-17)
bb b
This may also be written as
Po
7]p = Po + Pp + nR 1
X 100% (9-18)

Clearly, a large value of 7]p means a small value of P p for a given output.
SEC. 9-4] UNTUNED POWER AMPLIFIERS 175
This means that a smaller tube with a smaller plate-supply source may
be used.
It is possible to obtain an approximate expression for the theoretical
value of 'YIp for the series-fed and the shunt-fed circuits. Consider that

an ideal tube is used in an amplifier circuit. The plate characteristics of

such a tube would have the form illustrated in Fig. 9-10. Suppose that
the grid does not swing beyond Ee = 0 and may swing to give zero cur-
rent. Then, by the proper choice of E e ,
Ib = 1m
and
% (9-19)

For the series-fed circuit, the point marked EffiB.x = Ebb, and from the
diagram,

Hence Eq. (9-19) becomes

25
'YIp = 50 (E Em 1 )
2 m + mrp
%
from which, since

then
25
'YIp = 1 + Tp/Rl % (9-20)

The theoretical maximum plate-circuit efficiency for the series-fed ampli-

fier is 25 per cent. For the conditions of maximum power output, when
R = r p, 'YIp = 12.5 per cent. Actually, owing to the limited range of
operation without distortion, 'YIp seldom exceeds 10 per cent in practice.
Evidently, the linear vacuum-tube amplifier is an inefficient device for
converting d-c into a-c power.
In the shunt-fed system a means has been devised for eliminating the
d-c power loss in the load. This results in an improved plate efficiency.
176 ELECTRON-TUBE CIRCUITS [CH~. 9

If the static resistance is assumed negligible, then

Ebb = Eb = Em + 2lmrp
and
'fJp = 50 ~: = 50 Em !2lmr p %
This reduces to the form
50
% (9-21)
'fJp = 1 + 2rp/R z
The theoretical maximum plate-circuit efficiency of the shunt-fed or
transformer-fed amplifier is 50 per cent. Thus the elimination of the
static power loss in the load reflects itself as an improved plate efficiency.
For the conditions of maximum output power, when R = 2rp, 'fJp = 25 per
cent. However, since the static resistance is not negligible and since the
current lmin cannot be taken as zero if distortion is to be avoided, the
actual plate-circuit efficiency will be less than the 25 per cent figure.
9-5. Power Pentodes and Beam Power Tubes. The plate character-
istics of the power pentode are markedly different from those of a triode,

~80r-----~~--~-r------+-------r------1
E
·f·:
er: 60
L.
I--+-""-..
:J
o
Q)

~~~~~-t~~~~~~~~~~~r------J
c. I

' ...e
-25
-.70
L~====~======l;======f=~~~E=~~-J.~~~
00 100 200 300 400 500
eh, plC'lte volts
FIG. 9-11. Plate characteristics of a 6F6 power pentode.

and the graphical analyses given above are not valid for pentodes. Since
the characteristics of the beam power tubes are similar to those of the
power pentode, the discussion to follow applies for both the power pentode
and the beam power tube.
Power pentodes differ from triodes principally in the character of the
dynamic curve with increasing load resistances. In the triode, the dis-
SEC. 9-5] UNTUNED POWER AMPLIFIERS 177
tortion decreases as the magnitude of the load resistance increases. This
follows from the fact that the dynamic curve becomes increasingly linear
as the load resistance becomes higher. In the power pentode, the
dynamic characteristic is critically dependent on the load resistance, with
excessive curvature at both the high and the low values of load resistance.
Moreover, the critical load resistance to be used cannot be related
analytically with the plate resistance of the tube. This resistance is

IS
60 E
+--
<:
OJ
t:
~------+-------~--~~~------~40a
)!
IS
0..
r-------+-----~~-------+------~20·~

.~----~------~----~~----~o
-40 -30 -20 -10 0
ec,biOis
FIG. 9-12. Dynamic characteristics of a 6F6 for three values of plate load resistance.

always less than the plate resistance of the tube. It might appear there-
fore that the output power capacity of the tube would be too low to make
the tube feasible. Actually, owing to the large J.l.gm product, even with
the appropriate Rl the power output is usually higher than with the tube
connected as a triode, and this with a smaller grid driving-voltage
amplitude.
To examine the matter in somewhat greater detail, refer to Fig. 9-11,
which gives the plate characteristics of a 6F6 power pentode. It will be
supposed that the load is kansformer-coupled to the tube and that the
plate potential is maintained at 300 volts. Three load lines are shown,
2,500, 7,000, and 10,000 ohms. The corresponding dynamic curves are
given in Fig. 9-12. The following example WIll help clarify the situation.

Example: Calculate the output power, the plate-circuit efficiency, and the
second, third, and fourth harmonics for the 6F6 that supplies power to a loud-
speaker, the effective resistance of which is changed to have values of 2.5k , 7k, and
10k ohms. A 300-volt source is available, and the tube is biased at -20 volts.
Solution: The important data from Figs. 9-11 and 9-12 are included in the
tabulation. The general character of the results is given graphically in Fig. 9-13.
Notice that optimum power transfer occurs somewhat above 10k , which is very
small compared with the tube resistance of 78,000 ohms.
 178 ELECTRON -TUBE CIRCUITS [CHAP. 9

~ 2.5k 7k 10k
XI 2.5k _7k J
i
l()k

Ima.x 86 64 53 Bo 9 4 1
Iy, 52 4!J 46 B, 44 35 31
h 25 25 25 B. 9 4 I 1
Ly, 6 6 6 B3 -1 -4 -5
Imin 1 1 1 B4 +>~ -72 I
I

~ 2.5k 7k 10k
~ 2.5k 7k 10"

D. 20.5 11 .5 3.2 I a-c 45.0 35.5 31

D3 -2.3 -11.5 -16 P a-c 2.5 4.4 4.8
D4 +1 -1.4 I d-c 34 29 26
P d-. 10.2 8.7 7.8
'1p 24.5 50.6 61.6

9-6. Push-pull Amplifiers. The use of two tubps in parallel provides

twice the output power of a single
tube with the same distortion. A
5
push-pull amplifier circuit is a much
more desirable connection for two
4
tubes. In this circuit, the two tubes
are arranged as shown in Fig. 9-14.
The excitation voltages to the grids
of the two tubes must be of equal
magnitude, but of opposite phase.
..... In the circuit shown, a center-tapped
§IO
transformer is used to provide the
o °O~----~~5Tk------~~ two equal voltages that differ by 180
deg. A number of vacuum-tube
LOlild resisfGmce
circuits are possible for achieving
FIG. 9-13. Operating characteristics
of a 6F6 pentode as a function of load these results, and several of these
resistance. will be examined below.
To examine certain of the features
of such an amplifier, suppose that the input·grid voltage to one tube is of
the form
egl = Egm cos wt

The plate current of this tube will be represented in general by the expres-
/ sion (see Sec. 3-6)

i bl = Bo + Bl cos wt + B2 cos 2wt + Ba cos 3wt + (9-22)

SEc.9-6J UNTUNED POWER AMPLIFIERS 179
The corresponding signal to the second tube is
eg 2 = -Egm cos wt = Egm cos (wt + 71")
and the output plate current is
i b2 = B 0 + B 1 cos (wt + 71") + B 2 cos 2 (wt + 71") + B a cos 3 (wt + 71") +
which has the form
i b2 = Bo - Bl cos wt + B2 cos 2wt - Ba cos 3wt + (9-23)
But, from Fig. 9-14, the currents are in opposite directions through the
output transformer windings. The total output is then proportional to
the difference between the plate currents in the two tubes. This is
i = k(ibl - i b2 ) = 2(Bl cos wt + Ba cos 3wt + ...) (9-24)
This expression shows that the push-pull circuit balances out all even
harmonics in the output and leaves
the third-harmonic term as the
principal source of distortion. ]
Another feature of importance in
the push-pull system is evident
from the circuit of Fig. 9-14. It is
observed that the steady compo-
nents of the plate currents flow in FIG. 9-14. The basic push-pull circuit.
opposite directions in the windings,
thus opposing each other magnetically in the transformer core. This
eliminates any tendency toward core saturation and the resulting dis-
tortion that might arise from the magnetization of the transformer core.
The effects of ripple voltages that may be contained in the power supply
owing to inadequate filtering will be balanced out in the push-pull circuit.
This is so because the currents that are produced by this ripple voltage
are in opposition in the output transformer and hence will not appear in
the load. Of course, the effects of the ripple voltage that appear on the
grids of the amplifier will not be balanced out and will be noticeable with
the signal.
Another feature of this amplifier is that, under self-biased conditions,
there is no need for a by-pass capacitor across the cathode resistor. This
follows from the fact that the voltage which appears across the self-bias
resistor Rk is (ib1 + i b2 )R k. But this is, from Eqs. (9-22) and (9-23),
(ibl + i b2 )R k = 2R k (h + Bo + B2 cos 2wt + B4 cos 4wt + ...)
But for tubes operating in class A the harmonic amplitudes are very small
and are therefore not significant.
One of the particularly significant features of the push-pull amplifier
180 ELECTRON-TUBE CIRCUITS [CHAP. 9

is that the output power possible with the two tubes for a given total
distortion is higher than twice that of the single tube. This results
from the fact that, with the automatic cancellation of even harmonics in
the output, the tubes may be driven
harder until the third-harmonic
terms become significant. Not e
also that, with the increased grid
t-------'--t-----<q ~ Rl
EJ drive, the rectification component
2BoRk becomes significant and adds
to the bias 2hR k , if self-bias is used.
As a result, the effective bias in-
creases, with consequent reduction
FIG. 9-15. Equivalent circuit of the
class A push-pull amplifier of Fig. 9-14. of output power. This means that
the output power of a push-pull
amplifier under otherwise similar conditions will be higher with fixed bias
than with self-bias.
9-7. Equivalent Circuit of a Class A Push-pull Amplifier. Suppose
that both tubes of the push-pull ampli-
fier are identical and that J.l and rp are
constant over the range of operation. - ~
The equivalent circuit of the system flEgk
Rlf
then has the form given in Fig. 9-15.
Observe that the connection between the pEgk
cathode terminals and the mid-point of
the output transformer does not carry a
fundamental frequency component of FIG. 9-16. The simplified equiva-
current, owing to the cancellation that lent circuit of the class A push-pull
occurs. This connection may be omitted amplifier.
from the diagram without influencing the operation. The resulting cir-
cuit then has the form given in Fig. 9-16. In this diagram

(9-25)

The resultant current is given by

(9-26)

which may be written in the form

(9-27)
 SEc.9-8J UNTUNED POWER AMPLIFIERS 181
The total power delivered to the load is then

p = 12R~ = 2( J.l.Ea ,)2 Rf2

Rz
(9-28)
rp + "2
This expression may be interpreted to mean that the total output power
is twice the power of each tube considered to be working into the equiva-
lent load resistance RU2.
A more significant expression results by writing Eq. (9-28) in the form

p = ( (rp/2)
J.l.Eg
+ (RU4)
)2 "4Rf (
9-29
)
'fJQhiS may be interpreted to show that the class A push-pull amplifier may
be represented by a single composite generator which has emf MEg, with
an internal resistance r p/2, and which works into a load resistance equal
t RU4. It is possible, in fact, to derive a set of static characteristics of
the composite tube from the tube plate characteristics and to obtain
significant operating information from this.
9-8. Composite Static-characteristic Curves. The composite static
characteristics of the push-pull amplifier may be obtained from the plate

FIG. 9-17, The potentials in the push-pull amplifier.

characteristics of the individual tube by a graphical construction, 2 It is

assumed that the output transformer is ideal, whence the voltages across
each half of the transformer are equal. The situation is illustrated in
Fig, 9-17, The load voltage is given by

(9-30)

Also, in the ideal transformer, the net primary ampere turns are equal to
the secondary ampere turns, from which
(9-31)
Thus the load voltage is
. e = t'R z = ('tbl -
.)NI
N 2 RI
tb2 (9-32)
182 ELECTRON -TUBE CIRCUITS [CHAP. 9

By combining Eqs. (9-30) and (9-32) there results

ep = (ibl - ib2 ) (~~y R, (9-33)

which may be written in the form
ep = (. . )4"R:
~bl - ~b2 (9-34)
where R~ is the plate-plate resistance.
The following relationships are evident from an inspection of the dia-
gram of Fig. 9-17:

(9-35)

FIG. 9-18. To construct the composite static characteristics of a push-pull amplifier

from the plate characteristic.

It follows from these equations that, when the plate-cathode potential

ebl of tube T1 decreases from the quiescent-point value Eb by an amount
e p , then the corresponding potential eb2 of tube T2 increases beyond Eb
SEC. 9-8] UNTUNED POWER AMPLIFIERS 183

by a like amount. Also, when the grid-cathode potential eel increases

beyond Eee by the signal voltage eo, the corresponding value of ee2 decreases
below Eee by eo. These conditions are shown in Fig. 9-18 for three
different values of ep • In these diagrams point A corresponds to condi-
tions ebl = Eb - e p , eel = Eee + eo, the tube current being i bl. Point
B is that for eb2 = Eb + e p , ee2 = Eee - eo, and the tube current is ib2.
Point C has the ordinate ibl - ib2
for the chosen value of ep •
The composite static character-
istics are the family of curves of
i bl - i b2 vs. eb, with grid signal volt-
age as a parameter. Clearly, point
C is one point on the composite
static curve for the signal voltage
eo. Other points on this character-
istic are found by maintaining eo FIG. 9-19.
constant and by varying ep • The
construction for two other values
e~ and e~ are given in Figs. 9-18b and 9-18c. These locate two other points
Cf and Cff on the composite static characteristic. The complete composite
static is given in Fig. 9-19.
Several significant features are evident from Fig. 9-19. The composite
static characteristic extends above and below the zero current axis.
Also, the composite static curve is
much more linear than the plate
characteristics of the individual
• tube .
An alternative method for obtain-
ing the composite static characteris-
tics was described by Thompson. 3
According to this method, the plate
characteristics of the tube are
plotted in the usual way. The
curves are also plotted in an in-
FIG. 9-20. The Thompson method of verted manner, with the voltage
obtaining the composite static char-
acteristics of a push-pull amplifier. scale shifted so that the voltages Eb
of both sets of curves coincide with
each other. This construction is shown in Fig. 9-20. The inverted curves
represent the plate family of tube T2. The two methods are essentially
equivalent.
The foregoing discussion of the graphical construction is general and
applies to any type of tube under operation under class A, AB, or B con-
ditions. Although the illustrations are for triodes, the curves for other
184 ELECTRON-TUBE CIRCUITS [CHAP. 9

tube types are obtained in the same way. In fact, owing to their shape,
the composite static characteristics for pent ode-type tubes are more easily
obtained than the triode curves. The simplification results because the
current i b2 remains substantially constant for large variations in eo. The
composite static characteristics for the 6F6 pentode are shown in Fig. 9-21.

L ...::::::~~~~~i~~~~~~~8~.~~~
i oz !.. Q eh--eg~O
~---ep--~--ep--
FIG. 9-21. Composite static characteristics of a 6F6 pentode push-pull amplifier.

9-9. Composite Dynamic Characteristic. The composite dynamic

characteristic of a push-pull amplifier is obtained from the composite
static characteristics in precisely the same way that the dynamic curve
is obtained from the plate characteristics of a single-tube amplifier. This
requires drawing the effective
composite load line on the compos-
ite static curves and noting the
points of intersection of the two,
which are points on the composite
dynamic curve. This construc-
tion also permits obtaining the
load line into which each tube is
working and hence also the dy-
namic characteristic of the indi-
vidual tubes. The construction
FIG. 9-22. The load line of the composite
is shown in Fig. 9-22. Clearly,
circuit, and the load line of one tube of the the intersection of the push-pull
push-pull amplifier. load line with the composite plate
characteristics gives points on the
composite dynamic characteristic. That is, points A, B, C, D are points
on the composite dynamic curve. These points are replotted in Fig. 9-23
to give the composite dynamic curve.
To find the dynamic characteristic of each tube, the procedure is
essentially the reverse of that of Sec. 9-8. In particular, consider the
SEC. 9-10] UNTUNED POWER AMPLIFIERS 185
point B in Fig. 9-22. This point is a representation of i bi - i b2 corre-
sponding to the appropriately chosen value of ep • But the point i bi lies
vertically above this point, by an amount i b2 , and must lie on the plate
characteristic of the tube. This defines the point Ai. The other points
are obtained in a similar way, and the results are shown in Fig. 9-22.
These points are plotted also in Fig. 9-23. The points A 2 , B 2, C 2, D2
for tube T2 are obtained by symmetry from the corresponding points of
tube Tl.

wi

wt

FIG. 9-23. The single-tube dynamic characteristics and the composite dynamic curve
of the push-pull amplifier. The wave shape in each tube and in the output are also
shown.

I t is interesting to note that the composite dynamic curve is practically

a straight line, although the individual tube dynamic curves are markedly
curved. Thus, for a sinusoidal input, the total output current is closely
sinusoidal, although the current in each tube is not sinusoidal, in general.
If only the output current is desired, there is no requirement for the
individual tube dynamics. If the plate-circuit efficiency is required, the
individual tube dynamics are required in order to calculate the value of
h and the d-c component Bo that results from the partial rectification
due to the curvature of the dynamic. The total d-c power input to the
amplifier is 2Ebb (h + Bo).
9-10. Powei-Oufputand Distortion in Push-pull Amplifiers. Owing
to the fact that the dynamic curve must be an odd function, by virtue of
the manner of its construction, then for the composite circuit

(9-36)
 186 ELECTRON-TUBE CIRCUITS [CHAP. 9

Under theoo circUlllBtanceB the five-point Bchedule of Eq. (3-20) reduceB to

Bo = B2 = B4 = 0 }
Bl = %(lmax + 1J,i) (9-37)
Bs = ~(lmax - 21J,i)

The fundamental power output is given by the expression

(9-38)

By neglecting the harmonic components of power, the total power is

given by
(9-39)

where Em and 1m denote the peak values of the a-c output voltage and
current, respectively. These values are obtained directly from t,he
curves of Fig. 9-22, since 1m = 1A and Em = Eb - E A, whence
p = (Eb - E A )I A
(9-40)
2
:/ To find the maximum output power in the push-pull class A system
i utilizing triodes, use is made of the fact that the load resistance should
Iequal the internal resistance of the equivalent or composite generator.
I This follows from Eq. (9-29) for the class A amplifier and requires that
! R~/4 = r p /2. This requires that the slope of the effective load line must
I be equal to the reciprocal of the composite static characteristic, which
L,has a value of rp/2.
Suppose that the tubes are operated in push-pull class B. N ow, since
the tubes are biased to cutoff, then either one or the other of the two tubes
will be supplying current to the circuit and each contributes power for
one-half of each cycle. Consequently, the equivalent generator will be
one with an internal resistance equal to rp of the tube. The maximum
power under these conditions will be obtained for R~/4 = rp. It is
reasonable to expect that the internal resistance of the equivalent source
of a push-pull class AB amplifier will lie between the value for the class
A circuit rp/2 and that for the class B circuit rp. In all cases, however,
recourse should be had to the composite static characteristics, and then
R;;4 should be set equal to the reciprocal of the slope of these lines.
The situation for pent odes is different from that discussed above for
triodes and follows roughly the reasoning of Sec. 9-5. The optimum load
is that which yields the maximum power with low distortion. The opti-
mum load line is drawn through the point Q'so that it intersects the peak
SEC. 9-11] UNTUNED POWER AMPLIFIERS 187
composite grid-voltage curve in the neighborhood of the knee of the
curve. This is illustrated in Fig. 9-21 for the 6F6 tube.
9-11. Driver Stages for Push-pull Amplifiers. The driver may be
considered to comprise the circuit that supplies the two voltages of equal
magnitude but in phase opposition

E
to the grids of the push-pull power
amplifier. A variety of suitable cir-
cuits exist, the most direct of which
is illustrated in Fig. 9-24. This cir-
cuit consists of a simple amplifier
with a transformer in the output, the FIG. 9-24. A simple driver circuit for
a push-pull amplifier.
secondary of which is center-tapped.
The characteristics of this stage are determined by the grid driving-power
requirements of the power amplifier. If the push-pull circuit requires
substantially no driving power, then there are no serious requirements
imposed on the driver stage. If the driver stage is called upon to supply
power, and this would seldom exceed 15 per cent of the output of the
push-pull stage, although it is ordinarily less than this amount, the driver
stage must have a relatively low internal resistance if no distortion is to
be introduced into the grid circuit of the push-pull amplifier. It is
advisable in such cases that a step-down transformer be used to couple
the driver stage to the push-pull input in order to reduce the effective
resistance in the grid circuit.
If the power requirements are low, then anyone of a wide variety of
"paraphase " circuits may be used. A paraphase circuit is one which
provides two equal output potentials which are 180 deg apart in phase
from a single signal source.
Single-tube Paraphase Amplifier. A single-tube amplifier in which the
plate resistor is divided equally between the plate and cathode circuits
is the simplest form of paraphase amplifier. The circuit is given in
Fig. 9-25. The resistors Rl and Rk have the same value, whence the
amplitude of the voltage developed across each is the same, since the
same current flows through each. The polarity is opposite because the
cathode output is taken from the more positive end of Rk and the plate
output is taken from the less positive end of R l • An analysis of this
circuit will show that the gain of the stage is less than unity and is given
by the expression
K =_IL_ (9-41 )
IL+2
Two-tube Paraphase Amplifiers. In the two-tube paraphase amplifier,
one tube is used as a conventional amplifier, and a second tube is used as
a phase-1nverter amplifier. Figure 9-26 illustrates such a circuit. The
188 ELECTRON -TUBE CIRCUITS [CHAP. 9

resistors Rl and R2 comprise a voltage divider across the output of a

conventional amplifier, the ratio of the resistances being chosen so that
the gain from the anode of Tl to the anode of T2 is unity. Also, the
operating conditions of the tubes are carefully chosen to allow the curva-
ture of the characteristic of T2 to compensate for the curvature of Tl.
Thus the output potentials relative to ground are both slightly distorted

FIG. \)-25. A single-tube paraphase am- FIG. 9-26. A two-tube para phase am-
plifier. plifier.
to provide a comparatively undistorted potential difference between the
output terminals. This method is difficult to apply in practice because
the adjustments necessary to reduce distortion to a minimum are critical.
A second form of two-tube paraphase amplifier employs the differential
voltage between the outputs of two tubes as the input signal to the phase-
inverter section. This circuit, which is also referred to as the floating
paraphase amplifier, is illustrated in two versions in Fig. 9-27.
o

(aJ
FIG. 9-27. Two forms of floating para phase amplifiers.

In the circuit of Fig. 9-27 a tube Tl is an amplifier to increase the amp]i~

tude of the applied wave form. The cathode resistors R k , if not by-passed,
will provide some degeneration, which will help to reduce distortion. The
output from Tl is coupled through C 1 to Rl and R a, both of which have
 SEC. 9-11] UNTUNED POWER AMPLIFIERS 189
the same value as R 2 • The voltage which appears across R3 is applied
to the grid of T2. The output of T2 is passed through C2 and is applied
across R2 and R 3• Thus half the output of
both Tl and T2 appears across R 3• Since
these potentials are of opposite polarity, the
resultant voltage across Ra is the difference
between these two. The output of Tl is
larger than the output of T2, and in order
that this difference should be kept as small as
possible, pentodes are used, so as to take ad-
vantage of their high amplification.
The feature of the circuit of Fig. 9-27b is
that the difference between the output volt- FIG. 9-28. A cathode-coll-
ages is taken care of in the choice of the re- pled paraphase amplifier.
sistors Rl and R2 so that the output voltages
have the same amplitude. To do this requires that the following con-
ditions be satisfied,

(9-42)

where K is the gain of the stage.

The cathode-coupled paraphase amplifier was discussed in some detail
in Sec. 6-9. This circuit is used extensively to provide push-pull deflec-
tion potentials for the plates of a cathode-ray tube. It may also be used
as the driver of a push-pull amplifier. The circuit is redrawn in Fig. 9-28
for convenience.
REFERENCES
1. Nottingham, W. B., Proc. IRE, 29, 620 (1941).
2. Millman, J., and S. Seely, "Electronics," Chap. XX. McGraw-Hill Book Com-
pany, Inc., New York, 1941.
3. Thompson, B. J., Proc. IRE, 21,.591 (1933).

PROBLEMS
/ 9-1. A 6F6 tube is operated as a triode and supplies power to a 4,000-ohm load.
With Ebb = 300, Eee = -25 volts and with a peak signal of 25 volts, calculate the
following:

a. Output power.
b. Plate-circuit efficiency.
c. Per cent second-harmonic distortion.
d. Plate dissipation.

9-2. Repeat Prob. 9-1 when the load is transformer-coupled to the .tube.
190 ELECTRON-TUBE CIRCUITS [CHAP. 9

9-3. It is supposed that the plate dissipation at the operating point is kept
constant. Prove that for class A operation the plate load is made larger with
increasing values of Ebb and the plate efficiency increases.
9-4. A 6V6 is operated with Ebb = Eo2 = 250 volts and with E" = -12.5
volts. The grid signal is sinusoidal, with a peak value of 12.5 volts. Calculate
the following for a shunt-fed load of 5,000 ohms:

a. Output power.
b. Total distortion.
c. Plate dissipation.
d. Plate-circuit efficiency.

9-5. Repeat Prob. 9-4 if the load is 2,000 ohms; 8,000 ohms.
9-6. Two 6F6 tubes are connected as triodes and are operated in push-pull
class A from a 350-volt plate source, with a grid bias of -30 volts. A 30-volt
peak signal is used.

a. Draw the composite static characteristics.

b. From this, determine the plate-plate resistance for maximum output power.
c. Calculate the power output, third-harmonic distortion, and plate-circuit
efficiency.

9-7. Two 6F6 tubes are connected as pentodes and are operated in push-pull
class AB from a 350-volt plate source, with a grid bias of -25 volts. The screen
potentials are maintained at 250 volts.

a. Draw the composite static characteristics.

b. Plot the composite dynamic and the dynamic characteristic of each tube.
c. Calculate the output power, third-harmonic distortion, and plate-circuit
efficiency for a plate-plate resistance of 10,000 ohms. The peak grid signal
is 40 volts.

j,9-8. Two 2A3 triodes are operated in push-pull with Ebb = 300, Eo, = -60
volts.

a. Draw the composite static characteristic.

b. From this, determine the plate-plate resistance for maximum power output
c. Calculate the power output under these conditions.
d. Repeat c for Hand 2 times the optimum value.
e. Construct the paths of operation for the individual tubes.

9-9. A 6N7 zero-bias tube is connected as a class B push-pull audio amplifier

and is to furnish 10 watts into a dynamic speaker, the voice coil of which has a
resistance of 8 ohms. A 35: 1 step-down transformer is used. The plate supply
is 325 volts. Determine the following:

a. D-c plate current.

b. Grid driving voltage.
 UNTUNED POWER AMPLIFIERS 191
9-10. The typical operating characteristics of a 6L6 beam power tube when
used in push-pull class A are shown below. Values shown are for two-tube unless
otherwise specified.
Plate supply. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 270 volts
Screen supply. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 270 volts
Cathode resistor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 125 ohms
Zero-signal plate current. . . . . . . . . . . . . . . . . . . . . . . . . . . .. 134 ma
Maximum-signal plate current ........................ 145 rna
Zero-signal plate current. . . . . . . . . . . . . . . . . . . . . . . . . . . .. 11 rna
Maximum-signal screen current. . . . . . . . . . . . . . . . . . . . . .. 17 ma
Plate resistance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 23,500 ohms
Transconductance ................................... 5,700 j.lmhos
Effective load resistance (plate to plate) ................ 5,000 ohms
Maximum-signal power output. . . . . . . . . . . . . . . . . . . . . . .. 18.5 watts
The 6L6 tubes are to supply the 18.5 watts to the grids of a pair of 806 triodes
which are operating in class B push-pull. The required peak grid driving voltage
is 660 volts.
a. Calculate the turns ratio of the output transformer.
b. What is the peak a-c plate voltage on each tube?
c. What is the peak a-c grid-voltage swing on each tube?
d. Does grid current flow during any part of the input cycle?
e. Calculate the plate-circuit efficiency.
9-11. Verify Eq. (9-41).
9-12. Verify the conditions (9-42) imposed on the floating paraphase amplifier
for balanced output voltages. What conditions are imposed on R3?
9-13. It is suggested that the paraphase principle be combined with a push-pull
amplifier to yield push-pull operation without a separate driving source. The
push-pull amplifier feeds a dynamic speaker. Discuss the suggested operation
from the point of view of class of operation possible; of distortion.
 CHAPTER 10

TUNED VOLTAGE AMPLIFIERS

TUNED voltage amplifiers are used in those cases in which it is desired to

amplify a relatively narrow band of frequencies centered about some
designated mean or carrier frequency. Voltages whose frequencies lie
outside of this range are undesirable and should be rejected. The use of
tuned networks accomplishes this, as it is possible to adjust the tuned
network so that the impedance falls steeply to low values outside of the

(a) (6)

(c)
FIG. 10-1. The three basic tuned-amplifier circuits: (a) single-tuned, direct coupling;
(b) single-tuned, transformer coupling; (c) double-tuned.

desired frequency band, with the consequent reduction in amplifier gain

to negligibly low values. The resulting nonlinear distortion that is
produced in these amplifiers is very small, both because the stage is •
operated under class A conditions and because the tuned plate-circuit
impedance may be very low for any harmonic frequencies that might be
generated.
There are three basic tuned-amplifier circuits, and these are illustrated
in Fig. 10-1. Pent odes are ordinarily used in such amplifiers, and the
circuits are drawn showing such tubes. In two of these types, a single
resonant circuit is used, which may be included directly in the plate
circuit (direct-coupled) or which may be inductively coupled to the plate
circuit (transformer-coupled). In the third type, a double-tuned band-
192
SEc.10-1J TUNED VOLT AGE AMPLIFIERS 193
pass arrangement is used, both the primary and secondary circuits being
tuned. An analysis of the operation of each of these amplifier circuits
will be given.
10-1. Single-tuned Direct-coupled Amplifier. The equivalent cir-
cuit of a typical single-tuned direct-coupled stage is given in Fig. 10-2.

G P

FIG. 10-2. The equivalpnt circuit of a single-tuned direct-coupled class A amplifier.

Included in this diagram are the output tube capacitances C [seeEq. (4-20)]
and the input and wiring capacitances to the following stage. The
coupling capacitor may be neglected, as its capacitance is presumed to be
large.
In accordance with the discussion of Sec. 4-7, the gain of the amplifier
can be written directly as
(10-1)

where Z is the total load impedance. This impedance has the form

1 1 1 1
-=-+-+-
Z rp Rg Zt
(10-2)

where Zt is the impedance of the antiresonant circuit, and comprises the

inductor L and the sum of the various capacitances Ct = C2 + Cg + C'
where C 2 is ddefined as in Eq. (4-20). ( \ ' {! . .t ~ 1'5
The impe ance Zt has the form C,. l.tll of'!
- j :C (Rl, + jwL)
Zt = (10-3)
R l_ + j (WL - w~)
which may be written in the form

(10-4)
194 ELECTRON-TUBE CI.RCUITS [CHAP. 10

By writing

(10-5)

the impedance function becomes

Z -
(1 .1"'0)
R L Q2 -J--
Q '"
t -. [ 1 + (:0 _:0)]
jQ

RLQ2 [ (1+ 0) - j ~ ]
(10-6)
= (1 + 0) + jQo(2 + 0)
At resonance", = "'0, and 0 = O. Then

(10-7)

Since Q for the circuit used is usually high, that is, Q 2:: 10, then with
good approximation
(10-8)
This result shows that the shunt impedance Ro of the antiresonant circuit
is essentially resistive, for circuits with Q > 10, and at the resonant
frequency.
By combining Eq. (10-8) with Eq. (10-2), the gain at resonance becomes

(10-9)

This expression may be written in the form

K r.. = -gm"'oLQe (10-10)
where Qe, the effective Q of the amplifier, is

(10-11)

This is the equivalent Q of the resonance curve of the tuned amplifier and
is the Q of the actual resonant circuit as modified by the shunting resist-
ances Rg and rp.
SEC. 10-1] TUNED VOLT AGE AMPLIFIERS 195
To find the gain of the amplifier when the input frequency and the
resonant frequency of the tuned circuit are slightly different from each
other, it is supposed that w == Wo, whence 5 is small. Then, from Eq.
(10-6), it follows that

(10-12)

The corresponding value of the gain, given by Eq. (10-9), is

K= (10-13)

The gain ratio KIKres is then

K 1
(10-14)
K ro• = 1 + j25Q.
from which the amplitude ratio is

(10-15)
I/feal = VI +\25Qe)2
A plot of these results is given in Fig. 10-3. This is essentially the
"universal resonance" curve. Note from Eq. (10-15) that when

25Qe = 1
then

\:.,\= ~
But since the band width of the circuit is the frequency width between
the 3-db power points,
B - 2(f
-
') - 2(f - fo)fo
3db - JO - io
or
B = 25fo = fo (10-16)
Q.
In order that the voltage gain at resonance be large, the resonant
impedance of the tuned circuit, and the grid resistor R g , must be large
compared with Tp. Note, however, from Eqs. (10-5) and (10-8) that an
increase of Zt at resonance by increasing the Lie ratio is accompanied by
a decreased Qe, with a corresponding increase of band width or decreased
frequency selectivity. This is an undesirable condition if a narrow band
196 ELECTRON -TUBE CIRCUITS [CHAP. 10

width is desired, but it is an important consideration in wide-band tuned

amplifiers. Note also that if the circuit Q of the tuned circuit is increased
at fixed values of Wo and Lie ratio, then the effective Qe of the amplifier
10

r---r-~~~---4---+-/~~
[\\
8
1/. "
II 1 Q=oo;
1----\---t---I---f---I--0.7 75 -;\.~----l-~-..--_-~-lF:::~Q='=,...J~~j~:=t===l

f----I---l---j.--+-/-f--I-/~0.6
s:: so-gl7";~\/+----+---+------i--1 o/"

/ ~
I----\---t---I---II-+-t-.~::: 0.5 12~-F-+----\',\:+--+---+---+---I
~=!q ~ II _
t---t---I-::---t---I:AL'-j-'--l- ] 0.4 0 ~ ~
Q=ooJ// ;'1,~' ~lj=IO
0.3 2S~"" ~
.......--___ O
----- [....0'/
;'
/ ..£:~
"
0.
IS
.
..... ~Q=oo:~
- - : : ,...- Q=oo 4" 0.2 50 l'r----t---I----+----==""-l"'"-.:::.::'""-...,,-l

-- - L.-J. -.;:'/
1:=~:::::====F_:::::rF...:..::.-r---t-O.1 75 -.+---1----+---1-----1---1
._ --r
--- Q=!OI' o
3.0 2.5 2.0 1.5 1.0 0.5 o 0.5 1.0 1.5 2.0 2.5 3.0
oQ
FIG. 10-3. Plots showing the amplitude K/ K res and phase of the output of the single-
tuned direct-coupled amplifier. (From Terman, F. E., Radio Engineering, 3d ed.,
McGraw-Hill Book Company, Inc., New York, 1947.)

is increased, with a corresponding increase of frequency selectivity or

decreased band width.
G P
t-"---r--,

K
FIG. 10-4. The equivalent circuit of the single-tuned transformer-coupled voltage
amplifier.

10-2. Single-tuned Transformer-coupled Amplifier. The general

behavior of the single-tuned transformer-coupled amplifier is quite
similar to that of the direct-coupled circuit. To examine the operation
of the circuit in some detail, refer to Fig. 10-4, which gives the equivalent
circuit of this amplifier. An approximate expression for the voltage gain
of the amplifier is readily obtained. If it is noted that rp is usually large
SEc.10-2J TUNED VOLTAGE AMPLIFIEHS 197
compared with R1 and L 1, then the voltage induced in the secondary of
the transformer is given by
(10-17)
The output voltage, which is the potential across the capacitor Cz, is
then seen to be
Eo = }wM(gmEg) _.1_ (10-18)
Rz +} (wLz __ 1_) + (wM)
wC z rp
2
JwC 2

where (wM)2/rp is the reflected impedance of the primary into the second-
ary circuit. The expression for the gain then becomes

(10-19)

The corresponding expression for the voltage gain at resonance is

(10-20)

This result may also be expressed in terms of Qz = WoL2/R 2 = 1/w oC2R 2

and is
WOMQ2
K reB = gm 1 + (w~M2/rpR2) (10-21)
,vhich may be wrItten as
Kr"" = gmwoMQ. (10-22)

(10-23)

A comparison of this expression with Eq. (10-10) shows that trans-

former coupling modifies the amplification by the ratio M fL. This
provides a means for controlling the gain of the stage and still retaining
the high Q required for selectivity. It might appear that there are no
limits on the gain, and that it continues to increase with increasing values
of M. This is not so, owing to the appearance of M in the denominator
of Eq. (10-20). An optimum value of gain exists, and this occurs when M
has the value required to make aK/ aM = O. This yields, for the optimum
value of M,
198 ELECTRON -TUBE CIRCUITS [CHAP. 10

or
M OIlt = -vr;JI;. (10-24)
Wo

Equation (10-21) becomes

~Q2
Kr ... OJ>t = gm 2

(10-25)

To find the band width of the amplifier, consider the general expression
for the gain given by Eq. (10-19). By writing, as before,
1
Wo = y'LC

o = ~-1 (10-26)
Wo

WoL2 1 1 IL2
Q2 = R2 = woC2R 2 = R2 '\J C2

and noting that in the neighborhood of resonance

wL 1 = {L; (~ _ wo)
*'
__
2
wC 2 VC; Wo W

== 20 = 20R2Q2 (10-27)
Eq. (10-19) becomes
K = p.M/C 2 (10-28)
r;R2 (1 + j R2~ 28 '\J('};)
(L;\ + w2M2
The gain ratio [Eqs. (10-28) to (10-20)] then becomes
K 1
(10-29)
Kres = 1 + j28Q.
which has the same form as for the direct-coupled connection [Eq. (10-14)].
The band width of this amplifier is, following the same reasoning as that
which led to Eq. (10-16),
B = fo (10-30)
Q.
The optimum value of M is not of much importance, owing to practical
limitations. This follows from Eq. (10-24), which shows that for pen-
todes, with the corresponding large values of rp , the value of M would be
large. In fact, to achieve these values of M, the distributed capacitances
of the windings may become excessive, and the self-resonant frequency
SEC. 10-3] TUNED VOLTAGE AMPLIFIERS 199
may be so low as to make the coils useless. Owing to this, the mutual
inductance is usually chosen far below the optimum value in the pentode
amplifier.
10-3. The Double-tuned Ampli- 6
fier. Both the single-tuned direct- +
+
coupled amplifier and the double- Ef1
tuned amplifier are extensively used --
in radar, television, and communi-
la
cation receivers. For the i-f ampli- FIG. 10-5. The equivalent circuit of a
fiers of both a-m and f-m types, the double-tuned amplifier.
double-tuned amplifier is commonly
used. This is so because such an amplifier can provide substantially
constant amplification over a band of frequencies and the gain falls more
sharply outside of this band than does the single-tuned stage.
To examine the operation of the circuit, refer to the equivalent circuit
of this amplifier. This circuit can be further simplified by applying
TMvenin's theorem to the portion of the circuit to the left of the points
aa. The equivalent generator has the potential

E = /lEg (1jjwC 1) (10-31)

Tp + (ljjwC 1)

where C1 = C~ + C~. But since Tp > ljwC 1 for the pentode, then with
good approximation

E == /lEg = gmEg (10-32)

jWTpC 1 jwC 1
The internal impedance of the equivalent generator will have the value

which is, to the same approximation as above,

Z == _1_ (10-33)
jwC 1
Then the equivalent circuit of
FIG. 10-6. The equivalent series form Fig. 10-5 reduces to the form of
of Fig. 10-5. Fig. 10-6.
This circuit is analyzed by the
IItandard methods of network analysis. Accordingly, if one writes

E = ZllI l + Zld2 } (10-34)

o= Z 12 I 1 + Z2d2
200 ELECTRON-TUBE CIRCUITS [CHAP. 10

then the current in the secondary is

(10-35)

where

Zll = Rl +j (WLI - W~)

Z12 = jwM (10-36)

Z22 = R2 +j (WL2 - w~)

The gain of the amplifier becomes

(10-37)

But it must be noted that both circuits are tuned to the same resonant
frequency. Thus
1 1
wo=~-~

Also define

(10-38)

and write

Similarly
(10-39)
and
Z12 = jwM
The expression for the gain lEq. (10-37) 1then becomes for frequencies near
resonance
SEC. 10-3] TUNED VOLTAGE AMPLIFIERS 201
or finally
K _ -jagm QIQ2 ~ (10-40)
- 1 + a 2 + j20(Ql + Q2) - 402QIQ2
The gain at resonance is obtained by setting 0 = 0 in this expression.
There results
(10-41)

The gain ratio at frequencies slightly different from resonance is given by

K 1
Kres - (1 _ 402QIQ2)
1 + a2
+ .20(Ql + Q2)
J 1+a 2
(10-42)

The exact shape of the response curve of the double-tuned system

depends upon the parameter a, or, correspondingly, on k, the coefficient
of coupling between the primary and secondary coils. The resonant gain
Ixl
Units Ofgm QI Q2Vii;R2
.----r--r--...-..-,--JO.5-r-~,-"7"""i,.._____r-__,_-_,

{)
o 0.5 1.0 1.5 2.0 2.5
oQ
FIG. 10-7. The response characteristics of a double-tuned amplifier for various
values of coupling.

is a maximum when a = 1, as may be verified by examining dK/da = O.

Moreover, if the primary and secondary Q values are the same and a = 1,
the response curve has the maximum possible flatness in the vicinity of
resonance. This is the condition for critical coupling kc• A value of a
greater than unity (over coupling) results in double peaks, whereas a value
of a less than unity (undercoupling) causes the response to be rounded
on the top. If the circuit is considerably undercoupled, then the gain
may be less than that at resonance. The situation discussed is illustrated
graphically in Fig. 10-7.
An analytic expression for the positions of the peaks in the overcoupled
circuit is readily possible. These are given, of course, by the values of
202 ELECTRON -TUBE CIRCUITS [CHAP. 10

frequency at which Eq. (10-40) is a maximum. To find these values, it

is noted that the gain is a maximum without regard to the phase. Thus
the square of the absolute value of Eq. (10-40) is differentiated with
respect to 0 and maximized. The results are

1 K 12 = «(JmQIQ2 ...jR 1R 2)2 (1 + a2 _ 402QIQ~22 + [20(Ql + Q2)P

(10-43)
and the derivative 01 K 1/002
= 0 yields
1 +a 2 _ 402QIQ2 = (Ql + Q2)2
2QIQ2
from which
(10-44)

Frequently the circuits are designed with Ql = Q2. Even if this condi-
tion is not true, ordinarily Ql does not differ too markedly from Q2, and
it is possible to assume that
VQIQ2 == Ql + Q2
2
Subject to this approximation, Eq. (10-44) becomes

o = +! Ik2 __1_ (10-45)

- 2'\j QIQ2
which becomes, in the manner of representation of Fig. 10-7, simply
(10-46)
The value of the gain at either peak Kr= is obtained by combining
Eq. (10-46) with (10-40). The result is

Kmax = -j(gmQIQ2 yR 1R 2) a
2(1 + j ya 2
- 1)
or
(10-47)
This shows that for the overcoupled case the maximum gain is the same
as that for critical coupling a = 1, and at resonance 0 = O.
The gain at the dip, or the frequency Wo, can be found readily by setting
o = 0 in Eq. (10-40). The result is
K,nin = -j(gmQIQ2 yR 1R 2) 1. ~ a 2
or
(10-48)
SEC. 10-31 TUNED VOLT AGE AMPLIFIERS 203
For the case where the primary and secondary Q values are not the
same, the flattest selectivity curve may be shown to occur when

(10-49)

although the mid-band gain is not a maximum under these conditions.

When Eq. (10-49) is satisfied, the circuit is said to be transitionally
coupled. The transitional value of coupling coefficient is, by Eqs. (10-49)
and (10-38),

(10-50)

For a coupling coefficient larger than this value, the selectivity curve
divides into two peaks. For values less than this, the curve has a single
peak.
It follows from Eqs. (10-46) and (10-47) that increased coupling
increases the frequency separation of the two peaks, but does not change
their amplitudes. If the coupling is very large, then the approximation
made in Eq. (10-40) is no longer valid. The effect of the factor w/wo
in this equation is to increase the lower frequency maximum and decrease
the higher frequency maximum.
It is interesting to compare the gain of the double-tuned circuit with a
single-tuned circuit having the same Q. The gain of the two circuits at
resonance are given by Eqs. (10-10), and (10-41) for optimum value of a,
and are

Single-tuned direct-coupled: K res = -gmwoLQe

Double tuned (with a = 1 and identical coils):

It is observed that for critical coupling the gains at resonance of the two
amplifiers are identical if it is assumed that the tuning capacitance of the
individual tuned circuit in the double-tuned circuit is one-half the tuning
capacitance for the single-tuned case.
Despite the fact that the response characteristics are optimum under
critical coupling conditions, the transformers in narrow-band double-
tuned amplifiers are usually under coupled slightly. This is done in
order that the frequency alignment of the tuned circuits may be made
easier, since, with undercoupled stages, each stage can be adjusted sepa-
rately to give maximum response at the specified frequency. If over-
coupled circuits exist, owing to the interactions between coils and the
204 ELECTRON-TUBE CIRCUITS [CHAP. 10

resulting double peak, this alignment is more critical. The critical

coupling case is likewise difficult to align.
The band width of the amplifier, under optimum conditions a = 1 and
with equal primary and secondary values of Q, is readily calculated.
Under these conditions Eq. (10-42) becomes
K 1
(10-51)
K r08 - (1 - 2ij2Q2) + j20Q
and the magnitude becomes

It.. I = VI : 40 4Q4
(10-52)

Since, by definition, the band width gives a measure of the frequency

spread over which the gain remains within 3 db of the maximum value,
then
40 4Q4 = 1

from which it follows that the band width is

B = 20/0 = v'2B (10-53)

A comparison of this result with Eq. (10-16) for the single-tuned stage
shows that the 3-db band width of the double-tuned circuit is 1.414 time,;
that of the single-tuned stage.
10-4. Cascaded Tuned Amplifiers. It is frequently necessary to
incorporate more than one stage of amplification in a given amplifier.
Although such a practice provides a higher gain, this higher gain is
accompanied by a narrower band width than for the single stage. Ana-
lytic expressions for the effect of cascading identical amplifiers are readily
possible.
Consider first n single-tuned stages in cascade. The gain of such an
n-stage amplifier becomes, from Eq. (10-15),

I%.08 r = [1 + (20Qe)2]~
(10-54)

To find tne corresponding band width, it is noted that

n
[1 + (20Qe) 2]2 = V2
from which
1
1 + (20Qe)2 = 2;;
so that
20Qe = ~'2~ -- 1
SEC. 10-4J TUNED VOLT AGE AMPLIFIERS 20~

But the band width is given by

. .~
"\12; - 1
BIn = 20jo = Qe/jo (10-55)

This may be expressed in terms of the band width of the single stage B l , in
the form

BIn = B 1 ~ 2~ - 1 (10-56)

Table 10-1 gives the band-width reduction function ~2~

- 1. It is
seen, for example, that two stages in cascade have a band width that is
TABLE 10-1
THE SINGLE-TUNED-AMPLIFIER BAND-WIDTH REDUCTION FACTOR

n ~2; - 1
1 1.0
2 0.643
3 0.510
4 0435
5 0.387
6 0.350
7 0.323
8 0.301
only 0.64 times that of a single stage. To maintain a given band width,
it is accordingly necessary that the Q of the individual stages be decreased
as the number of stages is increased.
A corresponding expression is possible for the double-tuned amplifier.
For such an n-stage amplifier, with critical coupling a = 1 and equal
primary and secondary values of Q, the relative gain becomes, from Eq.
(10-52),
ItJn =
(1
1 ~
+ 40 4Q4)2
It follows from this that

oQ =
J2~- 4 --1
The band width of the n-stage amplifier then has the form
- l

B 2n = 20jo = 2 J 2n - 1 jo
-4-"Q

which may be written in terms of one-stage band width as

}~
R 2n = B 2 \/2 n - 1 ( l0-57)
~06 ELECTRON-TUBE CIRCUITS [CHAP. 11

The band-width reduction factor is tabulated in Table 10-2. For a

two-stage double-tuned amplifier with the coils critically coupled, the
band width is 0.802 times that of the single-stage amplifier. Note that
TABLE 10-2
THE DOUBLE-TUNED-AMPLIFIER BAND-WIDTH REDUCTION FACTOR
FOR a = 1
~4r--r--
n "2:;; - 1
1 1.00
2 0.802
3 0.713
4 0.659
5 0.622
6 0.592
7 0.568
8 0.548
this reduction is considerably less than the corresponding reduction of the
two-stage single-tuned amplifier. This arises from the fact that the
amplification or selectivity curve of the double-tuned amplifier has
steeper sides than that of the single-tuned circuit and with successive
stages drops away less rapidly than for the single-tuned case. In par-
ticular, an ideal amplifier with a rectangular response curve would show
no band-width reduction with the addition of successive stages.
10-5. Gain-Band Width Product. It is of interest to tabulate the gain
at resonance of the three amplifier circuits that have been studied. These
follow:
Single-tuned direct-coupled: K re• = - gmwoLQe
Single-tuned transformer-coupled: K res = gmwoMQc
Double-tuned: Kre. = -jO.5gm QIQ2 yR 1R2
These expressions may be interpreted as showing that the gain in each
case has the form
(10-58)
where gm is the transconductance of the tube and Iz I is the effective
impedance of the load. Moreover, the foregoing analyses for these
amplifiers show that the band width in each case varies inversely with the
effective Q of the tuned circuit. Clearly, therefore, the higher gains are
accompanied by a decreasing band width. Consequently a specification
of the impedance or the gain at resonance of the tuned circuit is not in
itself a very good measure of the excellence of the amplifier.
It might appear that the product of the gain and the band width would
be a significant quantity. Such is actually the case, and this product
may be considered to represent the area under a perfect band-pass ampli-
fier having a constant gain over the full band width. Consider first the
SEC. 10-6] TUNED VOLT AGE AMPLIFIERS 207
gain-band width product of the single-tuned direct-coupled amplifier.
Combining Eq. (10-10) with (10-16), there results
f w 2L
K.esB = -gmwoLQe Q: = -g 2~
which may be written in the form

Kre. B = - i:c = (10-59)

Notice that this product is independent of the resonant frequency or of

the band width and depends only on the transconductance of the tube
and on the total tuning capacitance. If in an amplifier the tuning capaci-
tance is determined by the tube capacitance, then the gain-band width
product is a measure of the merit of the tube. What is desired is a large
transconductance in proportion to the input plus output electrode
capacitances. The 6AK5 and the 6AC7 tubes are both highly satisfac-
tory in this respect, the 6AK5 being slightly better than the 6AC7. An
average 6AK5 has a gain-band width product of approximately 55 mega-
cycles/sec, an average 6AC7 having a corresponding value of 50 mega-
cycles/sec when allowance is made for socket and wiring capacitance.
The gain-band width product of the double-tuned amplifier is found
by combining Eq. (10-41) with (10-53) and is

KreaB = - j g2mQIQ2 yR 1R 2 y'2 _ ~

v QIQ2
-j ~2 gm yQIQ 2R 1R 2 ;;

. gm 1
- J- --::=----== (10-60)
211" y2 yC1C2
This expression shows that the gain-band width product of the double-
tuned amplifier is y2 as great as that for single-tuned circuits. That is,
by splitting the tube input and output capacitances by the use of the
double-tuned circuit, there is an increase in the gain-band width product.
10-6. Stagger TuningJ If it is desired to build a wide-band high-
gain amplifier, one procedure is to use either single-tuned or double-tuned
circuits which have been heavily loaded so as to increase the band width.
The gain per stage is correspondingly reduced, by virtue of the constant
gain-band width product. The use of a cascade chain will provide for
the desired gain. For example, a particular amplifier comprising nine
cascaded single-tuned stages, each of 6 megacycles/sec band width, has
an over-all band width of 1.7 megacycles. A nine-cascaded double-tuned
amplifier, each also of 6 megacycles band width, yields an over-all band
208 ELECTRON-TUBE CIRCUITS [CHAP. ]0

width of 3.2 megacycles. However, double-tuned stage~ are difficult to

align and also are more sensitive to variations in tube capacitance and coil
inductance than that of the single-tuned circuits.
A means is available for achieving large band width and other charac-
teristics of double-tuned circuits by using single-tuned circuits. This
consists in taking two single-tuned circuits of a certain band width and
staggering their resonance peaks by an amount equal to their band width.
The result will have a band width that is V2 times as great as that of

1,;
!:
0
a.
V)

'"'"
<lJ
>
:..;:"
~
'"
a::

-5 5

FIG. 10-8. The adjustments of frequency of a stagger-tuned pair.

each of the individual circuits; the over-all selectivity function will be

identical with the corresponding single-stage double-tuned system. The
general situation is illustrated in Fig. 10-8.
An analytic expression is readily obtained for the over-all characteristics
of the stagger-tuned pair. If it is noted that the general selectivity
function of the single-tuned direct-coupled circuit is, from Eq. (10-14),
K 1 1
Kes = 1 + j20Qe - 1 + jx
and the band width between the 3-db points is, from Eq. (10-16),
/0
B = 20/0 = Qe
then the corresponding selectivity functions of the two circuits are,
respectively,

and
(:Jl 1
= 1+j(x+1)

(:')2 1
= 1 + j(x - 1)

2 - x2 + j2x
 SEC. 10-6J TUNED VOLTAGE AMPLIFIERS 209
which becomes, on considering the magnitude of the resulting function,

I :., 111 : .. 12 = V4 \ X4 = ~ VI +140~Q4 (10-61)

where 00 is the value of 0 referred to the new frequency Wo, and where
Q is the value of Q6 for each circuit referred to woo A comparison of this
expression with Eq. (10-52) for the double-tuned circuit shows that the
forms of the variation are identical.
A calculation shows that the gain-band width product of the stagger
pair is 1.10 times that of two synchronously tuned stages. However, the
stagger pair is smaller than two double tuned stages by the factor
1.60.
The principle of stagger tuning can be extended, and staggered triples
are in fairly common use in radar receivers. In this case a centered
single-tuned circuit of relative band width 2 and relative resonant gain
Yz is combined with two single-tuned circuits, each of relative band width
1 and relative gain 1, staggered such that their resonance peaks are
± v3/2 from the band center. The resultant sensitivity function is of
the form

I:J=~ (10-62)

This sensitivity function has the same form as that for a triple-tuned
circuit.
The advantage of stagger-tuned amplifiers, and the principle may be
extended to n-uples, lies in the fact that simple single-tuned circuits are
used throughout. This makes the alignment of the stages very easy, as
they are independent of each other.
REFERENCES
1. Wallman, H., M.I.T. Radiation Lab. Rept. 524, Feb. 23, 1944.
Wallman, H., Electronics, 21, 100 (May, 1948).
2. Ai' a general reference, consult Sturley, K. R., "Radio Receiver Design," Part I,
John Wiley & Sons, Inc., New York, 1943.
PROBLEMS
0/10.1. A 6SJ7 pentode is used in a certain class A r-f amplifier, with Ebb = 250
volts, Ec,2 = 100 volts, and E"l = -3 volts. At these conditions the tube
parameters are approximately Om = 1,600 JLmhos, Tp = 1.2 X 10 6 ohms. A
single-tuned load consists of a 1-mh coil in parallel with a 100-JLJLf capacitor.
The resonant Q of the load is 200.
a. Determine the voltage gain of the stage at the resonant frequency.
b. Determine the voltage gain of the stage 10 kc above and below resonance.
10.2. In a single-tuned direct-coupled amplifier stage using a 6SJ7 tube that is
'tuned to 1,100 kc, it is found that the band width is 18 kc. Determine the Q of
the circuit.
 210 ELECTRON -TUBE CIRCUITS [CHAP. 10

/ 10-3. It is discussed in the text that the response of the single-tuned direct-
coupled or transformer-coupled amplifier for small deviation 8 and high Q is
given by either Eq. (10-14) or Eq. (10-29).

a. Calculate the error in each case for Q = 3, 8 very small.

b. Repeat for Q large, 8 = O.I.

10-4. A single-tuned circuit employing a 6SK7 tube feeds a diode detector.

Oiode defector
equivg/enf /04fdSO Ir

The equivalent output circuit is illustrated. The tank is tuned to 10 = 1 mega-

cycle, L = 0.5 mh, Q = 60. Assume for the 6SK7 that rp = 10 6, gm = 2,000
IlmhoS. Determine
a. Gain at resonance.
b. Gain at resonance if detector is removed.
c. Band width with and without the detector circuit.
10-5. A direct-coupled single-tuned amplifier has a band width of 150 kc and a
l"esonant shunt impedance of 50k • What must be the value of the shunting
resistance across the tank if the gain is to be constant within 10 per cent over the
150-kc band?
10-6. A direct-coupled single-tuned amplifier has a band width of 50 kc, when
C = 25 Illlf. Calculate the band width if C is increased to 100 Illlf, and
a. If the resonant frequency is kept constant.
b. If L is maintained constant.
/ 10-7. A single-tuned direct-coupled stage has a Q = 100 when tuned to 800 kc.
Two equal signals are fed to the grid, one of which is 50 cps off resonance and the
'other of which is 5,000 cps off resonance. What will the amplitude ratio be in
the output of the amplifier?
10-8. A direct-coupled single-tuned amplifier is to have a band width of
200 kc at 4.7 megacycles. If the total capacitance is 251l1lf,
a. Calculate the maximum impedance and the value of L.
b. Over what frequency band is the total phase shift through this amplifier
less than 30 deg?
10-9. Repeat Prob. 10-8 for the transformer-coupled single-tuned amplifier,
for optimum coupling.
/ 10-10. A double-tuned circuit with C = 12 J.l.J.I.f has a band width of 1 megacycle
at 10 megacycles for the critically coupled stage. Determine the value of L
and Q if both primary and secondary windings are identical.
10-11. A single-stage double-tuned amplifier using a 6SK7 tube is critically
 TUNED VOLTAGE AMPLIFIERS 211
coupled. It operates at 455 kc and has a band width of 12 kc. The total primary
C and total secondary C are each 26 Illl£. The coils and loading are the same.
a. Calculate the values of shunt resistance of each circuit, Land M.
b. Calculate the mid-frequency gain. Choose Om = 1,500 Ilmhos.

10-12. A 6SK7 double-tuned circuit comprises two identical 200-llh coils, with
Q = 80, which are tuned to 500 kc.
a. Calculate the critical coefficient of coupling.
b. Calculate and plot the gain of the stage as the mutual inductance is varied
from zero to twice the critical value.
10-13. Determine the proper design for the winding of an i-f transformer with
L1 = L2 and each winding tuned to resonance by a capacitance of 100 IlI.Lf. The
secondary voltage is not to fall below 0.88 of the peak value in a 10-kc band,
centered at 465 kc. Find k, L 1, L 2, Q1, Q2 and the secondary voltage, with 1 volt,
465 kc to the primary. Assume critical coupling.
10-14. A 6SJ7 is used in a double-tuned circuit which feeds a diode detector and
automatic-volume-control circuit. The significant portion of the circuit is shown.

I
'JlT I Diode defeefor and alit:
I equivalent load 200k
I
I
Ql=IOO Qz=80
i-f=456 kc

The rms potential across the secondary feeding the detector-automatic-volume-

control circuits must be 10 volts.
a. What is the coefficient of coupling?
b. What is the value of eo to give the required output?
10-15. The i-f amplifier section in a radar receiver consists of four identical
stages, each of the form illustrated. The maximum over-all gain of the four

stages is 10,000, and the gain at 28.5 and 31.5 megacycles is 7,070. 6AC7 tubes
are used, with Om = 9,000,umhos, Ci = 11 J..LJ..Lf, Co = 5 1lJ..Lf.
a. Calculate the value of the plate-load resistor.
b. Calculate the wiring capacitance.
c. Calculate the value of the inductance of the coil.
212 ELECTRON-TUBE CIRCUITS [CHAP. 10

10-16. A six-stage single-tuned i-f amplifier using 6AC7 tubes has a maximum
over-all gain of 4,100 and an over-all band width of 6.0 megacycles. If the over-
all gain had to be obtained from four stages instead of six stages, what would
have been the band width?
10-17. A six-stage single-tuned amplifier using 6AC7 tubes has a maximum
over-all gain of 530,000 and an over-all band width of 2 megacycles.
a. If it is found that the over-all band width need not be greater than 1.5
megacycles, what would be the corresponding over-all gain by an appro-
priate change in the value of the effective shunt resistance of each stage?
b. If the original over-all gain of 530,000 had been obtained from four stag€~
instead of six, calculate the over-all band width that would result.
10-18. Suppose that three identical stages having the characteristics of Prob.
10-2 are connected in cascade. Calculate and plot a curve of relative gain in
decibels vs. frequency. Carry out the calculations to frequencies at which the
gain is down at least 75 db below the optimum value.
10-19. A three-stage direct-coupled single-tuned amplifier is used in a broad-
cast receiver. A three-ganged 165-JLttf capacitor is used to tune the receiver
over the range from 550 to 1,650 kc. The loading is chosen to give a minimum
band width of 10 kc.
a. Determine the variation of band width as the receiver is tuned over the
entire range, assuming that Q remains constant.
b. Repeat for the ~ase where C is fixed at 100 JLJLf and L is varied.
10-20. Refer to a tube manual, and prepare a table of the merit of the following
tubes: 6AB7, 6AC7, 6AG7, 6AK5, 6C5, 6K7, 6L6, 6SF5, 6SJ7, 6SK7, 6V6, 6Y6.
10-21. A two-stage direct-coupled single-tuned amplifier using 6AC7 tubes
operates at 60 megacycles, and is to have a 2-megacycle band width.
a. What gain is possible if both stages are tuned to the same frequency"?
Assume that the shunt capacitance is 25 JLJLf.
b. If the stages are to be stagger-tuned to be critically fiat, what gain is
possible?
10-22. Show that, by choosing the three single-tuned stages in the manner
discussed in the text to yield a staggered triple, the relative response function has
the form given by Eq. (10-62). Sketch the individual response characteristics
and that of the resultant staggered triple.
 CHAPTER 11

TUNED POWER AMPLIFIERS

T HE basic circuit of a tuned power amplifier is substantially that of the

single-tuned direct-coupled type discussed in Sec. 10-1. The essential
differences are in the magnitude of the grid-bias supply voltage E cc , the
corresponding value of the grid input signal Cg, and the amount of power
involved. A schematic diagram of a tuned power amplifier is given in
Fig. 11-1.
Owing to the negative bias on the tube, which is adjusted approximately
to plate-current cutoff)n the class B amplifier and which is adjusted
beyond plate-current cutoff in the class C amplifier, harmonic currents
are generated in the plate which
are comparable in amplitude with
the fundamental component. L
However, if the Q of the tuned
plate circuit has a value of 10 or
more, the impedance of the tank
]
circuit to the second or higher
harmonics will be very low. As
it result, the higher harmonic po- FIG. 11-1. Schematic diagram of a tuned
power amplifier.
tentials across the tank will be
very small compared with the fundamental potential. That is, the effect
of the harmonic generation in the tube plate current is largely suppressed
by the tuned plate load.
But the requirement that the Q of the tank circuit must be high in order
to suppress harmonics in the output imposes a limitation on the fre-
quency-response characteristics of the amplifier, since the gain is constant
over a very narrow band of frequencies. Consequently such amplifiers
.are confined in their operation to narrow frequency bands. In fact, the
, class B amplifier may be used to amplify a narrow band of frequencies of
differing amplitudes, whereas the class C amplifier is confined to a narrow
. band of frequencies of constant amplitudes. Despite these severe restric-
tions, both classes of amplifier are extensively used in restricted applica-
tions, the class B amplifier to amplify an a-m r-f carrier wave, the class
C amplifier as a frequency multiplier or as a source for the production of
an a-ill carrier wave.
213
214 ELECTRON-TUBE CIRCUITS (CHAP. 11

11-1. Properties of the Tank Circuit. The tuned plate load in the
diagram of Fig. 11-1 is drawn as a simple parallel resonant circuit.
Ordinarily the load is coupled inductively to the plate tank, and a more
typical coupling network is that shown in Fig. 11-2. The capacitor C2
is assumed to be so adjusted that
1/2n- VL2C2, the resonant frequency
of the secondary circuit, is equal to
the operating frequency of the ampli-
fier. Because of the resonance in
FIG. 11-2. A typical tuned-amplifier the secondary circuit, only a resistIVe
tank circuit.
component R~ = (WM)2/(R'{ + R 2)
is reflected into the primary of the tuned circuit~-7he equivalent circuit
then becomes that shown in Fig.
11-3.
If the characteristics of the tank
circuit were ideal, the impedance at
resonance would be resistive and
equal to the shunt resistance Ro of FIG. 11-3. The equivalent circuits of
. network . Th'
the resu Itmg e Imped- Fig. 11-2.
ance would be zero at any ~f the harmonic frequencies. That is, the
impedance would be

Z(Wo) = Ro (11-1)
Z(nwo) = 0 n = 2,3,4, ... }

These ideal conditions do not prevail in practice, although it is possible

to achieve relatively low impedance for Z(nwo). To examinE' this, refer
to Eq. (10-6) for the impedance function of the simple tuned circuit.

_ 2 (1 + 0) - j(I/Q)
Z - RLQ (1 + 0) + jQo(2 + 0) (11-2)

At resonance w = Wo, and 0 = O. Equation (11-2) reduces to

Note, however, that, if Q = 10, then

Z(wo) = RLQ2 X 1.005/-5.7°
which shows that the impedance of the tank circuit is essentially resistive
and is given by
(11-3)
SEC. 11-1] TUNED POWER AMPLIFIERS 215
I

Under these conditions it follows that :' i ,,/ '

Ll {Lt (11-4)
Ro = RLQ2 = WoLIQ = RLC = Q '\)Ct
1
Now consider the situation at the second-harmonic frequency. When
w = 2wo, 0 = 1 and Eq. (11-2) reduces to

w
1- j
Q
i
Z(2 o) = RLQ2 1 + j1.5Q = RLQ2
0.25 - j +
1 + 2.25Q2
(~ 1.5Q)
(11-5)

For Q = 10 this reduces to

1
Z( 2wo) =. R LQ2 j1.5Q . 1 (Lt (11-6)
= - J 1.5 '\)Ct

The ratio of the second harmonic to the fundamental frequency impedance

is then

In fact, under the extreme conditions when I p2 = I p1 , the relative power

ratio is
P L1 = I~l<RZ(wo) = RLQ2I~1 = 4(1 + 225Q2)
P L2 I~2<RZ(2wo) [ RLQ2I~2] .
4(1 + 2.25Q2)

where <R denotes "the real part oC" With Q = 10, this reduces to
P L1 = 900
P L2
Clearly, therefore, the second-harmonic power is negligible under these
conditions.
Obviously, there will be losses in the tank circuit owing to the resistive
component of the coils, and perhaps the capacitor. The power delivered
to the effective load is r..;.o,";1'f..-,
'\ ,-
,,_ "" 2 "'(WM)2 R';
PL - (Ql p1 ) R';+ R2 R'; + R2 (11-7)

and the power lost in the tank circuit is

(11-8)

The circuit transfer efficiency, which is defined as the ratio of the power
delivered to the load to that supplied to the tank circuit, is given by
216 ELECTRON-TUBE CIRCUITS [CHAP. 11

(11-9)

By defining a quantity Qu, as in Eq. (11-10), and this is the Q of the

circuit when the coupied load is removed, and writing the loaded Q of
the circuit as QL, then
WOLI
Q" = (WM)2R 2 (11-10)
Rl + +
(R'f R2)2
and the transfer efficiency may be written in the form
1/ = (l/QL) - (l/Qu) = 1 _ QL (11-11)
l/QL Qu
Thus, for high tank-circuit efficiency, QL should be low, and Qu should be
high. However, since QL must be 10 or greater in order to provide for a
low harmonic content in the output, then it is important that the value
of Qu be high. The values of Q" that are possible in practice will vary
with the power output, the character of construction of the coil, and the
frequency of operation. Typical values for coils of conventional design
vary somewhat as follows for frequencies in the range from 500 to 1,500 kc:
Qu""' 100 to 200 for low-power coils
Qu ""' 500 to 800 for high-power coils.
11-2. Class B Tuned Amplifier. Under class B operation, the grid-
bias supply voltage Ecc in Fig. 11-1 is made negative by an amount

FIG. 11-4. The output wave shape from a class B stage, with a linear dynamic curve.

sufficient to reduce the plate current to zero for zero signal voltage ego
If the dynamic characteristic of the amplifier is linear over the range of
 SEC. 11-2] TUNED POWER AMPLIFIERS 217
operation, then for sinusoidal input signal voltage the current will consist
of half-wave rectified pulses. The construction for deducing the output
wave shape is sketched in Fig. 11-4.
It is important that it be recognized that Fig. 11-4 represents an ideal-
ized picture, which depends upon a linear dynamic curve. This is not
completely true, although, in the analysis to follow, it will be assumed
that the linear relation does apply. If the dynamic curve is not linear,
then a graphical solution must be used in order to determine the shape of
the plate-current curve and the linear class B analysis is not valid.
To find the operating path of an amplifier with a tuned load, a special
construction is required, since the conditions are different from those of

--0N"-;-n-,-----=="""!O~
eb
60
120
180

ali wt
FIG. 11-5. The construction for determining the plate-current wave shape graphically
from the plate characteristics.
an amplifier with a pure resistance load. This is so because of the inter-
relation of a number of factors and the different manner of operation of
, . the circuit. Among the important factors that must be considered are
the allowable plate dissipation of the tube, the Q of the circuit, the effec-
tive shunt resistance of the tank circuit, the grid driving potential, the
/ shape of the plate-current wave, and the corresponding harmonic com-
ponents in the plate current. Ordinarily a method of successive approxi-
ations is necessary, in which a given set of conditions is assumed and a
calculation is made. If a consistent solution is not found, a second trial
must be made. This procedure must be continued until a consistent
solution is found.
Although the determination of the operating path is not essential for
the linear analytical solution to follow, the method will be discussed here,
218 ELECTRON -TUBE CIRCUITS [CHAP. 11

since it will permit a check on the validity of the linear assumptions.

Moreover, it is a general method and will also be used later in the discus-
sion of the tuned class C amplifier. The details of the construction are
illustrated in Fig. 11-5.
To find the operating path, it is assumed that the plate-voltage swing
is sinusoidal when the grid input signal is sinusoidal. Also, as a starting
i point, it is assumed that ebmin is ap-
proximately 10 per cerrt,-' of Ebb.
Tlievalue of eemax must not be al-
lowed to reach an instantaneous
r---T----I----L_~,C<lt positive potential that is higher
than the plate potential ebmin; other.!
wise the current to the grid will in-
e crease very rapidly. This may
cause serious damage to the tube.
Even if no damage results, the in-
creasing grid current is accompa-
nied by a decreasing plate current,
and in consequence the analysis
will no longer be valid owing to the
resulting nonlinearity of the dy-
1-7"~~-----'''?-''''::'''..,.--~C<lt namic curve. With this choice of
conditions, the analysis can be com-
pleted, and a caICl1~l'l.. tion can be
FIG. 11.6. The important wave shapes -Q'"
in a class B tuned amplifier. made of the following: the d-c pp;ver
. from the plate-supply source; the
a-c power output to the load, and"'the plate dissipation. If the plate
dissipation is within the rating of the tube, then the resulting calculations
will indicate the adjustments of the circuit parameters that are necessary
to achieve the indicated results.
The specific procedure is the following (refer to Fig. 11-5): Select any
particular instantaneous grid potential ee, such as that corresponding to
the point Ae. Determine the corresponding instantaneous plate poten-
tial eb by locating the point Ab at the same phase angle in the operating
cycle. By projecting Ab up to its intersection with the curve for the
selected grid potential, the point A on the operating path will be located.
Other points are determined in a similar manner. For class B operation,
the operating path should be approximately linear and should intersect
the plate-voltage axis at Ebb, approximately.
To determine the shape of the plate-current pulse as a function of the
phase angle, the current corresponding to each point A on the operating
path is plotted as a function of the appropriate phase angle. The corre-
sponding plate-current pulse is plotted in the above diagram.
SEC. 11-3] TUNED POWER AMPLIFIERS 219
The curves of Fig. 11-6 illustrate the important wave shapes of the
amplifier.
11-3. Analytic Solution of Tuned Class B Amplifier. l An analytic
solution of the tuned class B amplifier is based on finding an analytic
form for the tube characteristics. From Eq. (2-10), the general relation-
ship between the plate current and the plate and grid potentials is of the
form

Actually, it is found that for power triodes over a wide range of parameters
the plate current is of the form

ib = k (ec + ;)
which may be written in the more complete form

.
~b = gm ( ec +-;
eb) (11-12)

This is, of course, simply the first term in the Taylor expansion for the
current.
The instantaneous potentials are of the form
ec = Ecc + Egm cos wt }
(11-13)
eb = Ebb - E plm cos wt
But since the current is zero when the grid signal is zero, then, for ib = 0,

which requires that

Ecc + Ebb
Jl.
= °
or, for cutoff,
Ebb
(11-14)
/l

By combining Eqs. (11-12) and (11-13), the expression for plate current
becomes

'l.b. = gm (E cc + E gm cos wt + Ebb

--;; - Eplm
-/l- cos wt)

= gm (E gm cos wt - EPlm)
-/l- cos wt

gm Eplm) cos
-/l- wt (11-15)
= (E gm -
220 ELECTRON-TUBE CIRCUITS [CHAP. 11

which is written in the form

--<--

I.
I ~b = I bm cos wt - 11"-2 < wt <-11"2
11"
-2 < wt <-
2
311"
l (l1-Hi)

where

I~ = 9~(E~rr: - E~Im) (11-17)

The average value of the plate current is

2
Ib = ;11" 10 11" ib d(wt) (11-18)

or

(11-19)

Also, by Fourier analysis, the amplitude of the fundamental component

of the plate current is
I pim = -
1 1211" ib cos wt d(wt) (11-20)
11" 0
or

(11-21)

But at resonance Z(wo) = Ro is resistive, and the fundamental-frequency

potential difference across the load is
(11-22)
Combining Eq. (11-22) with (11-21) and (11-17),

,
E"Im =
R~Ibm Ro
~2-· = 2 9m
(E
am -
EpIm)
M
It follows from this that
E Ro E pim Ro E
pim + "2 gm -p.- = "2 gm om
or
G I pim
= R
0
J.l.E. m \
Ro + 2rp\
(11-23)
I+-
\
-~.- ----- . - )

which yields, for the rms value of the fundamental-frequency component

of current, the expression
I J.l.Eg (11-24)
pI = 2rp + Ro
SEC. 11-3] TUNED POWER AMPL(IFIERS~ "'-~ ).cf' 221
Also, from Eqs. (11-21) and (11-24), cM1 (! I ~ I I) ;..If'...., ') ) I{'
hY'~

(11-25)

The gain of the amplifier is given in Eq. (11-23) and is

K = _ p,Ro (11-26)
2rp + Ro
The d-c power input to the plate circuit, which is equal to the average
power furnished by the plate supply when the d-c power dissipated in the
plate load resistance is negligible, is given by
1
P bb = 211" Jof27r Ebb~b. d(wt)
This becomes
(11-27)

The a-c power output of importance is that at the fundamental frequency

and is given by

(11-28)

(11-29)

The plate dissipation is given by

1 f27r .
Pp = 211" Jo eb% d(wt)
or
Pp = ;11" 121' (Ebb - eL)ib d(wt) =\E:J;,-~P~~ (11-30)

which becomes, by virtue of Eqs. (11-27) to (11-29),

p =.\- P (1 -
T/p)P b; " (11-31)
\
222 ELECTHON -TUBE CIRCUITS [CHAP. 11

It is of some interest to calculate the results corresponding to the opti-

mum conditions e cmax = ebmin. For this condition

ecmax :. Ecc + Eam } (11-32)

ebmin - Ebb - E pim
from which

By Eqs. (11-14) and (11-23), this yields

Ji.Ro Ebb
Ea m + E om2-rp+ R0 = Ebb +-Ji.
or
(11-33)

The corresponding expressions for the fundamental-frequency component

and the d-c components of current are, respectively,
,~~\~ -
;,Y{,. ,\}U\. I _ Ebb(Ji. + 1) 1
(11-34)
\ , pI - v'2 2rp + (Ji. + l)Ro
and '
(11-35)

The corresponding values of the optimum Pbb, Pac, and TIp are readily
calculated from these expressions for I pI and lb. The expression for the
plate-circuit efficiency is found to be
Ebb(Ji. + 1) 1 )2'R
_ I';'IRo _ ( v'2 2rp + (Ji. + l)Ro 0
TIp - Ebbh - 22 1
;. Ebb(ft +-
1) 2rp + (Ji. + l)Ro
which reduces to
RoCJi. + 1)
rJp = 78.5 X 2rp + Ro(Ji. + 1) % (11-36)

Ordinarily the plate dissipation is the limiting factor on the output

pow~r. The appropriate value of Ro is then specified, since all aspects
of the circuit may be expressed in terms of it. To examine this, note that
Pp = Ebbh - l~IRo

which may be written as

P = E2 ~ (
P bb 7r Ji.
+ 1)
2rp
1
+ (Ji. + l)Ro
_ (Ebb (
y2 fJ.
+ 1)
2rp
1
+ (p, + l)Ro
)2() l'\.c

This expression may be rearranged and yields the following quadratic

 SEC. 11-4] TUNED POWER AMPLIFIERS 223
expression for R o, from which Ro may be evaluated:

~ R5
4r p
+ [ J.I + 1 -
E~b
Pp
(2 1)] R° +
;: - 2
[4r;
(J.I + 1) 2 -
E~
P p (J.I
4r P
+ 1)11"
]
=
0
(11-37)

11-4. Semigraphical Analysis of Class C Amplifiers. An analysis of

the operation of the tuned class C amplifier can be made on the basis of
the assumption of a linear tube characteristic, essentially as an extension
of the method of Sec. 11-3. 2 This analysis is considerably complicated
by the fact that E~c is no longer the single value chosen to yield a zero

Plcrle
currenf
wcrve

eb D· 60· 120· 180· cut

--O..,-..;:;:-r~TI-""'-""---===-"10-eb
60
120
180

wt
FIG. 11-7. The construction for determining the plate-current pulses in a class C
amplifier.

current for zero excitation but is now a parameter. Moreover, it is no

longer valid to assume that the operating characteristic is linear. Hence,
although such a linear-tube-characteristic analysis is possible, it is a poor
approximation. It does have the advantage over other methods of giving
an explicit solution for the optimum operating conditions. Owing to its
approximate nature, other methods are preferred.
To see that the operating path is not linear, the construction of Fig. 11-5
is again employed. The only differences that exist arise because the
grid bias Ecc is adjusted beyond the cutoff value. With such values of
Ecc and with the appropriately increased value of grid driving potential,
the results have the form illustrated in Fig. 11-7. The curves of Fig. 11-8
illustrate the important wave shapes in such an amplifier.
224 ELECTRON-TUBE CIRCUITS [CHAP. 11

A comparIson of these curves with those of Fig. 11-6 indicates that in

the class C amplifier the plate current consists of pulses, the duration of
which is less than 180 deg of the cycle. Also, it is not possible, in general,
to derive easily an analytic expression for the shape of the plate-current
pulse.
In the case of class C operation there will be no output for small grid
signals since the plate current is zero. Consequently the output voltage

PI ode omo grid currents

Plate voltage eb

Grid vOltorge ec
~rL-T~~~-L+-~~--~wt

FIG. 11-8. The wave shapes at various points in the tuned amplifier.

is not proportional to the input voltage, and these amplifiers cannot be

used where such a linear relation must be maintained. They are used
extensively for amplifying a signal of fixed amplitude. They are also
used extensively in radio communication as either low-level or high-level
modulation stages. This latter application will be examined in detail in
Chap. 15. When the amplifier is biased to class B operation, a linear
relation between the output and input voltages does exist and such
amplifiers find extensive use in those applications requiring this charac-
teristic. The most important application is to increase the power level
of a modulated carrier wave.
A second method of obtaining the operating path of a tuned power
amplifier is possible. This makes use of the fact that the operating line
 11-4] TUNED POWER AMPLIFIERS 225
SEC.
.
appears as a straight line on the constant-current (eb,e ) characteristics
o
of the tube. These constant-current tube characteristics are available
for transmitting-type tubes and are provided for this particular purpose.
To verify that the dynamic characteristic is a straight line on the con-
stant-current characteristics, use is made of Eqs. (11-13) for the grid and
plate potentials, viz.,
eo = Ecc + Egm cos wt }
(11-38)
eb = Ebb - E:plm cos wt
This latter expression is valid when the Q of the tank circuit is 10 or
greater. Now combine these expressions by writing

- eo = -Ecc + coswt
Egm Eam
~ - Ebb _ coswt
E:plm - E p1m
Adding these expressions gives
ec + eb _ E cc + Ebb
Egm E p1m - Egrn E p1m
This may be written in the form

eo = - Egm eb
-E + (E cc + ~E
Egm E bb) (11-39)
plm plm

which is the slope-intercept form of the equation of a straight line. The

results are illustrated in Fig. 11-9.

+
t
o ~~~----~~~~~-q~ ~=-~---.eb
Plaie current(ma)
!
FIG. 11-9. The operating line on the constant-current curves of a power tube.

In order to establish the range of operation, it is necessary to specify

the end points of the region of operation. Ordinarily this is done by
226 ELECTRON-TUBE CIRCUITS [CHAP. 11

specifying Ebb, ebmin, eCDlllX, quantities which are determined from con-
siderations of economy, power output desired, efficiency, and tube
ratings. The manner of this dependence will be investigated below.
With these factors specified, the operating characteristics of the amplifier
are obtained from the curves in the manner illustrated in Fig. 11-10.

eb
Locus OTconsfanf

r-----------~---------------eL

~ wt
FIG. 11-10. The operating characteristics of a class C amplifier.

11-5. Grid and Plate Currents in Class C Amplifiers. 3 In order to

obtain a numerical solution of the operational features of the amplifier,
such as power output, efficiency, grid driving power, and plate dissipa-
tion, the average and rms values of the grid and plate currents are
required. These must be deduced from the plate- and grid-current pulses
as obtained from the curves, as discussed above. It is well to examine
this matter before considering a detailed analysis of the amplifier
operation.
An inspection of Figs. 11-7 and 11-8 shows that the plate- and grid-
current pulses possess zero-axis symmetry. Consequently, these recur-
ring waves may be represented by a Fourier series involving only cosine
terms. In particular, the plate- and grid-current pulses may be repre-
sented analytically by series of the form
ib =h +I p1m cos wt +I p2m cos 2wt + ... } (11-40)
ie = Ie +I g1m cos wt +I g2m cos 2wt + ...
The average or d-c value of the plate current is given by the integral

h = 211"
1
Jo(2".
2b d(wt)
SEC. 11-5] TUNED POWER AMPLIFIERS 227

which becomes, by virtue of the zero-axis symmetry and the fact that
conduction proceeds over the angle (lb,

h = -1
71'
f~0 ib d(wt) (11-41)

This integral expresses the area under the plate-current pulse. Since,
however, an analytic expression for the current pulse is not available,
recourse is had to any of the available methods of numerical integration,
e.g., through the use of a planimeter; by dividing the base of the wave into
equal parts, approximating the mean ordinates of the resulting rectangles,

.zr .'Jtr
2 4"
FIG. 11-11. Current wave form and its approximate representation.

and then summing the areas of these rectangles; or through the use of
other methods devised for numerical integration.
The details of the second method are given. Suppose that Fig. 11-11 is
the current wave form, certain features of which are to be examined.
Suppose that the half recurrence period is divided into n equal parts;
hence each division is 71'/n = 180/n deg long. Since the current flow will
proceed for less than 90 deg in each half period, and taking account of the
symmetry, the integral for h is then given with good approximation by
the expression
(11-42)

The averge value of the grid current is found in a similar manner from
the graph of the grid-current pulse. It is

which has the form

Ie = -11~lie d(wt)
71' 0
(11-43)

where (lc denotes the grid-current conduction angle. In terms of the

228 ELECTRON -TUBE CIRCUITS • [CHAP. 11

approximate calculation, this becomes

Ie = ~ [ie~) +
k = 1,2,3'0 ••
2: e:)] ie (11-44)

The amplitude of the fundamental-harmonic component of the plate

current is obtained from considerations of the general Fourier series
representation of the current. This leads to the form

I plm = -1 10 2
.-
Zb
0

cos wt d(wt)
7r 0

which may be written, in view of the existing symmetry, in the form

I plm = - 21o~ 0
Zb cos wt d(wt) (11-45)
7r 0

This integral may be expressed as a summation by the approximate

methods that have been employed above. This becomes

I -'- Ii2 [ib(O) 2cos 0 + 4\' . (k7r)

o plm -
k7rJ
11: cos 11: Zb (11-46)
k

The amplitude of the fundamental-harmonic component of the grid

current is obtained in the same way as the corresponding component of
plate current. It is given by

I glm = -1
7r
10
2
.- ie cos wt d(wt)
which reduces to the form
8,
2
I glm = 210
-
7r 0
ie cos wt d(wt) (11-47)

In general, the grid current flows for a relatively small portion of the cycle
in the neighborhood of (}e = O. But the value of cos wt does not appreci-
ably differ from unity during this interval. Then approximately
8,
2
I g lm = -2100
7r
ie d(wt)
from which it follows that
(11-48)
In general, it is not necessary to plot the grid- and plate-current wave
forms, since the information may be taken directly from the curve of
Fig. 11-10 and combined in a table like Table 11-1 to yield the desired
results.
SEC. 11-6] TUNED POWER AMPLIFIERS 229
TABLE 11-1
AN ALYSIS OF CLASS B AND CLASS C TUNED AMPLIFIER
Tube
Ebb E •• Earn eCDlSX
ebmin
ehmin E p1m ibmax
e.uw.x
icJD.&% n k (h
ib - I - length of line PQ

1 k 0 1 2 3 4 5 6 7 8 9

2 (h
--1------)---------------------
3 cos Ok
---1------1-- - - - - - - - - - - - - - - - - - - -
4 I cos Ok
---1------1--- --- - - - - - - - - - - - - - - - -
5 i b( Ok)
--1------1----------------------
6 i.( Ok)
----1------1--- - - - - - - - - - - - - - - - - - - - -
7 ib( Ok) cos Ok
(-
\

,
<
1
\.
Ip1m = ~ [i'(O) ;os 0 + Lib e:) e:) ]
cos

11-6. Power Considerations in Class C Amplifiers. A number of the

results are the same as those considered in Sec. 11-3 for the class B ampli-
fier. Here too the d-c power input to the plate circuit, which is equal
to the average power furnished by the plate supply when the d-c power
dissipated in the plate load resistance is negligible, is given by
1
P bb = 27l" Jo(Z" EbbtbdCwt)
.
= Ebblb Cl1-49)

The a-c power output of importance is that at the fundamental fre-

quency and is given by
1
PI, = 27l" Jo(Z".
eI,tp d(wt)
1 (Z"
= 27l" Jo E p1m cos wtI plm cos wt d(wt)
which is
P I, = Eplmlplm _ E I (11-50)
2 - pI pI

The plate-circuit efficiency is

'T]Jl = PI, X 100% = Epdpi X 100% (11-51)
PbIJ Ebbh
230 ELECTRON-TUBE CIRCUITS [CHAP. 11

It might be noted that typical values for class C operation are (h in the
range 120 to 150 deg, with corresponding plate-circuit efficiencies approxi-
mately from 'YJp ,,:,J~O to 60 per cent.
The power dissipated in the plate of the tube is given by
1 (2" 1 (2"
Pp = 271"Jo Cbib d(wt) = 271" Jo
(Ebb - cL)ib d(wt)

which reduces to
Pp = Ebbh - Epripl = Pbb - P L (11-52)
By combining this with Eq. (11-51), there results
Pp = (1 - 'YJp)Pbb (11-53)
This expression shows that the plate dissipation decreases as the output
power increases, for a given plate power input.
The average grid power supplied by the driving source is given by
1 (2" .
Jo
P g = 271" cg~cd(wt)
.
This reduces, under the assumption that the grid potential is at its maxi-
mum value when the grid current flmvs and does not vary appreciably
during this interval, to

P g == E gm 271"
1
Jo(2" ic d(wt)
which is
(11-54)
The results of Thomas 4 have shown that the grid driving power is given
more accurately by the expression
-

P g = 0.9Egmlc) (11-55)
A somewhat better approximation is given by Maling, 5

Po = EornIc (0.85

Po
+ 0.16 cos~)
= EgmIc (0.81 - 0.20 cos ~)
for triodes

for tetrodes and pent odes

l (11-56)

The average grid dissipation is !iven by the expression

1 [2" .
Pc = 271" J0 Cc~c d(wt)
This may be written as
1 [2" .
Pc = 271" Jo (Ecc + cohc d(wt)
= EcJc + EomIc (11-57)
SEC. 11-7] TUNED POWER AMPLIFIERS 231
but the first term gives a measure of the amount of power that the grid
battery is absorbing from the input driving source, since

Pee = 27l'
1
Jo(2" Ecc~c. d(wt) = Eccle (11-58)

and Ecc is inherently negative. Hence the power dissipated in the grid
circuit is
(11-59)
Example: In order to illustrate the calculations for a typical transmitting tube,
consider the following specific problem: A type 806 triode having the constant-
320.----,~T--r-_r--,------,--_r_-,.._-_.____,

240J---p.,.,!,,-,-~

-160 I--I---I---I---\--f-""'-~f------II-r~

-2~L-__L __ _L -__L __ _~~L-~~~~~--~

FIG. 11-12. Constant-current characteristi9s of an 806 triode.

current characteristics shown in Fig. 11-12 is used ~~ a class C amplifier, under

the following conditions: ,
Ebl. = 2,500 volts E" = -1500 volts
eb=in = 1.0 E. m = 7&5 volts
ecmwc
Determine the following:
a. Power supplied by the plate power supply.
b. A-c power output. d. Plate dissipation.
c. Plate-circuit efficiency. e. Grid driving power.
(N ote: The details of the solution are given in Table 11-2.)
11-7. Design Considerations for Class C Amplifiers. The analysis
presented above is based on the assumption that the locus of the operating
point of the tube characteristic IS known. Frequently, however, the
engineering design carries with it the requirement for the selection of the
tube and the selection of the operating conditions that govern the locus
to give a high plate-circuit efficiency, and other specified results. A num-
232 ELECTRON -TUBE CIRCUITS [CHAP. 11

TABLE 11-2
AN ALYSIS OF CLASS C AMPLIFIER
Tube-806
Ebb = 2,500 Eee = -500 Eam = 755 eemax = 255
ebmin = l.0
ebmin = 255 E p1m = 2,245 hrnax = 825 rna
ecmax
Iemax = 185 rna n = 18 k = 9
Length of line PQ = 27.8 crn

1 k 0 1 2 3 4 5 6 7 8 9
- - - - - - -- - - - - - - -
2 Ok 0 10° 20 30 40 50 60 70 80 90
- - - - -- - - -- - - - - - - -
3 cos Ok 1.0 0.985 0.94 0.86 0.76 0.64 0.50 0.34 0.17 0.00
- - - - - - - - -- -- -- - - -
4 I cos Ok 27.8 27.4 26.1 24.1 21.3 17.9 13.9 9.5 4.8 0.0
_. - - -- -- -- -- -- -- -- -
5 i b( Ok) 825 800 750 640 410 150 0 0 0 0
- - -- -- -- -- -- -- -- -- -
6 ie( Ok) 185 170 120 55 12 0 0 0 0 0
---- -- - ~-
- -- -- -- -
7 i b( Ok) cos Ok 825 788 710 555 314 96 0 0 0 0
-, ;~

h = 718(82% + 2,750) = 176 rna

Ie = 718 (18% + 357) = 25 rna
Ip1m = ~iJ(82% + 2,463) = 319 rna
P bb = 2,500 X 176 = 440 watts
PL = 2,245 X 319 = 357 watts
2
1J = X 100% = 81%
35~~40
Pp =
(1 - 0.81) X 440 = 83.5 watts
P a = 0.9 X 755 X 0.025 = 17 watts

ber of factors are important in such a design, and it is desirable to examine

the influence of these.
The important factors that are involved in the enginering design of, a
elastl C amplifier are the following:
1. The peak space cur:rent that should be demanded of a given tube.
This is usually controlled by the values of Cbmin and Cernax , since the total
peak-space-current demand is given by
I,max = I bmax + I emax
= !(Cbmin, Cemax)

2. The minimum potential to which the plate falls, Cbmin.

3. The maximum value of the instantaneous grid potential, cemax •
4. The angle of plate-current flow, fh.
5. The angle of grid-current flow, Be.
SEC. 11-7] TUNED POWER AMPLIFIERS 233
6. The plate supply potential, Ebb.
The influence of each of these factors is considered in some detail.
Item 1. In so far as the total space current that may be safely drawn
in a vacuum tube is concerned, it is limited by the allowable emission
from the cathode, if saturation current may be drawn from the tube.
Although it might not be too unreasonable to draw emission saturation
current on the current peaks in a tube that is provided with a pure tung-
sten filament, it is unwise to drive a tube with either a thoriated-tungsten
or an oxide-coated cathode to such extremes. Reasonable figures for the
average emitter are
Tungsten filament-lomax approximately 100 per cent of total emission
current.
Thoriated-tungsten-I omax from 15 to 35 per cent of the total emission
current.
Oxide-coated cathode-I.max from 10 to 20 per cent of the total emission
current.
Items 2 and 3. The optimum values of ebmin and eemax will be such that
the total allowable peak space current will not be exceeded. Moreover,
their relative values must be so chosen that the maximum plate current
occurs at wt = O. This requires that the tube must not be driven so hard
that it operates in the region of rapidly falling plate current. Such a
condition is avoided by keeping ebmin > eem... However, high plate-
circuit efficiency results when ebmin = ecmax , although for low grid driving
power it is required that ebmin > eemax. Typical values of the ratio
ebmin/ eem.. usually range from 1 to 2.
Item 4. The range over which plate conduction occurs, i.e., the conduc-
tion angle 8b, influences both the average current h and the first-harmonic
current amplitude I pim. For a large value of the first-harmonic current
amplitude, it is desirable that 8b be made large. However, in order to
provide a high value of plate-circuit efficiency, small values of Bb are
indicated. Consequently, it is necessary to compromise between plate
efficiency and power output. Typical values for class C operation, as
already discussed, are Bb in the range from 120 to 150 deg, with correspond-
Ll'; plate efficiencies 7] from about 80 to 60 per cent.

With the choice of I., ebmin, ecmax , and 8b specified, the other operating
conditions are established. It. is desired, therefore, to examine the rela-
tion that expresses the grid bias, E ee , and also the grid conduction angle
Be, in terms of the fixed parameters. To find an expression for E ee , it is
noted that the plate current becomes zero when wt = 8b /2. At this
point, the grid signal is given by

(11-60)
234 ELECTRON-TUBE CIRCUITS [CHAP. 11

But at this point it is necessary that ec + ellP, = O. This follows from

the fact that the plate current may be written by an expression of the
form ib = f(e o + eb/ jJ.) and, for ib to be zero, eo + eb/J.l must be zero. By
virtue of this

fh
( Egm cos 2 + Ecc) +;.1 ( Ebb - E p1m cos Ob\ = 0
2)
But since
ecmax = Egm + Ecc
ebmin = Ebb - E p1m
it follows that

[ (eomax - EGo) cos ~ + EccJ + ~ [ Ebb - (Ebb - ebmin) cos ~] =0

from which

E cc = -Ebb + (e cmax
+ebmin) eos
8
~ (11-61)
jJ. J.l cos...!! - 1
2
The angle of grid flow is readily determined, since the grid current
becomes zero when wt = Oc/2. At this point

ec = Egm cos ~ + Eco = 0

from which it follows that
00 Eoc
cos- = - E- (11-62)
2 gm

where Eco is obtained from Eq. (11-61).

11-8. Approximate Analytic Solution. 6 The foregoing development
permits an exact analysis of the performance of a class C amplifier. This
analysis requires that the instantaneous plate and grid voltages and
currents should be obtained and plotted and then from these curves
should be derived such information as the plate loss, the power output,
and the grid driving power. This method, while it possesses the virtue
of yielding an exact solution, does have the disadvantage of being fairly
laborious.
An approximate calculation of the performance can be obtained without
recourse to the point-by-point analysis. This approximate calculation
takes advantage of the fact that the total space current can be expressed
quite accurately by an expression of the form

is = ib + ic = k (ec + ;:r ;or (ec + ;:) > 0 (11-63)

SEC. 11-8] TUNED POWER AMPLIFIERS 235
The constant a ordinarily lies in the range from 1 to %. The accuracy
of the results that follow under this approximation will probably be well
within the accuracy with which tube circuit conditions are known and
within the reasonable variations of individual tubes from the average of
the group.
Evidently, when the current is specified by this relationship, the pulses
of current have a form that may be analytically expressed. However,
the current pulses so specified are for the space current, and not for the
plate current. Consequently the method yields information regarding
the space-current pulses. By making reasonable assumptions, informa-
tion is also obtained regarding the grid-current pUlses. The difference
between these two must then be the plate current. Since the grid current
is ordinarily a small fraction of the total space current, then an error in
the choice of the grid-current pulse will not introduce a very large error
in the resulting plate current.
To examine the situation in some detail, use is made of the known rela-
tions for the grid and plate potentials [see Eqs. (11-13)], viz.,
ee = Eee + Earn cos wt
eb = Ebb - E p1m cos wt
with
Eam = eemax - Eec
Epl m = Ebb - ebmin
Now within the limits of current flow, i.e., within the limits from - fV2
to (h/2, the space current may be written in the form

. =
~s Ie1 rL(
eernax -
E ec ) cos wt + E + Ebb
ee --;; -
(Ebb -J.' ebmin) cos w tJ"
•
which is

.
~8 = Ie1 [(eema, ebmin) cos wt + (E cc + --;;
+ ---;;- Ebb) (1 - cos wt) J" (11-64)

But the maximum value of this expression, which occurs when cos wt = 1,
IS

(11-65)

Hence the ratio of the current at any instant to the maximum value is
given by
.
~8
I smax
=

l
cos wt +
Ecc

ecmax
+-
+
Ebb
fJ..
ebmm
-
fJ.
(1 - ros wt)
j" (11-66)
 236 ELECTRON-TUBE CIRCUITS [CHAP. 11

But the value of the bias voltage to provide for a conduction angle Ob is
given by Eq. (11-61). By combining Eqs. (11-61) and (11-66), the cur-
rent ratio becomes

78
Ismax
= [cos wt + c:s ~
cos - b - 1
(1 _ cos wt)]"
2
which reduces to the form

~ = cos "2
8b - cos wt)" (11-67)
Ismax ( cos
8b - 1
"2

This expression may be used to provide the average or d-c component

of the space current and any of the harmonip-component values. In
particular, the average or d-c component of the space current is given by
the expression

which reduces to the form

-
Is
Ismax
=-
1
11"
12 8, cos "2
(8b
8b
- cos wt)"
1
d(wt) (11-68)
o cos "2 -

• In a similar way, the fundamental component has the form

-181m = -1 f2 .. (-isI) cos wt d(wt)

1 sma.x 1(' 0 smax

which may be written as

I 2
I:: = ;;: 1~ o
2 cos "2 -
(8b 8b
cos -2 - 1
cos wt
)" cos wt d(wt) (11-69)

These expressions, for given values of a, are functions of 8b, and they may
be conveniently expressed in graphical form. The curves of Fig. 11-13
give the relation of the d-c and fundamental-frequency components of the
space-current pulse as a function of the angle of flow 8b, and the peak
amplitude Ism ....
To find the corresponding values of the plate current, the grid current
S1<]C. 11-8J TUNED POWER AMPLIFIERS 237
is approximated by assuming an analytic form for the equation of the
grid current. As the grid current is usually a small part of the total
space current, a reasonable choice for the grid current will provide good

0.60

L
Ide
m
0.50

0.40
I.t
1';n -..
./
......-:: ~ ~ ~
--
~ ~ ~ ~ I~.lS
00
~
""
I. 00
I. 25
I. SO

a
); ~ ~ ~ ~ ~!. 2S
00
/.so
" ~~~
~ ~ ~~:1.7S ./
0.30 ZOO
~~
Y" ~ ~ ~V-
Ide
" I~
II
~ 0.20 ~~ ./" ~ ~
~
IAf?' ~ ~ ~
0.10 A~ ~~
/. v~ ~P'
~~
00 40 80 120 160 200 240
{}
FIG. 11-13. Curves giving the relation of d-c and peak fundamental-f;equency COIll-
ponent of current as a function of the angle of current flow, and the peak amplitude.

results for the plate current. An expression that represents the grid
current with good approximation is *

. -_
1e k' ( Ce + Cb)"
- for Cc >0 (11-70)
p,

Consequently, by following a parallel development to that above, there

resul ts I~ ,I _

~ __ (cos ~ - cos wt)" ~(

(11-71)
lemax ee 1
cos 2" -

from which the d-c and fundamental-frequency components become

f
~ = ~ ~(cos ~2 - cos. wt)"d(wt) (11-72)
lemax 7r ee
o cos - - 1
2
* :\1aling finds that the exponent 2 is suitable for most triodes, and a value 1.4 seems
Buitahle for most tetra des and pentodes.
 238 ELECTRON-TUBE CIRCUITS [CHAP. 11

and
~(8
cos":: - cos wt)a

1
2
[elm = ~ 2 8 cos wt d(wt) (11-73)
[ emax'l1"
o cos 2'e1
These expressions are functions of 8e and have the same graphical form
as the space-current components. Consequently the curves of Fig. 11-13
are also a valid representation of the grid-current components.
'J Example: To illustrate the methods of Sec. 11-8, the example on page 231 is
repeated according to the methods of Sec. 11-8, and the corresponding results will
be compared.
Peak space current: I.m = I bmax + I,max = 825 + 185 = 1,010 rna
To find the grid current, choose a = 2; then
8, 500
2 = cos-1 755 = 480
and from Fig. 11-13
~ = 0.14 I'lm = 0.26
Iemax Icmax

Hence the currents of importance are

I, = 0.14 X 185 = 26 rna
I,lm = 0.26 X 185 = 48 rna
To find the plate current, the space current must be calculated. To do this,
choose a = %, and - 11: .
i'
8b 378 0 0
"2 = cos- l 755 = 60 8b = 120
Then, from Fig. 11-13,
I. I. lm = 0.35
- = 0.19
I. max IaJnax

from which it follows that

I. = 0.19 X 1,010 = 192 rna
I.!m = 0.35 X 1,010 = 353 rna

The plate-current components are then

Ib = I, - 1. = 192 - 26 = 166 rna
I plm = 353 - 48 = 305 rna
The significant quantities are
P bb = 2,500 X 166 X 10- 3 = 415 watts
PL = ~~ X 305 X 10- 3 (2,500 - 255) = 342 watts
Pp = 415 - 342 = 73 watts
p. = 755 X O·.z6' X 0.9 = 17.6 watts
7Jp = 84%15 X 100 = 82.3%
SEC. 11-9] TUNED POWER AMPLIFIERS 239

A more accurate calculation would require a logarithmic plot of Eq. (11-63) for
the particular tube and a determination from this of the exponent a. However,
the use of the approximate methods would not be justified under these circum-
stances in general since the effort involved would be comparable with that in
applying the methods of Sec. 11-5.

(a) (b)
Grid neutralizcrlion PIOIte neutr<'1lizOItion
FIG. 11-14. Grid and plate neutralization of a single-ended amplifier.

FIG. 11-15. Neutralization of a push-pull tuned amplifier.

11-9. Neutralization. When a triode is used as a power amplifier in

the circuit of Fig. 11-1, some feedback between the grid circuit and the
plate circuit exists through the interelectrode capacitance Cgp. This feed-
back is usually positive and tends to cause the stage to oscillate. It may
be balanced out by one of several methods of neutralization. Tetrodes
and pentodes usually do not require neutralization, as the capacitance
C gp is usually small enough in these tubes so that the amount of feedbaek
is very small. The principle of neutralization is very simple and consists
in providing feedback through external circuits in an amount equal to that
through Cgp, butin opposite phar,je.
1\v()-commonmethods of neutralization are illustrated in Figs. 11-14.
The first is called grid neutralization, and the seeond is called plate
neutralization. In Fig. 11-14a the plate is conneeted through the adjust-
able neutralizing capacitance Cn to a point in the grid circuit which has a
potential of phase op'posite to that of the grid. In Fig. 11-14b the grid is
connected to a point in the plate circuit at which the potential is of oppo-
site phase to that of the plate. The diagram of Fig. 11-15 shows the
simple method of neutralizing a push-pull tuned amplifier.
240 ELECTRON-TUBE CIRCUITS [CHAP. 11

REFERENCES
1. Everitt, W. L., "Communication Engineering," 2d ed., pp. 582-590, McGraw-
Hill Book Company, Inc., New York, 1937.
2. Everitt, W. L., "Communication Engineering," 2d ed., pp. 565-594, McGraw-
Hill Book Company, Inc., New York, 1937.
3. M.LT. Staff, "Applied Electronics," Chap. X, John Wiley & Sons, Inc., New
York, 1943.
Cruft Laboratory, War Training Staff, "Electronic Circuit and Tubes," Chap.
XIV, McGraw-Hill Book Company, Inc., New York, 1947.
Mouromtseff, 1. E., and H. N. Kozanowski, Proc. IRE, 23, 752 (1935).
4. Thomas, H. P., Proc. IRE, 21, 1134 (1933).
5. Maling, H. F., private communication.
6. Terman, F. E., and J. H. Ferns, Proc. IRE, 22,359 (1934).
Terman, F. E., and W. C. Roake, Proc. IRE, 24, 620 (1936).

PROBLEMS
11-1. A type 800 tube is to be used in a tuned power amplifier. Find the
operating curves for Ebb = 1,000 and the following three conditions of grid
characteristics:
E" = -55 volts Egrn = 170 volts peak
-95 =210
-135 = 250,

Would any of the indicated operating conditions yield class B operation? Choose
ebmin = CCmaxo

11-2. The type 800 tube is operated as an r-f power amplifier under class B
conditions under the following conditions:
D-c plate voltage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 1,000 volts
D-c grid voltage.. . . . . . . . . . . . . . . . . . . . . . .. - 55 volts
Peak r-f grid voltage. .............. 170 volts
p. . . . . . . . . . . . . 15
rp ..... 5,700 ohms
Calculate the following for ebmm
a. Power output.
b. Plate dissipation.
c. Plate-circuit efficiency.
d. The im.pedance of the tuned circuit at resonance.

11-3. A type 833A triode has the following maximum ratings for operation
as a class B r-f amplifier.
D-c plate voltage ............................... . 4,000 volts
D-c grid voltage ................................ . -120 volts
Peak r-f grid voltage .......................... . 180 volts
Plate dissipation ............................... . 400 watts
p .................•.•.....•.................. 35
Urn ......•.........••......••.•.•••........... 15,800 micromhos
 TUNED POWER AMPLIFIERS 241
It is planned to use this tube as a class B amplifier with a grid signal frequency
of 16.0 mc, a plate supply of 4,000 volts, and a value of ebmin/ecmax equal to 2.0.
Find the following:

a. A-c power output.

b. Current by the plate power supply.
c. Plate dissipation.
d. Plate efficiency.
e. Impedance of tuned circuit at resonance.
f. The values of Land C in the tuned circuit if the loaded resonant Q is 12.
g. What value of Ro should be used if the maximum allowable plate dissipation
is 400 watts and ebmin = e cmax ?

11-4. A class C amplifier uses an 851 tube and operates under the following
conditions: .

D-c plate voltage .......................... . 2,500 volts

D-c grid voltage ........ . -250 volts
Peak r-f grid voltage ........ . 450 volts
Plate dissipation....... . . . . . . ............ . 550 watts
Shunt resistance of tank circuit ....................... . 1,550 ohms
D-c grid current ................................... . 0.10 amp

Calculate the following:

a. A-c plate potential.

b. Output power.
c. D-c plate current.
d. Plate-circuit efficiency.
e. Grid driving power.
f. Grid dissipation

Assume that ebmin = Ccmax •

11-5. A class C amplifier uses an 852 tube and operates under the following
conditions:

D-c plate voltage ... . 3,000 volts

D-c grid bias ....... . -600 volts
Peak r-f signal. ........... . 850 volts
D-c plate current ... . 85 rna
D-c grid current. . . . . . . . ......... . 15ma
Fundamental component of plate current ... 120 rna peak
Calculate the following, assuming ebmin = e cmax :
a. Output power.
b. Plate-circuit efficiency.
c. Grid driving power.
d. The amplifier is to operate at 1,500 kc. Specify the elements of the plate
tank. Choose Q = 23.
242 ELECTRON -TUBE CIRCUITS [CHAP. 11

11-6. The typical operating conditions for the type 893 A-R transmitting
triode when used as a class C r-f power amplifier or oscillator are
D-c plate voltage .............. . 18,000 volts
D-c grid bias ......... . -1,000 volts
Peak r-f grid signal. ........................ . 1,630 volts
D-c plate current ................. . 3.6 amp
D-c grid current ........................... . 0.21 amp appro x
Grid driving power ..................... . 340 watts approx
Fundamental component of plate current ..... . 6.25 amp peak
Minimum value of plate voltage ebmin . . . . . . . . . . 1,000 volts

Calculate the following:

a. Power output.
b. The inductance required in the tank circuit. Assume that the effective
Q of the tank circuit = 5, resonant frequency = 1 megacycle.
c. The required capacity to tune to 1 megacycle.
d. The circulating current in the tank circuit.
e. The grid driving power. Compare this result with that shown III the
tabulation above. Explain any discrepancy.
f. The power input to the plate circuit.
g. The plate-circuit efficiency.

11-7. A type 851 triode is used as a class C amplifier. The operating condi-
tions are to be
D-c plate voltage ........................... . 2,500 volts
D-c grid voltage. . . . . . . . . .......... . -250 volts
Peak r-f grid voltage .............................. . 450
Ratio ebmin/e c_ x . .............................. . 1
Determine the following:
a. D-c plate supply power.
b. A-c output power.
c. Plate circuit efficiency.
d. Plate dissipation.
e. Grid driving power.
11-8. Repeat Prob. 11-7 using the semigraphical method of Sec. 11-8.
11-9. A class C amplifier is operated under the following conditions:
D-c plate voltage ..... 3,000 volts
D-c grid bias ... -200 volts
Peak r-f grid voltago . 360 volts
ebmin/ecmax . ............ . 2
Peak space current .............. . 2.2 amp
Conduction angle .................. . 120 deg
Ratio:
D-c grid current to peak space current ... . 0.15
D-c plate current to peak space current .. . 0.21
 TUNED POWER AMPLIFIERS 243
Peak plate a-c current to peak space current .... . 0.37
Frequency .................................... . 2 megacycles
Loaded Q ..................................... . 12
Calculate the following:
a. D-c plate current. e. Grid driving power.
b. Plate-circuit power input. f. Load impedance.
c. Plate dissipation. g. Tank-circuit inductance.
d. Plate efficiency. h. Tank-circuit capacitance.
11-10. Show the general character of the construction (like Fig. 11-7) for
determining the operating features of a class C frequency doubler. In this circuit
the output tank is tuned to a frequency that is twice the frequency of the grid
driving source. What can be said about the plate conduction angle?
 CHAPTER 12

OSCILLATORS

T HE effects of feedback in amplifiers were examined in Secs. 5-6 and

5-8. It is there shown that positive feedback is to be avoided in an
amplifier if stability is to be achieved. On the other hand, if the circuit
is provided with a sufficient amount of regenerative feedback, the vacuum-
tube circuit will serve as a generator of periodically varying waves. This
output may be sinusoidal, with a high degree of purity of wave form; it
may be an essentially square wave, hence being of high harmonic content;
or it may be of periodically recurring, though nonsinusoidal, shape.
A large variety of feed-back circuits which differ considerably in detail
are available for the production of self-sustained oscillations. These
possess certain features which are common to all. In each case a feed-
back circuit exists through which is fed back into the input circuit a cer-
tain fraction of the output and in such a phase and of such an amplitude
that self-excitation results. In the usual class of tube, the feedback
from the output to the input circuit is accomplished externally to the
tube itself by means of coupling networks. In certain special types of
tubes, e.g., a klystron of the reflex type and a magnetron of the running-
wave type, the feedback is accomplished through the electron beam itself.
Nevertheless even these can be represented by equivalent circuits which
have features that are common to all feed-back oscillators.
Conventional self-excited oscillators ordinarily operate as class C
devices, although class A oscillators are possible and will be discussed
below. It is desirable to consider the class C oscillator first before
examining the operating features of the class A type. However, it is
important to keep clearly in mind the fact that the theory of class C
oscillators is of necessity only an approximate one owing to the nonlinear
character of the region of operation of such devices. Therefore, this
theory, while contributing materially to an understanding of the opera-
tion of the device, must be recognized as a limited solution of the oscillator
problem. But by supplementing the theory with practical design and
operating data a generally satisfactory understanding is possible.
12-1. Conditions for Self-excitation. 1 To ascertain the conditions
that must be fulfilled for oscillations to be sustained in a vacuum-tube
circuit, refer to Fig. 12-1. In Fig. 12-1a, the circuit is supposed to be
244
SEC. 12-1] OSCILLATORS 245
open at the point A, as shown. If a voltage Egk is impressed on the grid
of the vacuum tube without regard to the source of this potential, then
the output current is given by gmEOk for an assumed linear operation of
the circuit. N ow because of the current I p to the input of the coupling
network a certain voltage will appear across the output terminals of this
coupling network. If this voltage is equal in magnitude and phase to the
G

A 0---+-+

Couplinq Coup//nq
nefwork network

(a)-Block diagram (b)-Equivalent circuit

FIG. 12-1. A vacuum-tube circuit with coupling between the output and input
circuits.

original voltage E gk , the circuit may be connected at the point A and the
system will continue to operate.
The transfer impedance of the coupling network is denoted by ZT and
is defined as the ratio of the output potential to the input current. From
the diagram of Fig. 12-1b, this is seen to be ZT = Egk/l p • Consequently
for oscillations to occur it is necessary that

(gmEgk + ~:k) Zr = E kg = -Eok

or

gmZr (Egk + E;k) = Eko = - Egk

But since the output voltage of the vacuum tube is related to the input
voltage by the gain,

and

gmZrE gk (1 + ~) = -Eok
Therefore, for oscillations to be sustained, it is necessary that

gmZT (1 + I~) = -1 (12-1)

It should be noted that both the transfer impedance ZT and the driving-
point impedance of the coupling network Z are involved in this expression,
246 ELECTRON -TUBE CIRCUITS [CHAP. 12

the latter impedance appearing in the expression for the gain. Equation
(12-1) may be considered as one form of the Barkhausen criterion for
oscillations.
A parallel expression for the conditions for sustained oscillations involv-
ing only the impedance Z is readily possible. Thus, noting that the
voltage E kp is given by JpZ,

(gmEOk + ~:k) Z = E kp = - E pk
This becomes

or
(12-2)

which may be written in the form

(12-3)

This is another form of the Barkhausen criterion for oscillations.

Also, if it is noted that the ratio EoklEpk is a measure of the fraction of
the output voltage that is fed back into the input circuit through the
coupling network, and denoting this ratio as {3,

(12-4)

Note that the feed-back ratio of. the network is related to the gain of the
amplifier circuit by the simple expression
13K = 1 (12-5)
This condition for oscillation was discussed in Sec. 5-8.
12-2. Influence of Transconductance gw The criteria for oscillation
given in Eqs. (12-1), (12-3), and (12-4) are valid only for the linear region
of the tube characteristics, since it is only for this region that the current-
source equivalent circuit is valid. Despite this limitation, the expressions
may be extended with significant results over the nonlinear portion of the
tube characteristics.
Refer first to Eq. (12-1). Note that the ratio KIp. is given in terms of
circuit elements external to the tube, and these elements are independent
of the voltage and current magnitudes. Clearly, gm is the only factor in
the expression that might vary and hence must be examined critically.
Likewise, in Eqs. (12-3) and (12-4), if it is assumed that p. remains sub-
SEC. 12-3] OSCILLATORS 247
stantially constant over wide excursions of signal amplitude (and this is
a reasonable assumption), and since K involves f.L and r p , or gm, then here
too gm is the only factor that must be examined critically.
The transconductance gm, which is the slope of the ee-ib curve, is not
constant for large changes of input grid voltage. In fact, it is precisely
this variable character of gm which ac-
counts for the successful operation of
oscillators. For a given set of circuit
parameters, the oscillations will build up
until the value assumed by gm is such
that the conditional equations [Eqs.
(12-1), (12-3) and (12-4)] are satisfied,
when sustained oscillations will result.
If gm cannot assume a sufficiently large
value for these equations to be satisfied,
then the voltage fed back from the out- FIG. 12-2. Sketch for defining tho
average transconductance gm.
put to the input circuit is insufficient for
maintaining the oscillations and they will die out. If gm were too large,
the voltage fed back would be greater than that required for the oscilla-
tions just to be sustained and the amplitude would continue to increase.
If the average transconductance gm is defined as the slope of the line
connecting the two extreme points on the transfer characteristic appro-
priate to the input signal, as illustrated in Fig. 12-2,

I bmax - I bmin
gm = (12-6)
Eernax - E emin

l
The Barkhausen criteria for sustained oscillations III modified form
become

< 1, decaying ~scillations

= 1, sustained oscillations (12-7)

i> 1, growing oscillations

12-3. Amplitude and Frequency Conditions. Since in Eqs. (12-7) the

quantities K, Z, and ZT are complex in general, the conditions for se1£-
sustained oscillations requires that both the real and the imaginary parts
of the expressions separately and simultaneously satisfy the appropriate
conditions. Refer specifically to the third equation of (12-7). Here the
conditions to be satisfied are
248 ELECTRON -TUBE CIRCUITS [CHAP. 12

(12-8)

where ffi and g denote the real and imaginary parts, respectively, or the
equivalent pair of expressions

(12-9)

The first of the two sets of conditions contains a great deal of information
concerning the amplitude of the oscillations and specifies, in fact, a value
of gm and in consequence determines the amplitude of the oscillations.
The second of the two sets of conditions contains information about the
frequency of oscillation.
An examination of the first of Eqs. (12-9) reveals the following general
infor:qJ.ation: Since jJ. appears in 'the denominator, then for large jJ. there
is an almost 1: 1 correspondence Between gm and K. Since, however, gm
will vary over a range from zero to some finite value, then any condition
that makes Z large (and K therefore reaches a constant value) would
permit most easily the production of sustained oscillations. That is,
sustained oscillations are favored by circuits for which Z > f p • Note
also that Eq. (12-5) is very sensitive to the angle of K. Thus since the
angle characteristic of an antiresonant circuit is very sensitive to fre-
quency near resonance, an oscillator with an antiresonant circuit permits
sustained o~cillations with good frequency stability.
12-4. Fixed Bias and Starting Characteristics. Before examming
particular types of oscillator circuits, it is well to examine the effect of the
grid bias on the operating features of the feed-back circuit. In particular,
suppose that the bias of the circuit, as indicated in Fig. 12-2, is set to a
yalue beyond the cutoff value of the tube. A number of possible operat-
ing conditions are illustrated in Fig. 12-3. Evidently the initial value
of gm = gml is zero, and Eqs. (12-7) show that oscillations cannot build up
since the circuit conditions correspond to decaying oscillations.
Suppose that a voltage eg 2 is applied, whether from an external source
or produced by a transient phenomenon. If the mean value of gm2 is
small, the conditions required by Eqs. (12-7) for sustained oscillations
may still not be satisfied and the oscillations may die out. If the signaL
say eg 3, appears on the grid and if this is sufficiently large for oscillations
to grow, the amplitude of eg 3 will increase until the conditions for sustained
oscillations are fulfilled.
SEC. 12-5] OSCILLATORS 249

Clearly, for an oscillator biased near to or beyond cutoff, the circuit

will not be self-starting. However, as is evident by comparing the results
illustrated in Fig. 12-3 with those in Fig. 12-2, the amplitude of the oscil-
lations is larger in the heavily biased oscillator. This results in an
increased efficiency, a feature that may be very desirable in a power

FIG. 12-3. A feed-back oscillator biased to the left of cutoff, and the corresponding
secants for determining the average transconductance.

Bl./I7d-up
period

~--Ec---.j
FIG. 12-4. An oscillator with grid resistor and grid-capacitor biasing. The build-up
conditions are sketched.

oscillator although it is of small consequence in low-power sources. Also,

since it is usually quite easy to induce oscillations, the over biased condi-
tion is not objectionable from this point of view.
12-5. Grid-resistor-Grid-capacitor Biasing Circuits. The use of a
grid-resistor and grid-capacitor combination, as illustrated in Fig. 12-4,
not only allows the self-starting feature but also provides for an operating
bias at or beyond cutoff. The operation of the circuit is illustrated in
250 ELECTRON-TUBE CIRCUITS [CHAP. 12

Fig. 12-4, and i8 e88entially the following: When the circuit is first placed
in operation, the grid bias is zero, and the operating point is high on the
characteristic, where the value of gm is large. The third of the criteria
[Eq. (12-7)] applies, and growing oscillations occur and continue to
increase in amplitude. On the positive portion of the swing, the grid
potential becomes positive, thus charging the capacitor. The time con-
stant of the grid resistor and capacitor is such that a substantially steady
bias is maintained. This bias displaces the operating point to the left,
as illustrated, with consequent increasing amplitude of oscillations. The
amplitude of the oscillations continues to increase until an equilibrium
condition is reached between this amplitude and the consequent bias Ec.
The magnitude of the bias may be controlled to any value between zero
and Ec by the proper choice of Rg and Cg. The values of Rg and Cg are
not critical, and they are generally determined experimentally.
As illustrated, the amplitude of oscillation will be such as to allow
a small grid current to flow during the positive peaks of the cycle. It is
this small grid current which serves to charge the grid capacitor. In
fact, the variation in grid current can be used as an indication that oscil-
lations have been established and also as a rough indication of the
amplitude.
If the time constant RgCg is too large, the bias voltage across Cg adjusts
itself slowly to sudden changes in the amplitude of oscillation. If this
rate of adjustment is so slow that the oscillations can die out before the
bias voltage can change appreciably, then with sudden changes in the
amplitude the action is very much as though fixed bias were used. As a
result, it is possible that the oscilla-
tions will die out. Hence a possible
condition is one in which the oscilla-
tions first build up in amplitude to the
tIp equilibrium value. Any slight irreg-
~="'_-.J ularity that tends to reduce the ampli-
tude of the oscillations will cause the
oscillations to die out owing to the
substantially steady bias that exists.
FIG. 12-5. Tuned-plate oscillator.
Once the grid capacitor discharges
through the grid resistor and the bias reduces sufficiently, the oscilla-
tions will again build up, until the above process repeats itself. This
intermittent operation can be overcome by decreasing the time constant
RgCg. For stability to exist, it is necessary that the bias reduce as the
amplitude of oscillations decreases.
12-6. Tuned-plate Oscillator. The tuned-plate oscillator is one in
which an antiresonant circuit is connected directly in the plate circuit
SEC. 12-6] OSCILLATORS 251
of the vacuum tube, the grid excitation being supplied by inductive
coupling to this plate circuit. The complete circuit has the form illus-
trated in Fig. 12-5. It should be specifically noted that this is just the
circuit of a tuned class C amplifier, but with the circuit providing its own
grid excitation. Consequently the analyses of Chap. 11 of the tuned
class C amplifier apply for the tuned plate oscillator except that the grid
driving power reduces the total available power output.
In order to examine certain of the properties of the oscillator the condi-
tional equation for sustained oscillations [Eq. (12-7)] will be examined.
For simplicity, it will be assumed that the grid current may be neglected.
The grid excitation is then simply
E gk = jwMIL == ZMIL
The transfer impedance is given by the relation
ZT = E kg = ZMh ' (12-10)
Ip IL+lc

ZT = (12-11)
Also, the quantity
1 +K = 1 _ _ Z_ = _r_p_ (12-12)
J.L rp +Z rp + Z
where
(12-13)

The conditional equation for sustained oscillations may then be written

in the form
(12-14)
which is

or
-M
Om - = R+o(L
C
J W - -
wC
1) - JoR
--
wCrp
+ -Crp
L (12-15)

By equating the real and the imaginary terms, there results

and
Om-
-M
C
=

wL-2--~=O
wC
R +-L
Crp

wCrp
l (12-16)
252 ELECTRON-TUBE CIRCUITS [CHAP. 12

The first of these equations may be written in the form

p.RC
gm = p.M - L (12-17)

which specifies the average value of gm and which thus provides informa-
tion, at least in principle, concerning the amplitude of the oscillations.
The second equation becomes

w2 = .J:.-
LC
(1 + R) rp
(12-18)

If the quantity Wo is defined by the relation

1
Wo == y'LC
Eq. (12-18) becomes

(12-19)

The amplitude equation (12-17) gives a relation among the circuit con-
stants from which the approximate amplitude of oscillation may be
evaluated. To do this, one must refer to the static curves of the tube
and determine from these that amplitude for which the value of gm has
the value required by Eqs. (12-7).

LL
FIG. 12-6. The sinor diagram of a tuned-plate oscillator.

The frequency equation [Eq. (12-19)] shows that the frequency of oscil-
lation will be approximately the resonant frequency of the circuit, the
factor involving the ratio Rjrp being small. However, the frequency of
oscillation wIll always be slightly higher than the resonant value. Clearly,
the tube plays only a minor part in determining the frequency of oscilla-
tion, the external circuit elements exercising the main control. In fact, '
the influence of the tube on the frequency becomes less as the shunt
resistance of the antiresonant circuit increases or, correspondingly, as the
series resistance in the tank decreases. If circuits of very low dissipation
are provided, the oscillator has a very high degree of stability.
A sinor diagram of the circuit in its steady oscillating state may he
SEC. 12-7] OSCILLATORS 253
drawn; this applies for the fundamental frequency. In the diagram
(see Fig. 12-6) the sinors are not drawn to scale owing to the different
orders of magnitudes that usually exist among the currents and voltages.
Also, angles are exaggerated for clarity.
Under most circumstances the angle a = tan- 1 (wL/R) = tan- 1 Q will
be very nearly equal to 90 deg, and the feed-
back angle (90 - a), that is, the angle between
-Epk and E gk , will be very small. This means
C
that the feedback occurs SUbstantially with 180
deg phase displacement, so that a decreasing R
G-...........>--~
plate potential reflects itself as an increasing
Tuned t;IIri~ oscillOltor
potential on the grid.
12-7. Other Oscillator Circuits. A variety of
vacuum-tube feed-back oscillator circuits exist,
each of which possesses some special character-
istics. The coupling networks of the more im-
portant types of oscillators (see Prob. 12-1 for HOIrtley
the amplitude and frequency equations) are con-
tained in Fig. 12-7. In each of these circuits
the operation is essentially class C, the essential
differences among them being in the coupling
network.
Each of these circuits provides an antireso- Colpitts
nant circuit of some type, with either inductive
or conductive coupling between the output and
input circuits. This does not imply that only
circuits which possess an antiresonant circuit
will operate successfully as an oscillator. In
L~P
C
tlP
fact, circuits in which the feedback is accom- Tuned grid tuned plate
plished through resistance and capacitance net- FIG. 12-7. The coupling
works will be examined in some detail. How- networks of the more im-
ever, the above networks do possess a feature portant oscillators.
that is common to all feed-back oscillator
circuits: they all provide for a 180-deg phase shift between the output
and input circuits. This is a necessary condition in order that regen-
erative feedback exist.
, A special word is desirable about the tuned-grid-tuned-plate oscillator.
This oscillator depends for its operation on the feedback that will be
possible through the grid-plate capacitance Cgpo In this circuit the plate
tank circuit will be tuned slightly below that of the grid tank. This
causes the plate circuit to be inductive, and a negative input resistance
results which overcomes the grid-circuit losses and thus allows oscillations
to occur.
 254 ELECTRON-TUBE CIRCUITS [CHAP. 12

12-8. Stabilization of Feed-back Oscillators.2 Vacuum-tube oscilla-

tors will suffer changes in frequency with variations in any operating
characteristic that involves either the tube or the circuit parameters.
For example, a change in temperature may cause a change in the induct-
ance and capacitance of the tank elements and may also cause a change in
the grid-cathode and the plate-cathode interelectrode capacitances. Also,
a change in plate potential will result in changes in the interelectrode
capacitances. A change in the coupled load causes a change in the shunt
resistance, with a consequent change in frequency. Although these

FIG. 12-8. Electron-coupled oscillator.

factors have been neglected in the explicit discussions given above of the
various oscillator circuits, they do play a part in determining the fre-
quency, since they will contribute to a variation of the tube or circuit
constants of the coupling network.
A number of corrective measures may be taken in order to improve the
stability of an oscillator. This would include the careful choice of the
inductance and capacitance, either with negligible temperature coefficients
or with such temperature variation that a change in one is counteracted
by an opposite change in the other. Of course, any changes that might
result from changes in the plate potential can be overcome by the use of '
adequately regulated sources.
The effect of changes in the load impedance on the frequency may be
eliminated by using an amplifier to separate the load from the oscillator.
This system is called a master-oscillator power-ampltjier arrangement,
. usually abbreviated MOPA.
The oscillator and power amplifier can be combined into a single tube,
by using a tetrode or a pentode. Such an oscillator is called an electron-
coupled oscillator. A typical electron-coupled oscillator circuit employing
a pentode is illustrated in Fig. 12-8. Here the cathode, grid 1, and grid 2
are operated as a conventional Hartley oscillator, grid 2 acting as the
ordinary anode in a triode. The current to grid 2 is small, but it is
sufficient to maintain the oscillations. The main part of the space
current serves to produce the power in the load impedance. The plate
current is controlled by the oscillator portion of the tube, but since the
plate current is substantially independent of the plate potential, except
SEC. 12-8] OSCILLATORS 255
at the very low plate potentials, there is very little reaction between the
output circuit and the oscillator section of the tube.
In such electron-coupled oscillators it is found that increasing the plate
potential causes the frequency to decrease slightly, whereas increasing
the screen potential causes the frequency to increase slightly. Hence,
by obtaining the screen potential from a voltage divider, as shown in
Fig. 12-8, and by locating the screen tap at the proper point (and this is
determined experimentally), it is possible to make the frequency sub-
stantially independent of the plate supply voltage.
The effects of a varying plate resistance rp on the frequency can be
materially reduced through the use of resistance stabilization. In this, a
resistance R j is added between the plate of the tube and the tank circuit.
This added resistance serves to make the total effective resistance in the
plate circuit so high that changes in the plate resistance of the tube have
very little effect on the frequency. The 6
resistance also serves as a convenient
means of controlling the feedback and
hence the amplitude of the oscillations.
It is ordinarily desirable that the resist-
ance be made so high that the oscilla-
tions will just barely start.
Llewellyn 2 has shown that the fre- '--------1 Z 3 1 - - - - - - '
quency of oscillation can be made to
FIG. 12-9. The basic circuit of
approac h t h e resonant f requency 0 f t h e impedance stabilization.
tuned circuit by inserting suitable react-
ances in series with the grid or with the plate, or both. This might be
called impedance stabilization. It follows from the equivalent circuit
of an oscillator shown in Fig. 12-9 that
p.E gk = ZllI l + Zld2 + Z13I3}
o= Z 21 I 1 + Z 22I 2 + Z23I3 (12-20)
o= Z 31I 1 + Z3d2 + Z33I3
where
Zll = rp + Zl + Z5 Z22 == Zo = Zl + Z2 + Z3 + 2ZM}
Z12 = -Zl + ZM Z23 = -(Z2 + ZM) (12-21)
ZI3 = -ZM Z33 = rg + Z2 + Z4
and
(12-22)
The resulting expressions that obtain by equating the real and the imagi-
nary terms are
X O[rp(X 2 + X 4) + rg(XI + X + p.rg(X + X M)(X 2 + X M)
5)) I
= XOXMp.r o - (Xl + XM)2rO + (X2 + XM)2rp (12-23)
256 ELECTRON-TUBE CIRCUITS [CHAP. 13

from the reals and

Xo[rprg - (Xl +
X S)(X 2 +
X 4 )] - 2XM(X l + X M)(X + X M)
2
= -xox~ - (Xl +
XM)2(X2 X 4 ) - (X2 + + XM)2(Xl + Xs) (12-24)
from the imaginaries. If in Eq. (12-24) X 4 and X s have such values as to
satisfy the condition
2X M(X l + Klf)(X2 + X M)
= (Xl + XM)2(X2 + X + (X2 + XM)2(Xl + Xs)
4)

which contains all terms not containing Zo, then the resonant frequency
is exactly that to cause X °to become zero and
to remain so independently of r p , r g , and fJ..
That is, the frequency of oscillation is exactly
Cs
the series-resonant frequency of the tuned
circuit.
L -_ _ _Ir-:-c__-.-;...J As a particular example, consider a plate-
stabilized Hartley oscillator, as shown. The
FIG. 12-10. A plate-stabil-
ized Hartley oscillator. cond··
ItlOn f or sta b·l·· b b· .
1 IzatlOn ecomes, y WrItmg
X 4 = 0,

Ll - M)
X S = 2wM ( £2 +
M - wL 2 L2
(Ll ++ M)2
M - wL I

which requires that X s be negative. By setting

1
Xs = -
wC s
and since, for Xo = 0,
2 _ 1
w - C(LI + L2 + 2M)
then
Cs =C Ll + L2 + 2M
L
1
+L 2
(Ll
L2
+
+ M)2
M
_ 2M (Ll
L2
+
+ M)
M
For the ideal case in which the effective tank circuit Q is extremely high,
the compensation is perfect, and the frequency is independent of the tube
voltages. In the actual case, the compensating reactances must be
adjusted experimentally, and the compensation, although not perfect,
represents a substantial improvement in independence of frequency from
tube variations.
Several other cases are illustrated in Fig. 12-11.
12-9. Crystal Oscillators. 3 The frequency stability of an oscillator
can be made very high by utilizing piezoelectric crystals as antiresonant
circuits. Such crystals, which are sections cut from a quartz crystal in
SJ;;C. 12-9] OSCILLATORS 257

such a \yay that the flat sides are perpendicular to an electrical axis,
when stressed or compressed along this axis, are accompanied by the
appearance of electric charges on the surface of the crystal. Conversely,
when such crystals are placed in an alternating electric field, they are set

lc 2
CS =C'_k 2

.k=~
YL1 L z
FIG. 12-11. Impedance-stabilized oscillators.

into mechanical vibration. If the applied electrical frequency is very

near to that which produces mechanical resonance, the amplitude of the
vibrations will be very large.
A vibrating crystal can be replaced by an equivalent electrical circuit,
as shown in Fig. 12-12. In this circuit, em represents
the capacitance of the crystal, and its mounting when
it is not oscillating, the series combination L, e, and
R, represents the electrical equivalent of the vibra-
tional characteristics of the material. L is the electri-
cal equivalent of the crystal mass that is effective in
vibration, e is the electrical equivalent of the FIG. 12-12. The
crystal compliance, and R is the electrical equivalent equivalent elec-
trical network of a
of the coefficient of friction. The values of Land C vibrating qua r t z
are in series resonance at the frequency of mechanical crystal.
resonance.
One of the most common types of crystal-controlled oscillators is illus-
trated in Fig. 12-13. When the
crystal is replaced by its equivalent
circuit, it is seen that the oscillator is
essentially of the tuned-plate-tuned-
grid type, the crystal making up the
tuned-grid· circuit. Owing to the ex-
FIG. 12-13. A crystal-controlled os- I' Q f h . I
cilia tor. treme y hIgh 0 t e eqmva ent
circuit of the crystal, which may be
100 times as high as that of a conventional electrical circuit, the crystal
can oscillate only over a very narrow frequency range. As a result, the
258 ELECTRON -TUBE CIRCUITS [CHAP. 12

frequency stability of such an oscillator is very high. When the tempera-

ture of the crystal is maintained constant, the frequency drift may be
made less than 1 part in 106 •
12-10. Class A o scillators. 4 Some stabilization is made possible by
operating an oscillator in class A instead of class C, for this eliminates the

FIG. 12-14. A linear-stabilized oscillator.

grid current and any nonlinear effects resulting from it. Moreover, the
output wave shape from a class A oscillator will be sinusoidal, with a high
degree of purity of wave form. Owing to the manner of its operation, the
oscillating frequency is determined by the resonant elements alone.
However, since the self-regulating amplitude-control feature of the non-

FIG. 12-15. The circuit of a linear-stabilized oscillator.

linear tube characteristic is no longer being employed, other methods

must be provided in order to stabilize the amplitude of the output. One
way for providing for linear stabilized operation is illustrated in Fig.
12-14. In this circuit, the output is coupled to a rectifier, the output of
which is used to control the d-c bias of the oscillator. If the gain of the
amplifier before rectification is large, thus yielding a large d-c output, the
oscillator will operate as a linear amplifier.
A circuit showing the details of such an amplitude-stabilized oscillator
is given in Fig. 12-15.
Amplitude stabilization may be effected by providing an amplitude-
sensitive network to contrQI the output of the oscillator. Such a method
SEC. 12-11] OSCILLATORS 259

is employed in the Wien bridge oscillator, which is described in the next

section.
12-11. Resistance-Capacitance Oscillators. A form of coupling net-
work that has been used extensively in relatively low frequency oscillators
is given in Fig. 12-16. The phase shift through such a network as this
is a fairly sensitive function of the fre-
quency, and such RC oscillators are quite
stable. However, such a simple network
will not provide a large phase shift be-
tween the input and output terminals, and

FIG. 12-16. An RC coupling net- FIG. 12-17. An RC coupled

work for an oscillator. oscillator.

it is necessary to incorporate a second vacuum tube in the circuit in order

to provide an additional 180-deg phase shift. The circuit, when drawn in
the manner of the previous circuits, is as shown in Fig. 12-17.
This circuit may be analyzed in a direct manner by an application of
Eqs. (12-7). Refer to Fig. 12-18, which shows the complete coupling

FIG. 12-18. The complete coupling circuit of the RC oscillator.

network, in which the amplifier is replaced by a "black box" which

provides a gain K and a phase shift of 180 deg. An inspection of this
diagram shows that
E gk = KEp'k
This may be written in the form
Egk = KIZ g
if it is assumed that the input impedance to the amplifier is very high.
The transfer impedance of the network becomes
ZT = E kg = _ Kl$g
Ip IL + I
But as
260 ELECTRON -TUBE CIRCUITS [CHAP. 12

then
KIZ. KZgR L
(12-25)

Also, it is noted that

(12-26)

Then the term 1 + (KIJ.L), which appears in Eq. (12-7), becomes

1+K = rp (12-27)
J.L + RL(Zg + Z/)
rp RL + Zg + Z/

and Eq. (12-7) for sustained oscillations becomes

gmKZgRL rp = 1 (12-28)
rp(RL + Z" + Zf) + RL(Zg + Z/)
This may be written in the form

-gmK [ ZyRL - fJ.RL

K
(Zg + Zj) ] = RL + Zg + Z,
from which it follows that

KR
_ RL + Zy + Z, (12-29)
gm
1-.J--~
L - ( )
Z
g J.LK J.LK
By including in this expression the known values of Zg and Z/, namely,

and equating the real and the imaginary terms, two expressions result.
They are
- RL ( 1 )
-gm- RIR2 -
J.L
2C C = R2(R 1
W12
+ R L) - 2C1 C W12
(12-30)
{1m IlL (R2 + Rl _ (J.LK _ 1) R2) = _ (R2 + RL + R + R2) t

J.L C 1 C2 C2 Ct C2 C2
The first of these yields for the frequency the expression
1
(12-31)
SEC. 12-11J OSCILLATORS 261
Ordinarily Rl = R 2» R L; RL > rp; and Eq. (12-31) reduces to

",2 == 1 (12-32)
C l C 2R l R 2
The amplitude equation leads to the expression
- }.t[(Rl + R2 + RL)Cl + R2C2] (12-33)
gm = RL\[(}.tK - 1)R2 - Rl]C! - R 2C2}
For the speeial conditions for which
Rl = R2 = R
C l = C2 = C
then

'" == C~3}.t 1.' 3

gm == (}.tK - 3) RL = KRL
I (12-34)

The use of the phase-inverting amplifier stage ordinarily introduces so

much gain that a voltage divider must be incorporated in the circuit to
control the output.
Ordinarily the circuit is altered slightly, with distinctly superior results.
The modifications of the circuit provide for class A rather than class C

Po__------,

) - - -....K

FIG. 12-19. Wien bridge oscillator.

o---t====--__ --4G
FIG. 12-20. The Wien-bridge-oscillator
coupling network.

operation. The altered circuit provides the phase-reversing stage with

negative feedback, thus providing for an improved wave form. Also,
there is incorporated in the circuit an automatic amplitude control. The
praetieal circuit 5 of such an improved RC oscillator, which is known as a
Wien bridge oscillator, is given in Fig. 12-19. The reason for the name
is to be found in the form of the coupling network to tube Tl. A sketch
showing the bridge nature of the network is given in Fig. 12-20.
The resistor R4 is usually a tungsten filament lamp, which possesses a
positive temperature c6effieient of resistance; the hotter the filament of
the lamp, the higher its resistance. This provides automatic control of
the amplitude of the oscillations. If the amplitude of the oscillations

-.
262 ELECTRON-TUBE CIRCUITS [CHAP. 12

tends to increase, the lamp becomes hotter and R4 increases. As a result,

the grid-cathode potential of Tl is reduced, with a resulting decrease of
amplitude. If Ra is properly adjusted initially, the amplitude of the
oscillations is maintained at a substantially constant level, even though
the frequency may be varied over very wide limits.
A simple phase-shift 6 oscillator is possible which incorporates RC net-
works to provide the requisite 180-deg phase between input and output
potentials. Consequently, only a single tube is required in the circuit.
The circuit is illustrated in Fig. 12-21. It
may best be understood by supposing that
each L section, consisting of a C and R
combination, shifts the phase by 60 deg.
The use of three such sections will shift
the phase of the output by a total of 180
deg relative to the input.
FIG. 12-21. A phase-shift oscil- This oscillator operates as a class C de-
lator. vice, but the wave form of the output is
very nearly sinusoidal if the bias on the
tube is adjusted to a value which barely allows oscillations to be main-
tained. The frequency stability of the circuit is also quite good.
Owing to the fact that the range over which the frequency can be
changed is not very great, such phase-shift oscillators are relatively
limited in their applications. The frequency c.an be changed by changing
any of the phase-shifting capacitors or resistors. Large frequency
changes are not possible unless both Xc and R are changed, although
the ratio Xc/R must be maintained in order that the phase-shift of the
section remain unchanged.
12-12. Negative-resistance Oscillators. One'may consider the fore-
going analyses of feed-back oscillators as a demonstration of the fact that
it is possible to devise circuits containing vacuum tubes in which the
power generated is sufficient to overcome the losses of the circuit and also
to provide the power that is transferred to an external circuit. If the
total loading or dissipation within the circuit is represented by a certain
equivalent resistance in the plate circuit, then one might consider the
tube as representing a negative resistance of such a magnitude as just
to overcome the total dissipative terms. The oscillations in the circuit
will then be sustained at the stable level required by the variations of the
negative-resistance properties of the circuit.
If one is able to find a device that possesses a negative resistance, i,e"
a device in which a positive increment of current through it is accom-
panied by a negative potential increment across it, then this can be used
to neutralize the positive resistance representing the total dissipation.
SEC. 12-12J OSCILLATORS 263
Such negative-resistance devices do exist, the simple tetrode operating
with a plate potential below that of the screen being a common example.
The connections of such a device and the plate characteristics which show
the region of negative plate resistance are shown in Figs. 12-22.

L-------------------~eb

FIG. 12-22. A negative-resistance tetrode circuit, and the plate characteristics,

showing the region of negative plate resistance.

A circuit that exhibits an effective negative resistance and at the same

time avoids the objectionable features of secondary emission is illustrated
in Fig. 12-23. A pentode is operated with a plate potential that is lower
than the screen potential, and the suppressor grid is maintained slightly
negative relative to the cathode. Since the plate is at a low positive
potential, it does not exert much force on the electrons and under these
conditions the suppressor grid re-
pels most of the electrons that c
manage to get past the screen
grid, with a resulting higher screen
current.
R A
If the suppressor voltage is in-
creased slightly, i.e., is made less
negative, then there will be less
repelling action by the suppressor FIG. 12-23. A pentode circuit that ex-
and more plate current will flow hibits a negative output resistance.
at the expense of the screen cur-
rent. The electrons that were previously being repelled by the suppressor
and returned to the screen will now pass to the plate, with a consequent
reduced screen current. Note that even if the screen potential is increased
by the same potential as that applied to the suppressor grid the net effect is
still a reduction of the screen-grid current. That is, if the screen current
were to increase somewhat with the increase of screen potential, the
decrease in screen current owing to the action of the suppressor grid is so
much greater that the net effect is a reduction of the screen current.
Therefore, with the circuit shown, there is a decrease of current through
the terminals AB with an increase in potential across these terminals,
with a consequent negative resistance.
264 ELECT RON -TUBE CIRCUITS [CHAP. 12

To analyze the circuit analytically, 7 it is assumed that the change in

screen current is a linear function of the changes in the suppressor-grid
and screen-grid voltages and also that the suppressor-grid current is
negligible. That is, it is assumed that

The factor g32 has the dimensions of a conductance and is such that
g32 Ae e3 gives a measure of the influence of a change in current ie2 due to a
change in potential of the suppressor grid. Note from the foregoing dis-
cussion that g32 is inherently negative since a positive Aee3 is accompanied
by a negative Aie2 . The factor TvZ is a measure of the change in ie2 due to
a change in ee2.

FIG. 12-24. A negative-resistance, or dynatron, and a negative-transconductance,

or transitron, oscillator.

If it is assumed that g32 and Tu2 remain constant over the range of opera-
tion, and by noting that with a large C and R a change in voltage Ae e 2
appears on the suppressor as a change Ae e 3, then

Aie2 = (g32 J..)

+ Tu2 Aee2

The input resistance between points A and B is then

which is negative when

-g3ZTy2 > 1
To examine the operation of the circuit of such negative resistanees as
part of an oscillator, suppose that a tank circuit is coupled to the terminals
AB of the two eireuits shown in Fig. 12-24. These eircuits may be drawn
in the manner of Fig. 12-1. Since the feed-back voltage is zero, the eir-
cuit simplifies to that shown in Fig. 12-25. This may be drawn as a
simple coupled eireuit, in the form illustrated in Fill;. 12-26.
To evaluate the charaeteristies of the cireuit, apply Kirehhoff's law to
the two-mesh network. This yields
SEC. 12-12]

- - 11
Cp
1
OSCILLATORS

J....)
( r p + Cp 11 - Cp 12
1
~
+ R + Lp + Cp
- 12 = 0
= 0 l 265

(12-35)

FIG. 12-25. The equivalent and simplified circuit of the negative-resistance oscillator.

To solve for 1 2 , the current through the inductance and load, the following
differential equation must be evaluated:

~) dIdt + (RrpCL
+ (~L + rpC
2
d I22
dt
+ rp) I 2
2
= 0 (12-36)

If it is assumed that r p , which is inherently negative, remains substan-

tially constant over the range of operation,
this equation may be solved directly to give,
for the oscillatory case,
l(R 1) L
12 = Ae -2 r+rpC t sin (wt + fJ) (12-37)
FIG. 12-26. The basic equiv-
where A and fJ are constants. The expres- alent circuit of a negative-
resistance oscillator.
sion for 11 has exactly the same form, although
with different values for A and fJ. The angular frequency of oscillation is

or
w = {l- (rp +
'\j LC rp
R) _! (~L + ~)2
4 pC r
(12-38)

Under the assumed oscillatory conditions, the expression for 12 indi-

cates that the amplitude of the oscillations may decrease, remain con-
stant, or increase, depending upon the exponential term in the expression.
If the term (RjL) + (1jrpC) is positive, then the oscillations which might
have been started in any manner will ultimately fall to zero. If the quan-
tity (RjL) + (1jr p C) is negative, the oscillations will tend to increase
in amplitude with time. For the critical case for which the quantity
(RjL) + (1jr pC) equals zero, the exponential factory is unity and the
266 ELECTRON-TUBE CIRCUITS [CHAP. 12

amplitude of the oscillations remains constant. For thIS condition

L
rp = - RC (12-39)

and the corresponding frequency is

(12-40)

Such negative-resistance oscillators are self-regulating in much the

same manner as the normal feed-
back oscillators. Thus, owing to the
variation of the negative-resistance
characteristic of the tube circuit that is
used, if the quantity (RjL) + (ljrpC)
were negative, thus allowing for con-
tinually increasing amplitude of oscilla-
tions, these oscillations would increase
until the region of operation extended
to the point where rp = -LjRC, when
the build-up condition would cease.
FIG. 12-27. The amplitude of oscil- These conditions are illustrated in Fig.
lation increases until the value of rp 12-27. It should be noted from the di-
assumes the critical value LIRe.
agram that even with an assumed
sinusoidal output potential, and this is not a required condition, the out-
put current will be nonsinusoidal.
REFERENCES
1. Barkhausen, H., "Lehrbuch der Elektronenrohren," vol. III, S. Hirzel, Leipzig,
1935.
M.I.T. Staff, "Applied Electronics," Chap. XI, John Wiley & Sons, Inc., New
York, 1943.
2. Terman, F. E., Electronic8, 6, 190 (1933).
Llewellyn, F. B., Froc. IRE, 19,2063 (1931).
Jefferson, H., Wirele88 Eng., 22, 384 (1945).
3. For further details, see Terman, F. E., "Radio Engineering," 3d ed., Sec. 8-3,
McGraw-Hill Book Company, Inc., New York, 1947.
4. ARGUIMBAU, L. B., "Vacuum Tube Circuits," p. 320, John Wiley & Sons, Inc.,
New York, 1948.
5. Terman, F. E., R. R. Buss, W. R. Hewlett, and F. C. Cahill, Proc. IRE, 24, 649
(1939).
6. Ginzton, E. L., and L. M. Hollingsworth, Proc. IRE, 29, 43 (1941).
7. Herold, E. W., Proc. IRE, 23, 1201 (1935).
PROBLEMS
12-1. Show that the amplitude and frequency of the oscillators illustrated in
Fig. 12-7 are the following:
 OSCILLATORS 267
Tuned grid:
1 _. p,RL.C
Om = M(p,L - M)

Hartley:
_ . C(RI + R 2 )(L l + L2 + 2M)
w= Om = (Ll + M)(L2 + M)

Colpitts:
_ . p,R(C l + C2)
w = ~~(~l + ~2 + r~) Om = L(p, - C1/C 2)

12-2. Two identical triodes are connected in a Franklin oscillator. Determine

in terms of the circuit parameters

a. The expression for the critical value of the resistance R zo at which oscillations
will just begin.
b. The frequency of oscillation.

Assume that the power absorbed in the tuned circuit, which determines its Q,
may be represented as that absorbed in a resistance in shunt with the inductance
and capacitance.
12-3. Obtain an expression for the operating frequency of the cathode-coupled
oscillator shown in the diagram. *

12-4. Using the 806 tube, whose constant current characteristics are given in
Chap. 11, calculate the performance 'when it is used in the oscillator circuit
shown in the figure.
*Crosby, M. G., Electronics, 19, 136 (May, 1946).
268 ELECTRON -TUBE CIRCUITS [CHAP. 12

L-2.It9mh
L=1.96mh R=.1.12JJ..
R=O

M,,7./ph
12-5. A type 852 triode has the following ratings as an r-f power amplifier and
oscillator (key down conditions without modulation per tube):
D-c plate voltage. . . 3,000 volts
D-c grid voltage... . - 600 volts
Peak r-f grid voltage. . . 850 volts
D-c plate current.. . 85 rna
D-c grid current. 15 rna approx
Driving power. 12 watts approx
Power output..... 165 watts approx
The tube is operated under rated conditions in a tuned-plate oscillator, operat-
ing at a frequency of 1 megacycle. Determine the following:
a. Grid dissipation.
b. Plate dissipation.
c. Power output.
d. Oscillator efficiency.
e. Resonant impedance of tank circuit.
J. Grid leak required.
g. Mutual inductance between grid and plate coils.
Assume that the maximum grid voltage is equal to the minimum plate poten-
tial. Neglect the leakage inductances and resistances of the grid and plate coils,
and assume that the tank circuit has a Q = 20.
12-6. A type 806 triode when used as an r-f power amplifier has the following
ratings:
D-c plate voltage..... ............ 2,500 volts
D-c grid voltage. - 500 volts
D-c plate current.... ................. 195 rna
D-c grid current. . . . ....................... 25 rna
Driving power. . . . . ............... 17 watts
Grid resistor. . . ............ 20,000 ohms
Power output. . . . . . . . .. ........ 370 watts
This tube is connected as a Hartley oscillator and is operated under the condi-
tions specified. The tank tuning capacitor is 250 J.!J.!f; the resonant frequency is
2 megacycles; the loaded Q is 23.5. Determine the following:
a. The inductance, and resistance of the tank circuit.
b. The power output.
c. How far from the bottom of the tank coil is the cathode connection?
d. The oscillator efficiency.
 OS CILLAT ORS 269
Assume ec""", = ebmin.
12-7. An 833A transmitting tube has characteristics that may be represented
approximately by the equations
ib = 4 X 10- 4 (25e c + eb) amp for (25e c + eo) >0
for (25e c + eb) <0
f=Smc
0=/5
C=//8ppf

This tube is to be operated as a power oscillator in the ci~uit shown in the accom-
panying figure. Assume that ebmin = ec""",. Calculate the following:

a. Power input to the plate circuit.

b. A-c power output.

When calculating the grid signal voltage, neglect the grid driving power.
12-8. Typical constants of a crystal are

R = 1,500 ohms
L = 250 h
C = 0.04 p.p.f
Cm = 8 p.p.f

n. Calculate the Q of the crystal.

b. Calculate the series- and parallel-resonant frequencies.
c. What is the percentage change in the series and parallel frequencies if Cm
is doubled?

12-9. Crystals may be used in what are called series-resonant crystal-controlled

oscillators. * Discuss the operating features of the accompanying figures.

12-10. An amplitude-controlled oscillator uses a 6J5 tube. It operates at

5 megacycles. A measurement of the network and rectifier when the oscillator
'Butler, F., Wireless Eng., 23, 157 (1946).
270 ELECTRON-TUBE CIRCUITS [CHAP. 12

Recfif"ier
and
°F/lrer
Oscillerror

tube is removed indicates that with an input of Ep = 8 volts rms the potentials
are Eg = 0.4 volts rms, E" = 11.4 volts d-c. What is the value of Ro for the bias
to adjust itself to -7.5 volts?
12-11. A cathode-coupled Wien bridge oscillator is shown in the diagram for

•R.

this problem. Determine the critical value of Rz at which oscillations will just
start.
12-12. The plate characteristic of a 24A tube connected as a tetrode is given
in the figure. The d-c potential Ebb is adjusted to 45 volts. A parallel-resonant
circuit tuned to 1 mc is used, with C = 250 f.Lf.Lf.

i'1PJ==f En1
00 50 100
ell
150 200

a. Determine the minimum value of R for which oscillations will be sustained.

b. Plot the oscillation amplitude as a function of R.
c. Plot the current wave shape for maximum oscillation amplitude.
12-13. Discuss the operation of the transition oscillator illustrated in the
diagram.
 CHAPTER 13

RECTIFIERS

ANY electrical device having a high resistance to current in one direction

and a low resistance to current in the opposite direction possesses the
ability to convert an a-c current into a current which contains a d-c com-
ponent in addition to a-c components. An ideal rectifier would be one
with zero resistance in the forward direction and with an infinite resistance
in the reverse direction. A number of devices possess nonlinear charac-
teristics, among which are high-vacuum thermionic diodes, gas-filled and

,
,,
ln~fi?l'c
i
FIG. 13-1. A simple half-wave rectifier circuit.

vapor-filled thermionic diodes, pool-cathode mercury arcs, and certain

crystals.
The important rectifiers for power purposes fall into two general
groups, depending on their inherent characteristics. The vacuum
rectifier possesses an infinite resistance on the inverse cycle, as the tube
will not conduct when the plate is negative with respect to the cathode.
On the forward, or conducting, portion of the cycle, the vacuum diode is
characterized by an almost constant and low value of resistance. The gas
or vapor rectifiers also possess an infinite resistance on the inverse cycle,
but as discussed in Sec. 2-17, they are characterized by a substantially
constant tube drop during conduction. Owing to these differences, the
resulting operation in a circuit is slightly different. A detailed discussion
is included below.
13-1. Single-phase Half-wave Vacuum Rectifier. The basic circuit
for half-wave rectification is shown in Fig. 13-1. It is assumed that the
load is a pure resistance. Also, it is supposed that the power transformer
is ideal, with negligible resistance and leakage reactance.
An application of Kirchhoff's law to the load circuit yields
(13-1)
272 ELECTRON-TUBE CIRCUITS [CHAP. 13

where e if) the instantaneous value of the applied potential, eb is the

1nstantaneous voltage across the diode when the instantaneous current is
ib, and Rl is the load resistance. This one equation is not sufficient for
the determination of the two unknown quantities ib and e~ that appear
in the expression. Here, as for triodes and multielectrode tubes, a second
relation is contained in the static plate characteristic of the tube. Con~
sequently a solution is effected by drawing the load line on the plate
characteristic.
There is one significant difference between the solution of the diode as
a rectifier and that for the other tubes as amplifiers. With the rectifier,
an a-c potential is applied from a source, this source supplying the power

FIG. 13-2. The static and dynamic characteristics of a rectifier.

to the circuit. A vacuum tube as an amplifier converts direct current

from the plate supply into alternating current.
The dynamic characteristic for the rectifier is obtained somewhat
differently from the corresponding curve for an amplifier. The pro-
cedure is illustrated in Fig. 13-2. For an applied potential e, the current
is the intersection of the load line with the static characteristic, say point
A. That is, for the particular circuit, .the application of the potential e
results in a current A. This is one point on the dynamic curve and is
drawn vertically above e in the diagram. The slope of the load line does
not vary, although the intersection with the eb axis varies with e. Thus,
when the applied potential has the value e' , the corresponding current
is A'. The resulting curve so generated is the dynamic characteristic.
If the static characteristic of the tube were linear, the dynamic charac-
teristic would also be linear. Note from the construction, however, that
there is considerably less curvature in the dynamic curve than there is in
the static characteristic. It will be assumed in what follows that the
dynamic curve is linear.
To find the wave shape of the current in the output circuit, the pro-
SEC. 13-1] RECTIFIERS 273

cedure followed is that illustrated in Fig. 13-3. This procedure is very

much like that used to find the wave shape in a general amplifier circuit;
in fact, the situation here is quite like that of a class B amplifier, except
that cutoff of the tube exists at zero input.

wt
FIG. 13-3. The method of obtaining the output-current wave form from the dynamic
characteristic.

If it is assumed that the relation

(13-2)
is valid during conduction, and this supposes that the static characteristic
is linear, then from Eq. (13-1) it follows that
(13-3)
or
~b
.
+ rp sm wt = lm 'sm wt
= R ! Em. when 0 .::; wt .::; 7r
1'b = 0 when 7r .::; wt .::; 27r
(13-4)
where
Em
1
m = R + Tp
The d-c power supplied to the load is defined as the product of the read-
ing of a d-c ammeter in the load circuit and a d-c voltmeter across the
load. Thus
(13-5)
Clearly, the reading of the d-c ammeter is represented by

1d-c = -1
27r
1 0
2
" •
% dex = -1
27r
f'" 1m sm ex dex = -1m7r
0
(13-6)
and so
P _ 12 R _
d ... - d-c I -
(1)2
:;;: (rv
E;"RI
+ RI)2 (13-7)
274 ELECTRON-TUBE CIRCUITS [CHAP. 13

The power supplied to the circuit from the a-c source, and this is the
power that would be read by a wattmeter with its current coil in the
line and with the potential coil across the source, is given by the integral

(13-8)

This becomes, by'Eqs. (13-2) and (13-3),

1 [27r
Pi = 211" Jo
~l(rp +R l) dO! (13-9)

which may be written in the form

Pi = J';ms(rp +R l) (13-10)
where the rms current has the value

I rruB -- ~ 1 f27r· 2 d _ 11 J{7r0 12m SIn

. 2 d _ 1m
211" J0 ~b a - '\/ 211" a a - 2 (13-11)

The efficiency of rectification is defined by the relation

Pd-c 10001 X
Tir = -po X
,
10 = 12 (ncRl
rp + R l )
rInS
10001
10

which becomes
ld_c)2 100
Tir = (1rms 1 + (rp/Rl) (13-12)

By combining this with Eqs. (13-6) and (13-11), there results

2
Im/1I") 100 (13-13)
Tir = ( 1m/2 1 + (rp/R 1)

This indicates that the theoretical maximum efficiency of the single-phase

half-wave rectifier is 40.6 per cent. But it may be shown that maximum
power output occurs when R = r p , with a corresponding theoretical plate-
circuit efficiency of 20.3 per cent.
There are several features of such a rectifier circuit that warrant special
attention. Refer to Fig. 13-1, which shows the complete wiring diagram
of the rectifier. On the inverse cycle, i.e., on that part of the cycle during
which the tube is not conducting, the maximum potential across the
rectifier tube is equal to the transformer maximum value. That is, the
peak inverse voltage across:the tube is equal to the transformer maximum
value.
Note also from the diagram that with the negative terminal of the out-
put connected to ground the full transformer potential exists between the
primary and the secondary windings of the filament heating transformer.
SEC. 13-2] RECTIFIERS 275
This requires that the transformer insulation must be adequate to with-
stand this potential without rupture. Evidently if the positive terminal
is grounded, then the transformer need not have a high insulation
strength.
13-2. Ripple Factor. Although it is the object of a rectifier to convert
a-c into d-c current, the simple circuit considered does not achieve this.
N or, in fact, do any of the more complicated rectifier circuits, either
single-phase or polyphase, accomplish this exactly. What is achieved is
a unidirectional current, certain periodically fluctuating components still
remaining in the output. Filters are ordinarily used in rectifier systems
in order to help decrease these fluctuating components. A measure of the
fluctuating components is given by the ripple factor r, which is defined as
rms value of the a-c components of the wave
r=
avg or d-c value of the wave
and which may be written as

(13-14)

where I:ms and E:ms denote the rms values of the a-c components only.
An analytical expression for the ripple factor is readily possible. It i"
noted that the instantaneous a-c component of the current is given by
i' = i - I d -c
But by definition

~2~ 10 ~;7f' 10
2 2
I:ms = .. (i - hc)2 da = .. (i 2 - 2ihc + nc) dq

This expression is readily interpreted. The first term of the integrand

when evaluated yields the square of the rms value of the total wave I;ms"
The second term yields
2~ 10
2
71' 2ihc da = 2n..
The rms ripple current then becomes
I:ms = Vr;ms - 2I~_o + no = VI;ms - nc
By combining these results with Eq. (13-13)
2
r = VI;ms - IL = I(Irms) 1 (13-15)
I d-c '\j I d-c
This expression is independent of the current wave shape and applies in
general, since the development was not confined to a particular wave
276 ELECTRON -TUBE CIRCUITS [CHAP. 13

shape In the case of the half-wave single-phase rectifier, the ratio

Irma Im/ 2 7r 1 7
he I m/7r ="2 = .5
and hence
r = y!1.57 2 - 1 = 1.21 (13-1(\)

It follows from this that the rms value of the ripple voltage exceeds the
d-c potential of the output. This merely tends to show that a single-
phase half-wave rectifier without filter is a relatively poor device for con-
verting a-c to d-c potential.
13-3. Single-phase Full-wave Rectifier. The circuit of the single-
phase full-wave rectifier, given in Fig. 13-4, is seen to bear some resem-
blance to a push-pull circuit. Actually the circuit comprises two half-
wave circuits which are so connected that conduction takes place through

"Ad\;
l
.
I
I
I
I
I
I
I
I
I
I
I
I
I I I I
I I I I
I I I I
I I I I
FIG. 13-4. Schematic wiring diagram of a FIG. 13-5. Transformer voltage and
single-phase full-wave rectifier. output load current in a single-phase
full-wave rectifier.

one tube during one-half of the total power cycle and through the other
tube during the second half of the power cycle. The output current
through the load has the form illustrated, where the portions of the wave
marked 1 flow through tube T1 and the portions of the wave marked 2
flow through tube T2

l
The d-c and rms values of the load current are found from Eqs. (13-6)
and (13-11) to be

he = 2~m
(13-17)
I rms = Im_
y2
where 1m is the peak value of the current wave. The d-c output power
SEC. 13-4] RECTIFIERS 277

is then

Pd-o _- 12d-e R Z -_ (~)2

7r
E';.Rz
(rp + R Z)2 (13-18)

By comparing this expression with Eq. (13-7) it is seen that the power
delivered to the load is higher by a factor of 4 in the full-wave case.
However, the power depends on the circuit parameters in the same way
as for the half-wave circuit.
The input power from the a-c source is readily found to have the same
form as Eq. (13-10), viz.,
(13-19)

The efficiency of rectification is

81.2 01
(13-20)
TJr = 1 + (rp/Rz) /0

This expressions shows a theoretical maximum that is twice that of the

half-wave rectifier.
The ripple factor is readily found when it is noted that

Inns = Im/ -Y2 = 1 11

he 21m/1r .
From Eq. (13-15),
r = Vl.112 - 1 = 0.482 (13-21)
Thus the ripple factor has dropped from 1.21 in the half-wave rectifier to
0.482 in the present case. What has been accomplished in the full-wave
rectifier therefore is that the rectification process has become more
efficient, with a higher percentage of the power supplied to the circuit
being converted into the desired d-c power, and with a consequent smaller
fraction remaining in a-c form, which, while producing heating of the
load, does not contribute to the desired d-c power.
A study of Fig. 13-4 indicates that when one tube is conducting, say
T1, then tube T2 is in the nonconducting state. Except for the tube drop
ibrp in T1, the peak-inverse-potential difference across T2 is 2Em , or twice
the transformer maximum voltage. The potential stress between wind-
ings of the filament transformer is seen to be the full d-c potential, if the
negative is grounded, and is sensibly zero if the positive is grounded.
13-4. Circuits with Gas Diodes. Gas diodes may be used in the half-
wave and full-wave circuits discussed above. Owing to their different
plate characteristics, the results are somewhat different. For these
tubes a sensibly constant voltage appears across the tube when the tube
is conducting, but conduction does not begin until the applied potential
exceeds the breakdown potential of the tube. The tube will consequently
218 ELECTRON-TUBE CIRCUITS [CHAP. 13

conduct for less than 180 deg in each cycle. The situation is illustrated
in Fig. 13-6.
The equation of the potential across the load during conduction is, by
applying Kirchhoff's law to the plate circuit,
(13-22)
and the corresponding expression for the current is
Em sin a - Eo
~b = Rl (13-23)

where Eo is the constant tube drop during conduction.

The d-c plate current is found by taking the average value of the
instantaneous current and is

he = 2~ i~' Em si~ - Eo da (13-24)

*--*---f--~Ot where al is the angle at which the tube fires

and a2 is the angle at which conduction
ceases. Ordinarily the applied plate poten-
tial is much larger than Eo, and the angles al
FIG. 13-6. The applied po- and 7r - a2 are very nearly zero. Conse-
tential and the current wave quently the limits on the integral of Eq. (13-
shape in a half-wave rectifier
circuit using a gas diode. 24) may be changed to 0 and 7r without
appreciable error in the result. When this
is done and the integral is evaluated, it is found that

Ide = Em _ Eo = Em
- 7rRI 2Rl 7rRI
(1 _~ EmEO)
2
(13-25)

The load voltage E d _, may be written as

E = ~m (1 - ~ ~:)
d -c (13-26)

This equation does not contain the load current. This means, of course,
that E d _c is independent of the load current, with consequent perfect
regulation.
To calculate the efficiency of rectification, it is necessary to calculate
the input power to the plate circuit. This is given by
I {" . d 1 {" E . (Em sin a - EO) d
P i = 27r J
0 e~b a = 27r 0 m sm a J Rl a

where the limits are again taken as 0 and 7r. This expression reduces to

Pi = E;"
4Rl
(1 _! EmEO)
7r
(13-27)
SEC. 13-5] RECTIFIERS 279

The efficiency of rectification is then

P d -c
. (1
4
7r
- "2 Em
Eo)2
'l/T = - ="2 (13-28)
Pi 7r 1 _ ~ Eo
7r Em
which may be reduced to the form

'l/T = 40.6 (1 - 1.87 ~:) % (13-29)

Note that this value is independent of the load current or load resistance.
To the same approximation, namely, Em » Eo, the ripple factor is given
by

r = 1.21 (1 + 0.5 ~:) (13-30)

which is slightly higher than the value with the vacuum diode. This
increased ripple results because the tube conduction is less than 180 deg.
The corresponding properties of the full-wave circuit with gas tubes
will follow a completely parallel development and yield results that bear
the same relation to the vacuum-tube case as the foregoing results do
toward the corresponding half-wave vacuum-rectifier case.
13-5. Miscellaneous Single-phase Rectifier Circuits. A variety of
other rectifier circuits exist which find widespread use. Among these are
bridge rectifier circuits, voltage-doubling circuits, and voltage-multiplying
circuits. The bridge circuit finds /~-.:---+-- - - - - - ,

extensive use both as a power 7'1

/J
..-':
rectifier and also as the rectifying (
/

system in rectifier-type a-c meters. L _ _ _ '*- _ _ _ _ -,

The rectifiers for power use utilize

thermionic diodes of both the vac-
uum and gas varieties, whereas
those for instrument use are usu-
ally of the copper oxide or crystal FIG. 13-7. Single-phase full-wave bridge
rectifier circuit.
types.
To examine the operation of the bridge circuit, refer to Fig. 13-7. It is
obServed that two tubes conduct simultaneously during one half of the
cycle and the other two tubes conduct during the second half of the cycle.
The conduction paths and directions are such that the resulting current
through the load is substantially that shown in Fig. 13-5.
The primary features of the bridge circuit are the following: The cur-
rents drawn in both the primary and secondary of the plate-supply trans-
former are sinusoidal. This permits a smaller transformer to be used for
280 ELECTRON-TUBE CIRCUITS [CHAP. 14

a given output power than is possible with the single-phase full-wave

circuit of the two-tube type. Also, the transformer need not have a
center tap. Since each tube has only transformer voltage across it on
the inverse cycle, the bridge circuit is suitable for high-voltage applica-
tions. However, the transformers supplying the heaters of the tubes
must be properly insulated for the high voltage.
A rectifier meter is essentially a bridge-rectifier system which utilizes
copper oxide elements. The voltage to be measured is applied through a
multiplier resistance to two corners of the bridge, a d-c milliammeter
being used as an indicating instrument across the other two corners.
But as the d-c milliammeter reads average values of current, the scale
of the meter is calibrated to give rms values of sinusoidal waves by apply-
ing a sinusoidal voltage to the input terminals. The indication on such
an instrument is not correct for input signals that contain appreciable
harmonics.
A common voltage-doubling circuit is shown in Fig. 13-8. The out-
putl from such a circuit is approximately equal to twice the transformer
maximum voltage. It operates by alternately charging each of the two
capacitors to the transformer peak voltage Em, current being continually
drained from the capacitors through
the load. This circuit is character-
+ ized by poor regulation unless very
large capacitors are used. The peak
R·
I invers~ voltage is twice the trans-
former peak voltage. If ordinary
rectifiers are used, two separate fila-
ment sources are required. If a
FIG 13 8 A full-wave voltage-doub-
. -. relatively low voltage system is built,
ling circuit.
and these are used extensively in
a-c/d-c radio sets, special tubes such as 25Z5 are available. These tubes
are provided with separate indirectly heated cathodes. The cathodes in
these tubes are' well insulated from the heaters, which are connected in
series internally.
An alternative voltage-doubling circuit 2 is shown in Fig. 13-9. The
output potential from this circuit, like that from Fig. 13-8, is approxi-
mately equal to twice tlie transformer maximum voltage. It operates
by charging capacitor C 1 during one half cycle through tube Tl to the
transformer peak voltage Em and during the next half cycle charges C2
through tube T2 to the potential determined by that across C 1 and the
transformer in series, the peak being approximately 2Em • The peak
inverse voltage across each tube is twice the transformer peak voltage.
This circuit may be extended to a quadrupler by adding two tubes and
SEC. 13-6] RECTIFIERS 281
two capacitors, as shown in Fig. 13-10. It may be extended to provide
n-fold multiplication, odd or even.
13-6. Controlled Rectifiers. It is sometimes necessary to provide a
control on the amount of the rectified current in a rectifier. Such con-
trolled currents are required in motor speed control, in certain electric
welding operations, in lighting-control installations in theaters, in torque
controls of various types, and in a variety of industrial control applica-

I
I
I

,
I
I
I

I
I
,
1
I

I
I
I
Ouf/;u~
I
I
I
Ol/ffUf
I
I
I
i
:
+ +
FIG. 13-9. A half-wave voltage-doubling FIG. 13-10. A half-wave voltage-quad-
circuit. rupling circuit.

tions. Such control of the output of a rectifier may be accomplished by

controlling the voltage of the power transformer feeding the rectifier, or
this might, in some cases, be controlled by inserting a controlling resistor
in the output circuit. The first method may require expensive control
equipment, and the second would be characterized by poor efficiency.
The development of thyratrons and ignitrons has made control a direct
and relatively inexpensive process. However, the basic analyses are
common to both types of tubes, and no loss of generality is incurred by
confining our attention to thyratrons.
In the discussion in Sec. 2-19 it was indicated that the complete elec-
trostatic shielding of the cathode by the massive grid structure in the
thyratron provides a means for controlling the initiation of the arc in the
tube by controlling the grid potential. Thus since the arc is extinguished
once each cycle on each negative half cycle, provided that the arc is
initiated regularly, control is possible if the point in each half cycle at
which the arc is initiated can be controlled. Such regular control is
possible, thus providing a control on the average rectified current.
In order to analyze the action of a thyratron in a controlled circuit, use
is made of the critical grid breakdown characteristic of Fig. 2-26. As
indicated in Sec. 2-19 a knowledge of this single curve is sufficient to
predict the behavior of the thyratron in the control circuit. This curve
gives the minimum grid potential required for conduction to occur for
282 ELECTRON -TUBE CIRCUITS [CHAP. 13

each value of plate potential. Thus if a sinusoidal plate voltage is applied

to the tube, the potential on the grid just to permit conduction at each
point in the cycle is found from the critical grid curve. The conditions
are illustrated in Fig. 13-11. On this diagram, corresponding to any
time t1 in the positive half cycle, the plate potential is eb1. The corre-
sponding critical grid potential is obtained directly from the critical grid
breakdown characteristic and is shown in the figure. This means that,
unless the grid potential is more positive than that given by ecg at the
particular instant, conduction will not take place. Of course, once con-

raj (6)
NegOitive control tube Positive control tube
FIG. 13-11. The grid-control curve for an applied sinusoidal plate potential to a
thyratron.

duction begins at any point in the cycle, the grid loses control of the arc
and cannot again control until the arc is extinguished.
Suppose that the circuit is so arranged that the grid potential exceeds
the critical grid breakdown value at some angle, say <p. Conduction will
start at this point in the cycle. But the voltage drop across the tube

~rtf:l
mput d
L
____
=--rt
--- l.
(
~-----'~
t
Plate .supply
poknftGrI
FIG. 13-12. A thyratron circuit with a-c FIG. 13-13. The plate potential and
plate and grid excitation. the plate current in a thyratron.

during conduction of the thyratron, like that in any gas arc discharge,
remains SUbstantially constant at a low value Eo that is independent of
the current. Consequently the current in the plate circuit of the tube is
readily found. Refer to Fig. 13-12, which gives the schematic diagram
of a thyratron circuit, and to Fig. 13-13, which shows the voltage and
current wave shapes in the thyratron.
If the tube drop after conduction has begun is Eo, then the current in
the plate load of resistance Rz during the conducting portion of the cycle is

(13-31)
 RECTIFIERS 283
SEc.13-6J

where Em is the maximum value of the applied potential. Clearly, the

current is zero until conduction takes place, after which it assumes the
value dictated by Eq. (13-31). The current follows the form of Eq.
(13-31) until the supply voltage e falls below Eo at the phase 'If" - <po.
The curr,ent remains zero until the phase <p is again reached in the next
cycle.
The average rectified current, i.e., the value read on a d-c ammeter, j~
given by the expression

I d-e = 2'1f"
1 1."-'
'I' 1'0. da
tb

E
=~ f"-'I'O(s i n a -E-o) da
2'1f"R! 'I' Em
which integrates to

he = ::'R! [cos <p + cos <Po - ~: ('If" - <po - <p) ] (13-32)

where a = wt, and where <po is de-

fined by the relation
Eo = Em sin <Po (13-33)
If the ratio Eo/Em is very small, I
then <Po may be taken as zero. I
Equation (13-32) reduces under I I
I I
this condition to the form

t~ I
I
he == i:.R! (1 + cos <p) (13-34)

The limits of variation of the angle ..

I

I~
are from 0 to 'If". I
It is clear from this analysis
that the average rectified current
can be controlled by varying the
position in the cycle at which the 120 0
~
grid voltage exceeds the critical FIG. 13-14. The character of the conduc-
grid starting value. The maxi- tion in a thyratron for various angles of
initiation of the arc.
mum current is obtained when the
arc is initiated at the beginning of each cycle. The minimum current
is obtained when the arc is not initiated, and this would occur if initiation
were adjusted to occur at the end of each cycle. Sketches showmg the
character of the results are given in Fig. 13-14.
The potential across the thyratron throughout the cycle will vary in the
manner illustrated in Fig. 13-15. During the portion of the cycle when
284 ELECTRON-TUBE CIRCUITS [CHAP. 13

the tube is not conducting current, the full applied potential appears
across the tube. During the portion of the cycle when the tube is conduct-
ing, the drop across the tube is Eo, which is assumed constant and inde-
pendent of the tube current.
The reading of a d-c voltmeter placed across the tube will be

This integrates to

E d -c = 2~ (l<P Em sin ada + ["-<po Eo da + i:<po Em sin ada)

If the peak transformer voltage Em » Eo, this reduces to

Ed"" == - t (1 + cos If) (13-35)

The appearance of the negative sign

merely means that the cathode is
more positive than the plate for most
of the cycle. It is to be emphasized
that the d-c voltmeter does not read
the value Eo when an a-c potential
is applied to the tube.
FIG. 13-15. The potential across the The readings of d-c and a-c indi-
thyratron for various current-conduc- cating instruments may be calcu-
tion periods.
lated in somewhat similar ways.
For example, the reading of an a-c ammeter in the plate lead will be

which is
Irma = ~;7r ["-<po (Em Si~~ - EOY da (13-36)

This integration is readily effected. Similarly, the reading of a watt-

meter to indicate the total power supplied to the plate circuit is

which is
P -_ -2
1 !.,,-<p° E . (Em sin a - EO) d (13-37)
7r <P
m SIn a R
I
a

This integration is also readily effected. In particular, it is essential to

SEC. 13-7] RECTIFIERS 285

set up the basic expression for any quantity before proceeding in its
evaluation.
13-7. Phase-shift Control. In the phase-shift method of control, the
conduction point of the cycle is controlled by controlling the phase angle
between the plate and grid potentials. The situation is illustrated in
Fig. 13-16. In this figure, the grid-cathode potential eo lags the plate-
cathode potential eb by the angle (), as indicated in the voltage sinor
diagram. Note from the figure that the grid potential equals the critical

~e,
ec
FIG. 13-16. Phase-shift control of a thyratron.

grid breakdown potential at the angle cp and conduction begins at this

point in the cycle. The arc is extinguished when the plate potential falls
below the value to maintain conduction.
If the applied grid potential is large compared with the critical grid
potential at the point of conduction, the angles cp and () are approximately
the same. Also, it may be assumed under these conditions that the
critical grid curve coincides with Ide G!!.Cj!'!FE,,J../j::~~!f!!!.c!.~qc0'..~afl~.J.
the zero-voltage axis. Also, if I I I
Em» Eo, then Eq. (13-34) will give
the dependence of d-c load current
on the phase angle for all values of
~----~+-------r-----~-+O
cp for which the grid voltage lags 1T 2fT Jrr
behind the plate potential. When FIG. 13-17. The average load current
the grid potential leads the plate as a function of phase angle between the
potential by any angle, conduction grid and plate potentials.
will occur very nearly at the beginning of each cycle, with full rectifica-
tion. A sketch showing these results is given in Fig. 13-17. The curve
possesses a discontinuity at the 180-deg position, since for an angle slightly
larger than 180 deg the plate current is at its full value, whereas for an
angle slightly less than 180 deg the plate current is zero.
For those cases where a small plate potential or a small grid potential
lS used, the foregoing simplifications are not valid. However, the analy-
sis follows the same general form, due account being taken of the differ-
ence between cp and () and of the angles 7r - CPo and 7r.
Circuits for achieving the phase-shift control in a single-phase system
are readily analyzed. A common circuit and the voltage sinor diagram
286 ELECTRON -TUBE CIRCUITS [CHAP. 13

that applies to this circuit are given in Fig. 13-18. The phase between
the grid and the plate voltages is controlled by means of the two imped-

FIG. 13-18. A simple phase-shifting network, and the voltage circle diagram.
ances Zl and Z2. It should be noted that the voltage circle diagram in
Fig. 13-18 has been drawn under the assumption that
Zl = R (a resistance) Z2 = wL (an inductance)
Also, the voltage circle diagram applies only during the periods before
conduction begins in each cycle. Before conduction begins, there is no
current in the load R z, and points band P are the same. From the results
of the voltage circle diagram, it is clear that the phase B between the
plate and grid potentials can be varied over the range from 0 to 180
deg by varying the control resistor R in the phase-shifting network, with
() at 180 deg when R = 0 and with () at 0 deg when R = 00. Evidently,
the load current will decrease as the resistor R decreases. The phase
angle () is, from inspection of Fig. 13-18,
() E2 Z2
tan - = - = - (13-38)
2 E 1 . Zl
If the control impedances Zl and Z2 are interchanged, then the sinor
diagram that results has the following form,
which shows that E gk leads the voltage E pk •
Then, from Fig. 13-17, no control over the
plate current is possible, and I d - c is a constant
and equal to its maximum value for all values
of B. The use of Rand C as control imped- a K
ances is possible and is generally preferred over the use of Rand L. With
an RC phase-shifting circuit, control is possible for Zl = Xc, Z2 = R
but is not possible for Zl = R, Z2 = Xc.
13-8. D-C Bias Control. The magnitude of the d-c or average rectified
current of a thyratron may be controlled by applying a d-c bias to the grid
of the tube and controlling its magnitude. The plate supply must be
an a-c potential. The situation is best understood by reference to Fig.
13-19.
SEC. 13-10] RECTIFIERS 287
The tube will conduct at the point where Eo intersects the critical grid
curve, the angle IP in the diagram. Clearly, if the negative grid potential
is too large to intersect the critical grid
curve, no conduction will be possible.
The optimum bias is that for which the
grid-voltage line is tangent to the critical
grid curve, and the tube conducts for
one-half of the cycle. For less negative Be
bias, the conduction angle IP is less than ft'
90 deg. Control is evidently possible
over the range from full conduction to FIG. 13-19. D-c bias control of a
half conduction. The results of Fig. thyratron.
13-20 show the character of the control.
13-9. Bias Phase Control. The combination of d-c and a-c potentials
as a bias yields bias phase control. A circuit for such control is given
in Fig. 13-21. The network compris-
ing Rand C serves to introduce an a-c
potential of fixed phase with respect
to the plate potential.
Suppose, for convenience, that
R = Xc. The a-c grid potential will
-I 0 I 2 3 4 then lag the plate voltage by 45 deg.
Control grid YOltVlge
The amplitude of the a-c grid poten-
FIG. 13-20. The average plate cur- tial is 0.707 that of the plate potential.
rent in a thyratron as a function of
the d-c grid-bias potential.
Suppose also that the critical grid po-
tential coincides with the zero axis.
The general features of the operation of this control are illustrated in Fig.
13-22. The conditions corresponding to three different values of Ec are
illustrated. In the first, Ec is pos-
itive; in the second, Ec is zero; in
the third, Ec is negative. For the
circuit shown, the minimum recti-
fied current occurs when Ec is neg-
ative and equal to Egm. Conduc-
tion begins at 135 deg in the cycle R
and continues until the end of the FIG. 13-21. A circuit for bias phase con-
cycle. If the d-c bias is made trol of a thyratron.
more negative than this, no con-
duction is possible. It is evident from the diagrams that conduction will
begin at the start of the cycle when the d-c bias equals 0.707 times Egm.
13-10. On-off Control. A variety of circuits exist which permit on-off
control. Such circuits would be used when it is desired to use a thyratron
288 ELECTRON-TUBE CIRCUITS [CHAP. 13

as an arcless switch or contactor. A circuit for on-off control is given in

Fig. 13-23. With switch S open, no conduction occurs since Eee is so
adjusted that it is more negative than the maximum negative critical
grid value. When the switch S is closed, the grid is tied to the cathode
and approximately maximum rectified current is delivered. The resistor
R serves to prevent short circuiting of the battery source Eee when S is
closed.
13-11. Control of Ignitrons. As discussed in Sec. 2-21, the ignitron will
not conduct at any portion of the
cycle of an applied a-c potential to the
plate unless ignition is caused by ap-
plying a potential to the igniter rod.
Moreover, since this ignition pulse
may be applied at any point in the
cycle, control of the average current is
afforded by controlling the ignition of
the tube. However, the instantane-
ous power required by the igniter
circuit of the ignitron is higher than
that required by the grid circuit of a
thyratron, and the methods of thy-
ratron control are not applicable to
ignitrons. But with control accom-

\
,,_ ......
/

FIG. 13-22. Bias phase control, FIG. 13-23. An on-off thyratron control
showing current variations for three circuit.
different values of d-c component Eo.

plished the general discussion of Sec. 13-6 is applicable, and Eqs. (13-31)
to (13-37) express the results for the ignitron as well as for the thyratron.
Several common methods of establishing the point of ignition in each
cycle of an ignitron are illustrated in Fig. 13-24. These methods utilize
thyratrons for the control of the ignitron. In the first of these diagrams,
the ignitron current passes through the load resistor, whereas, in the
second, the igniter-rod current does not pass through the load.
Refer to Fig. 13-24b, and suppose that the thyratron is not conducting.
In this circuit the capacitor C will be charged to the peak value of the
transformer voltage through the rectifier. If the thyratron grid voltage
SEC. 13-11] RECTIFIERS 289
is adjusted to permit conduction, the capacitor charge will pass through
the thyratron and igniter-rod circuits, and the ignitron will conduct
provided that the ignitron anode potential is sufficiently positive to main-
tain the discharge. The current surge through the ignitron-rod circuit

FIG. 13-24. Several ignitron control circuits.

will quickly discharge the capacitor, the thyratron anode potential will
fall below that to maintain the arc, and the thyratron igniter-rod circuit
current will fall to zero. The capacitor will be recharged through the
diode rectifier circuit in time to control the ignition point in the next cycle.
REFERENCES
1. For an analysis of the operation, see Waidelich, D. L., Proc. IRE, 29, 554 (1941).
2. Cockcroft, J. D., and E. T. S. Walton, Proc. Roy. Soc. (London), 136,619 (1932).
Waidelich, D. L., Proc. IRE, 30, 534 (1942).
Mitchell, R. G., Wireless Eng., 22, 474 (1945).

PROBLEMS
13-1. A type 5U4G is connected in a half-wave circuit to supply power to a
1,500-ohm load from a 350-volt rms source of potential.
a. On a plate characteristic of the tube, plot the load line, and from this find
the dynamic curve.
b. Obtain a plot of the output-current wave shape for a sinusoidal applied
voltage.
c. Estimate the value of the plate resistance rp from the static characteristic
at four different values of current (50, 100, 150,200 ma). Use the average
of these as the r p of the tube.
d. Plot on the curve in part b the value obtained from Eq. (13-4), and compare.
13-2. The two sections of a 6X5 diode are connected in parallel and supply
power to a 1,000-ohm load from a 325-volt rms source of potential. The effective
plate resistance of the parallel combination of dioides is approximately 125 ohms.
Calculate the following:
a. The d-c load current.
b. The a-c current (rms).
c. The reading of a d-c voltmeter placed across the diode terminals.
d. The total input power to the plate circuit.
290 ELECTRON-TUBE CIRCUITS [CHAP. 13

e. The efficiency of rectification.

f. The regulation from no load to the given load.
g. The ripple factor.

13-3. Suppose that a 6X5 tube supplies power to a 1,000-ohm load from a
325-0-325 transformer. Repeat Prob. 13-2 under these conditions.
13-4. Show that the input power to a rectifier using gas diodes may be expressed
in the form
Pi = I;=RL + EoId .•
13-5. A gas diode for which the breakdown and maintaining voltage is
taken to be 10 'Volts supplies power in a half-wave rectifier circuit to a 1,000-0hm
load from a 325-volt rms source. Calculate the following:
a. The d-c current through the load.
b. The a-c (rms) current through the load.
c. The reading of a d-c voltmeter placed across the diode.
d. The reading of an rms a-c voltmeter across the diode.
e. The power input to the plate circuit.
f. The efficiency of rectification.
g. The ripple factor.

13-6. The peak inverse plate voltage rating of a 2X2/879 half-wave high-
vacuum rectifier is 12,500 volts. Calculate the maximum d-c voltage possible to
a load, without exceeding the peak inverse voltage, when such tubes are used in
a. A half-wave circuit.
b. A full-wave circuit.
c. A full-wave bridge circuit.
d. A full-wave voltage-doubling circuit.
e. A half-wave voltage-doubling circuit.
f. Specify in each case the insulation strength of each filament transformer
when the positive terminal is grounded.
13-7. Analyze the operation of the voltage-quadrupling circuit of Fig. 13-10.
Calculate
a. The maximum possible voltage across each condenser.
b. The peak inverse voltage of each tube.
c. The required insulation strength of each filament transformer.
13-8. The arc drop in a certain thyratron is 12 volts. The tube supplies power
tv a 100-ohm resistor from a 330-volt rms supply. Calculate the reading of a
d-c ammeter in the circuit when
a. The grid and plate voltages are in phase.
b. The grid voltage leads the plate voltage by 60 deg.
c. The grid voltage lags the plate voltage by 60 deg.
Assume a zero critical grid voltage for all values of plate voltage.
 RECTIFIERS 291
13-9. The circuit of Fig. 13-18 is used to control the current in a load of 100
ohms. The voltage Eab = 330 volts rms, and the power frequency is 60 cps. A
1,000-ohm resistor and a 2-t.tf capacitor are available.
a. Draw a voltage diagram that applies for the case where conduction occurs
for less than the full half cycle.
b. Calculate the d-c current in the load.
c. Calculate the a-c power supplied by the line to the plate circuit.
13-10. The full-wave controlled rectifier for a certain application is shown in
the accompanying diagram. The resistor in the phase-shift network is adjusted
to 1,000 ohms.

JJOV·II
60",

a. Draw a voltage diagram that applies for the case where conduction continues
for less than the full half cycle.
b. Calculate the d-c current in the load.
c. Calculate the a-c power supplied by the line to the plate circuit.
13-11. An FG-27A tube operates at a temperature of 50°C and supplies power
to a 100-ohm load from an input potential source of 330 volts rms.
a. A d-c bias potential of -4 is applied. Calculate the d-c plate current.
b. A 4-volt rms grid potential that lags the plate by 60 deg is applied. Calcu-
late the d-c plate current.
c. The two sources in parts a and b are connected in series. Calculate the
d-c plate curtent.
13-12. An ignitron is used in the circuit of Fig. l3-24b. The phase of the con-
trol circuit of the thyratron is adjusted for conduction to begin 60 deg after the
beginning of each cycle. The tube drop during conduction is 20 volts. If the
applied voltage is 330 volts rms and the load resistor is 20 ohms, calculate
a. The d-c load current.
b. The a-c power to the load circuit.
c. The power dissipated in the tube.
d. The total power dissipated in the load.
 CHAPTER 14

RECTIFIER FILTERS AND REGULATORS

I T IS usually the requirement of a power supply to provide a relatively

ripple-free source of d-c potential from an a-c line. However, as seen in
the last chapter, a rectifier actually provides an output which contains
a-c components in addition to the d-c term that is desired, the magnitude
of the a-c components being specified by the ripple factor. It is custom-
ary to include a filter between the rectifier and the output to attenuate
these ripple components.
The analysis of the action of such filters is complicated by the fact that
the rectifier as a driving source contains nonlinear circuit elements, thus
requiring the solution of circuits with nonlinear elements. It is possible
to make reasonable assumptions in order to effect a solution. In conse-
quence, the results obtained are only approximate.
14-1. The Harmonic Components in Rectifier Circuits. An analytic
representation of the output of the single-phase half-wave rectifier is
obtained in terms of a Fourier series. This series representation has
the form

i = bo + l'"

k~l
bk cos ka + l'"
k~l
ak sin ka (14-1)

where a = wt and where the coefficients that appear in the series are given
by the integrals
1 (271"
bo = 27r J0 i da

bk =
1
-
f271" i cos ka da (14-2)
7r 0

ak = -
1 f271" i sin ka da
7r 0

It should be recalled that the constant term bo that appears in the Fourier
series is the average or d-c value of the current.
The explicit expression for the current in a half-wave rectifier circuit,
which is obtained by performing the indicated integrations using Eqs.
(13-4) over the two specified intervals, yields .
292
SEC. 14-2] RECTIFIER FILTERS AND REGULATORS 293

. [1 1. 2 ~ cos kwt ]
~ = 1m ;; + 2" sm wt - ;; Lt (k + 1)(k - 1) (14-3)
k=2,4,6, .••

where 1m = Ernl(rp +
R!) and Em is the peak transformer voltage. The
lowest angular frequency that is present in this expression is that of the
primary source. Also, except for this single term of frequency w, all other
harmonic terms that appear in the expression are even-harmonic terms.
The corresponding Fourier series representation of the output of the
full-wave rectifier which is illustrated in Fig. 13-5 may be derived from
Eq. (14-3). Thus, by recalling that the full-wave circuit comprises two
half-wave circuits which are so arranged that one circuit is operating
during the interval when the other is not operating, then clearly the cur-
rents are functionally related by the expression i2(a) = ilea 7l'). The +
total load current, which is i ':= i l +
i 2, then attains the form

. [2 4 ~ cos kwt ]
~ = 1m ;; - ;; (kLt + 1)(k - 1)
(14-4)
k even
kr'O

where 1m = Eml(R! +rp) and where Em is the maximum value of the

transformer voltage measured to the center tap.
A comparison of Eqs. (14-3) and (14-4) indicates that the fundamental
angular-frequency term has been eliminated in the full-wave circuit, the
lowest harmonic term in the output being 2w, a second-harmonic term.
This will be found to be a distinct advantage in filtering.
The Fourier series representation of the half-wave and full-wave circuits
using gas diodes can be obtained in the same way as above, although the
form will be more complex. This is so because conduction begins at some
small angle CPo and ceases at the angle 7l' - CPo, when it is assumed that the
breakdown and the extinction potentials are equal. But since these
angles are usually small under normal operating conditions, it will be
assumed that Eqs. (14-3) and (14-4) are applicable for circuits with
vacuum or gas diodes. The Fourier series representation of the output
of a controlled rectifier is also possible, although the result is quite com-
plex. However, such controlled rectifiers are ordinarily used in services
in which the ripple is not of particular concern, and, as a result, no detailed
analysis will be undertaken. Some results will be given below covering
these rectifiers, however.
14-2. Inductor Filters. The operation of an inductor filter depends
on the inherent property of the inductance to oppose any change of
current that may tend to take place in the circuit. Consequently any
~udQ.en changes in current that might otherwise take place in the circuit
294 ELECTRON -TUBE CIRCUITS" [CHAP. 14

without an inductance would have these rapid current changes smoothed

out by the inductance.
In particular, suppose that an
inductor is connected in series
Rl with the load in a single-phase
] half-wave circuit, as illustrated in
Fig. 14-1. For simplicity, sup-
FIG.14-1. Half-wave rectifier circuit with pose that the tube and choke re-
choke filter.
sistances are negligible. Then
the controlling differential equation for the current in the circuit during
the time that current flows is
dib + R' E:.1m sm
. wt (14-5)
L dt ,Z'tb =

is V "\ wL=O

,I I\l
i 1\

'-\ ~~=4
L-.

/ V
II~ b:::: ~ V
./
~
-~kP=ro
o 90 '"' 180
wi--
270 360

FIG. 14-2. The effect of changing inductance on the wave form of the current in a
half-wave rectifier with inductor filter. The load Rz is assumed constant.

A solution of this differential equation may be effected. This solution is

complicated by the fact that current

]
continues over only a portion of the
cycle. The general character of the
solution is shown graphically in Fig.
14-2, in which is shown the effect of
changing the inductance on the wave
form of the current. Since a simple
inductance choke is seldom used with
a half-wave circuit, further details of
the analysis will not be given.
Suppose that an inductor filter is
applied to the output of a full-wave FIG. 14-3. Full-wave rectifier cir-
cuit with inductance choke, and the
rectifier. The circuit and a sketch of wave shape of the load current.
the output-current wave shape are
given in Fig. 14-3. Since no cutout occurs in the current, the analysis
SEC. 14-2] RECTIFIER FILTERS AND REGULATORS 295
assumes a different form from that for the half-wave case. Now, instead
of considering the circuit differential equation, as in Eq. (14-5), and adjust-
ing the initial conditions to fulfill the required physical conditions, an
approximate solution is effected. It is supposed that the equation of the
potential that is applied to the filter is given by Eq. (14-4). Moreover,
it is noted that the amplitudes of the a-c terms beyond the first, and this
is of second-harmonic frequency, are small compared with that of the
first term. In particular, the fourth-harmonic frequency term is only
20 per cent of the second-harmonic term. Furthermore, the impedance
of the inductance increases with frequency, and better filtering action
exists for the higher-harmonic terms. Consequently it is assumed that
all higher-order terms may be neglected.
In accordance with the discussion, it is supposed that the input poten-
tial to the rectifier and load has the approximate form

e = -2Em - -
4Em
- cos 2wt (14-6)
7[' 37['

The corresponding load current is, in accordance with a-c circuit theory,
. 2Em 4Em cos (2wt - 1/;)
~z = -7['-R- - -37['- -v--;;::;:R~1=+~47w=2~L~2 (14-7)
z
where
2wL
tan1/; =- (14-8)
Rz
The ripple factor, defined in Eq. (13-14), becomes
4Em 1

which may be expressed in the form

2 1
r = 3-y-2 -yJ=1C=+===;(=:=4w==:2=i=L2::=i/::::;::R~D (14-9)

If the ratio wL/R z is large, this reduces to

1 Rz
r=--- (14-10)
3V2wL
This expression shows that the filtering improves with decreased load
resistance or, correspondingly, with increased load current. At no load,
Rz = 00, and the filtering is poorest, with r = 2/3 y2 = 0.47. This is
also the result which applies when no choke is included in the circuit.
296 ELECTRON-TUBE CIRCUITS [CHAP. l4

[Compare with result with Eq. (13-21), which gives 0.482.

The difference
arises from the terms in the Fourier series that have been neglected.]
The expression also shows that large inductances are accompanied by
decreased ripple.
The d-c output voltage is given by

(14-11)

where Erma is the transformer secondary voltage measured to the center

tap. Note that under the assumptions made, viz., negligible power-
transformer leakage reactance and negligible transformer resistance, tube
resistance, and inductor resistance, the output voltage does not change
with load, with consequent perfect regulation. Because the neglected
effects are not negligible, the output voltage actually decreases with
increased current.
14-3. Capacitor Filter. Filtering is frequently effected by shunting
the load with a capacitor. During the time that the rectifier output is
increasing, the capacitor is charg-
ing to the rectifier output poten-

] c
~zl

R[

FIG. 14-4. A single-phase half-wave ca-

tial and energy is stored in the
capacitor. During the time that
the rectifier potential falls below
that of the capacitor, the capac-
itor delivers energy to the load,
pacitor filtered rectifier. thus maintaining the voltage at
a high level for a longer period
than without the capacitor. The ripple is therefore considerably
decreased. Clearly, the diode acts as a switch, permitting charge to
flow into the capacitor when the rectifier voltage exceeds the capacitor
voltage, and then acts to disconnect the power source when the potential
falls below that of the capacitor.
To examine the operation in some detail, refer to Fig. 14-4, which shows
a diagram of the circuit. The tube current during the conducting por-
tion of the cycle is
(14-12)
where
. ez eo
~z = Rz = R z (14-13)
and where
io = dqe = C dee (14-14)
dt dt
where qc is the capacitor charge. The controlling differential equation
SEC. 14-3] RECTIFIER FILTERS AND REGULATORS 297
of the charging current through the tube is then

ib = ec + C dec (14-15)
Rl dt
But .the voltage ec during the time that the tube is conducting is simply
the transformer voltage, if the tube drop is neglected. Hence the capac-
itor voltage during this portion of the cycle is sinusoidal and is
ec = e = Em sin wt
The corresponding tube current is

ib = ~7 sin wt + wCE m cos wt

This may be written in the equivalent form

ib = Em ~W2C2 + ~f sin (wt + if;) (14-16)

where
if; = tan- 1 wCR z (14-17)
A sketch of the current wave is illustrated in Fig. 14-5.

Cui-in
pOlnf
\
ec \
I \
I \
I \
I \

FIG. 14-5. The tube current and the load voltage in a single-phase half-wave capacitor
filtered rectifier.

Equation (14-16) shows that the use of large capacitors, in order to

improve the filtering, is accompanied by large tube currents. Therefore,
if a large capacitor is used for a given load in order to maintain the output
voltage more nearly constant, a very peaked current exists. In fact, for
a certain required average current demand by the load, the tube-current
pulse becomes more and more peaked as the capacitor is made . large.
This imposes serious duty conditions on the tube, since the average cur-
rent through the tube may be well within the tube rating and yet the
large peak current may injure the cathode. Vacuum diodes would not
be appreciably damaged by the high peak-current demands, since tem-
perature-saturated currents may be drawn without seriously injuring
the cathode. In the case of gas tubes, however, any attempt to draw
higher than temperature-saturated current will usually be accompanied
298 ELECTRON -TUBE CIRCUITS [CHAP. 14

by severe positive-ion bombardment, with consequent cathode disinte-

gration. It is for this reason that large capacitor input filters should
not be used with rectifiers that employ gas diodes.
When the tube stops conducting, ib = 0 and the controlling differential
equation during the nonconducting portion of the cycle is, from Eq.
(14-15),
c dec
dt
+ Rl
ec = 0 (14-18)

The solution of this differential equation is

t
ec = Ae -RIC (14-19)
This shows that the capacitor discharges exponentially through the load.
To determine the value of the constant A that appears in this expres-
sion, use is made of the fact that at the time t = t1, the cutout time,
ec = e = Em sin wt1
Combining this result with Eq. (14-19) gives
t.
A = Em sin wt1e R,C (14-20)
and Eq. (14-19) becomes
(t-t.)
ec = Em sin wt1e - R,C (14-21)
The quantity t1 that appears in this expression is known, since at t = t1
the tube current is zero. From Eq. (14-16) this requires
sin (wt1 + if;) = 0
from which it follows that
wt1 = 7r - if; = 7r - tan- 1 wCR I (14-22)
If t1 from Eq. (14-22) is substituted in Eq. (14-21), there results
(",t+.,- .. )
",CR,
(14-23)
To find the "cutin" point, it is noted that ec equals the impressed
transformer voltage e at this point. This requires
(",t2-.,-,,)
Em sin wt2 = Em sin wt 1e ",CR,
or
(",t2-"-")
sin wt2 = sin wt 1e ",CR, (14-24)
The evaluation of the cutin time t2 cannot be solved explicitly, for this is
a transcendental equation. Graphical methods can be used effectively
in this evaluation. The results are given in Fig. 14-6. Included on this
SEC. 14-3] RECTIFIER FILTERS AND REGULATORS 299
graph are a plot of Eq. (14-22) for the cutout angle and a plot of Eq.
(14-24) for the cutin angle.
The foregoing analysis permits a calculation of the d-c output potential
and ripple factor. However, such an analysis is very involved, and it is
180
0 expedient to make several reasonable
170 ~ approximations. The character of the
I\. approximations is made evident by
160
an inspection of Fig. 14-7, which
150 I" shows a trace of an oscillogram of the
f'\
140
K~ load voltage in a single-phase full-wave
~'30 capacitor filtered rectifier. The volt-
~<S' 120 I\. age curve may be approximated by
§ 110 I\. two straight-line segments, as shown
~100
rI\. in Fig. 14-8. If the total capacitor
o
"'..,3 90 :\\ discharge voltage is denoted as E r ,
.• 80 then, from the diagram, the average
~'" 10 value of the voltage is
~
~ 60
F50 I) (14-25)
os
wi2 (half' wave),
.!': 40
.,!.
.3 30
Vri
Wt2 (full wave}, V )
20
JO .v V
V
o .-l-n ~
o 10 20 30 40 50 60 10 80 90
fan- 1 wCR t
FIG. 14-6. Plot of cutin angle wt. FIG. 14-7. Oscillogram of the load
and cutout angle wtl vs. circuit voltage in a single-phase full-wave
parameters for the capacitor filter. capacitor filtered rectifier.

Also, the rms value of the triangular ripple voltage may be shown to be

(14-26)

Also, if it is assumed that the

capacitor discharge continues for
the full half cycle at a constant
rate which is equal to the average
load current I d • c , the fall in poten- FIG. 14-8. The approximate load-
tial during this half cycle is E r • voltage wave form corresponding to
That is, approximately the curves of Fig. 14-7.

E = I d- e (14-27)
r 2fC
300 ELECTRON-TUBE CIRCUITS [CHAP. 14

The ripple factor is then given by

(14-28)

1
r = --;=0--- (14-29)
4 v3jCR I
This expression shows that the ripple factor varies inversely with the
load resistance and the filter capacitance. At no load, Rl = 00, and the
ripple is zero. As Rl decreases, corresponding to increasing current, the
ripple becomes larger. Also, for given Rl, the ripple is smaller for large
capacitances. Actually, Eq. (14-29) is more nearly correctfor small values
of ripple than for the larger values, the value of ripple being generally
larger than that obtained experimentally. The results are adequate for
most purposes.
The regulation curve is obtained by combining Eqs. (14-25) and (14-27).
This yields
Id-c
E doc = E m - 4jC (14-30)

This expression represents a linearfall in potential with d-c output current.

A:= ±
Also, it shows that the simple capacitor filter will possess poor regulation

B
~_
.

e
_
~
_
unless the capacitance C is large.
Now refer to the circuit of Fig. 14-4 to
Rl ascertain the peak inverse voltage across
the tube. It is seen to be twice the trans-
former peak voltage. For the full-wave
FIG. 14-9. An L-section filter.
case, the peak inverse voltage is also twice
the transformer maximum voltage, as measured from the mid-point to
either end. Thus the presence of the capacitor increases the peak inverse
voltage in the half-wave circuit from Em to 2Em but does not affect the
relatively simple peak inverse voltage in the full-wave circuit.
14-4. L-section Filter. An L-section filter consists of a series induc-
tor and a shunt capacitor, as shown in Fig. 14-9. This filter is so arranged
that the inductor offers a high impedance to the harmonic terms, and
the capacitor shunts the load, so as to by-pass the harmonic currents.
The resulting ripple is markedly reduced over that of the relatively
simple filters of Secs. 14-2 and 14-3.
The ripple factor is readily approximated by taking for the voltage
applied to the input terminals of the filter the first two terms in the Fourier
series representation of the output voltage of the rectifier, viz.,

e = -2Em - 4Em
- cos 2wt (14-31 )
'11" 3'11"
SEC. 14-4] RECTIFIER FILTERS AND REGULATORS 301

But since the filter elements are chosen to provide a high choke impedance
and a very low shunting impedance, certain plausible approximations
may be made. Thus, since the choke impedance is high compared with
the effective parallel impedance of the capacitor and load resistor, the
net impedance between terminals AB is approximately XL and the a-c
current through the circuit is

, -'- 4Em ~ =
I rrna y2 E ~ (14-32)
3 V21l"XL 3 d·c XL

Likewise, since the a-c impedance of the capacitor is small compared with
R z, it may be assumed that all the a-c current passes through the capac-
itor and none through the resistor. The a-c voltage across the load
(the ripple voltage) is the voltage across the capacitor and is

..:.. V2 E d,cX
E 'rms-3 Xc (14- 33)
L

The ripple factor is then given by

V2 Xc V2 1 1 (14-34)
r = 3 XL = 32wC2wL
\vhich may be written, at 60 cps, with L in henrys and C in farads,
0.830
r = LC (14-35)

It should be noted that the effect of combining the decreasing ripple of the
inductor filter and the increasing ripple of the simple capacitor filter
for increasing loads is a constant ripple circuit, independent of load.
The above analysis assumes that no current cutout exists at any time
of the cycle. If it did, the analysis would follow along the lines of Sec.
14-3 and Eq. (14-31) for the potential would not apply. But since with
no inductance in the filter cutout will occur, whereas with sufficient
inductance there will be no cutout, it would be expected that there would
be some minimum inductance for a given current below which cutout
would occur, although for larger values than this critical value the con-
duction would continue for the full cycle. The situation is best illus-
trated graphically. Figure 14-10 shows the tube current for various
tcmounts of series inductance L.
If the rectifier is to pass current throughout the entire cycle, the peak
current delivered must not exceed the d-c component. But the d-c value
is Ed.c/R z• Also, the peak a-c current is (2Ed.c/3) (1/XL). Hence for
current flow during the full cycle it is necessary that
1-
E d-c > 2Ed-c
Rl - 3 Xl"
302 ELECTRON-TUBE CIRCUITS [CHAP. 14

or
x > 2Rl
L 3
-
(14-36)

from which the value for the critical inductance is found to be

2Rl
Lc=
3w
which has the value
(14-37)

(a) I
I
I
I

FIG. 14-10. The tube current in the full-wave rectifier with L-section filter, for
(a) L = 0, (b) L less than critical inductance Le, (e) L equal to L e , (d) L greater than L,.

for a 60-cps power frequency, where Rl is in ohms and Lc is in henrys.

However, owing to the approximations that have been made in this
analysis, it is advisable for conservative design to increase Lc above that
given in Eq. (14-37). A good practical figure is to choose the denomina-
tor as 1,000 instead of 1,130.
The effect of the cutout is illustrated in Fig. 14-11, which shows a regu-
lation curve of the system, for constant Rl and varying series inductance.
Clearly, when the series inductance is zero, the filter· is of the simple
capacitance type and the output voltage is approximately Em. With
SEC. 14-4] RECTIFIER FILT ERS AND kEG ULATORS 303
increasing inductance, the voltage falls, until at L = Lc the output poten-
tial is that corresponding to the simple L filter with no cutout, or 0.637 Em.·
For values of L greater than L c,
there is no change in potential, ex- 1.0
cept for the effects of the resistances
of the various elements of the circuit. Edc
It is not possible to satisfy the Em 0.5 I1
conditions of Eq. (14-37) for all val- I
1Lc
ues of load, since at no load this kX~--------------~L

would require an infinite inductance. FIG. 14-11. The regulation curve of a

However, when good voltage reg- rectifier with L-section filter as a func-
tion of series inductance, for constant
ulation is desired, it is customary output current.
to use a bleeder resistance across
the load so as to maintain the conditions of Eq. (14-37) even if the use-
ful load resistance is high.
A more efficient method than using a high bleeder current, with its
consequent power dissipation, is to make use of the fact that the induct-
ance of an iron-core reactor de-
pends, among other things, on the
amount of d-c current in the wind-
ing. Chokes for which the in-
ductance is high at low values of
d-c current and which decrease
markedly with increased d-c cur-
rents are called "swinging"
20 40 60 80 100 120 (rna) Ide chokes. A typical curve for such
FIG. 14-12. The inductance of a swinging a reactor is illustrated in Fig.
choke as a function of the d-c current
through it.
14-12. The advantage of such a
choke is that for high R l , and
therefore low d-c current, the inductance is high. As a result, the con-
ditional equation (14-37) is satisfied over a wider range of R l • Clearly,
however, when a swinging choke is used, the ripple factor is no longer inde-
pendent of the load.
The above analysis for the critical inductance of the L-type filter applietl
for the full-wave rectifier for which conduction continues for 180 deg in
each cycle. Consequently the results so obtained are not applicable
when an L-section filter is used with a controlled rectifier. The analysis
for a full-wave controlled rectifier is considerably more complicated than
that above, owing to the fact that the amplitude of the harmonics in the
Fourier series representation of the output depends on the delay angle,
and these are of such amplitude that they cannot be neglected in the
analysis. The results of such an analysis are given graphicallyl in
304 ELECTRON -TUBE CIRCUITS [CHAP. 14

Fig. 14-13. The curves give a measure of both the critical inductance
and the output voltage.
14-5. Multiple L-section Filters. If it is desired to limit the ripple to a
value that is less than that possible with a singleL-sectionfilter using com-
mercially available elements, two
10000

li,
or more L-section filters may be
5000
2000
V connected in cascade, as shown in
1000 / Fig. 14-14. An approximate solu-
,/~
/"
300 tion is possible by following the
'"' 200 ,...- , . /
~
,./ methods of Sec. 14-4. It is as-
" IDa
- I.0 sumed, therefore, that the choke
~ 50 o.
o. impedances are much larger than
~" 20 r;?,t'. >-. o.
the reactances of the capacitors.
Em": "-
I\~
10 Also, it is assumed that the re-
~ actance of the last capacitor is
3
2 o.m small compared with the resistance
I ODI
o 10 20 30 40 50 60 10 80 90 of the load. Under these assump-
De/"y "ngle. cleq.
tions, the impedance between As
FIG. 14-13. Critical inductance and d-c and Bs is X c2 • The impedance
output voltage as a function of the delay
angle in a full-wave controlled rectifier.
between A2 and B2 is X c1 , and the
impedance between Al and Bl is
X £1, approximately.
The a-c current II is approximately
V2 1

::t f~ f~'
11 =3 E d - e XLI

The a-c potential across C 1 is ap-

proximately
E A ,B = I 1X e1
2
B2 Ba
,R'
FIG. 14-14. A two-unit L-section filter.
The a-c current 12 is approximately

The a-c voltage across the load is approximately

1 2XC2 = IIX c2X

XCI v2 XCI XC2
= -3 Ed-eX LI -X
Ll L2

Thp. ripple factor is given by the expression

V2 XCI X C2 (14-38)
r = 3 X LI X L2
SEC. 14-6] RECTIFIER FILTERS AND REGULATORS 305

A comparison of this expression with Eq. (14-34) indicates the general-

ization that should be made in obtaining an expression for the ripple of
a cascaded filter of n sections. The expression would have the form

r=v'2
--X C1 X C2
-- - · · · XXCn
- (14-39)
3 X X
L1 L2 Ln

If·the sections are all similar, then Eq. (14-39) becomes

r = ~2 (~:) n = V;2 (167r 2J2Lc)n (14-40)

It follows from this that the required LC product for a specified ripple r
is given by
LC = 1.76 (0.!71)~ (14-41)

Note also that, to the approximation that the impedance between Az

and Bz is simply X c1 , the critical inductance is given by Eq. (14-37), as
for the single-section unit.
14-6. IT-section Filter. The use of a IT-section filter provides an output
potential that approaches the peak value of the a-c potential of the source,
the ripple components being very small. Such a filter is illustrated in
Fig. 14-15. Although such filters do provide a higher d-c output poten-
tial than is possible with an L-section
filter, the tube currents are peaked and
the regulation is generally poor, these re-
sults being common with the simple ca-
pacitor filter.
A study of the oscilloscope patterns at FIG. 14-15. A II-section filter.
various points of such a filter shows that
the action can be understood by considering the inductor and the
second capacitor as an L-section filter that acts on the triangular output
voltage wave from the first capacitor. The output potential is then
approximately that from the input capacitor, the ripple contained in this
output being reduced by the L-section filter. That is, the ripple factor of
the IT-section filter is given approximately by
(14-42)
where rc is given by Eq. (14-29) and rL is given by Eq. (14-34). This
result is only approximate, since it assumes in effect that the ripple out-
put from the capacitor filter is sinusoidal r~}her than triangular. A more
careful calculation shows that Eq. (14~4) is slightly higher than that
obtained by the complete calculation and so yields generally conservative
results.
306 ELECTRON-TUBE CIRCUITS [CHAP. 14

14-7. Glow-tube Regulator. The use of an electronic rectifier with

an appropriate filter serves to provide a low-ripple source of d-c potential,
the percentage of ripple present in the output depending upon the form
of the filter that is used. Such rectifier systems, while generally satis-
factory for many purposes, possess several shortcomings which may make
them inadequate for certain services. The output voltage depends criti-
cally upon the input voltage to the rectifier, and a poorly regulated power
system will be accompanied by a corresponding change in the output from
the rectifier. Also, since the output
impedance of the rectifier is usually
I ijl; 1 quite high, the rectifier system will
Unregulated I possess a poor regulation.
supply
vo!tage
Rr .!fo
) It is frequently necessary to con-
--,~,--____________1._·2~:..J-.l. struct a power supply the output po-
FIG. 14-16. Glow-tube voltage regu- tential of which is constant over wide
lator. ranges of input a-c voltage, so as to
provide a constant output voltage
source from a poorly regulated power line. Or it may be necessary to
maintain a constant output voltage for a varying output load. Elec-
tronic voltage regulators provide such a control device and are exten-
sively used for such service.
The simplest form of voltage regulator makes use of the substantially
constant voltage characteristic of a glow tube. A glow tube is a cold-
cathode discharge tube which is characterized by a fairly high tube drop
and a low current-carrying capacity (see Sec. 2-15). The voltage across
the tube over the operating range is fairly constant and independent of
the current. When connected in the circuit shown, the voltage across
the load will be a constant and equal to the tube drop of the glow tube,
over a range of currents. Specifically, if a VR-150/30 is used, the voltage
across the load will be approximately 150 volts provided that the current
through the tube does not exceed the
rated 30 ma of the tube.
If a voltage is desired that is higher
eh
1------:
I
B

I
than that of a single glow tube, several I I
I I
tubes may be connected in series. This I I
I I
will provide a constant voltage source
that is the sum of the tube drops of the 5 (mi
40 b

tubes that are used. For example, the FIG. 14-17. A typical glow-tube
volt-ampere characteristic.
use of a VR-150 and a VR-105 in series
will provide a constant 255-volt source. The supply voltage must be
greater than the breakdown voltage of the tubes in order to make opera-
tion possible. The difference between the supply voltage and the oper-
ating tube voltage drop will appear across the stabilizing resistor R 8•
SEC. 14-7] RECTIFIER FILTERS AND REGULATORS 307
An analysis of such a circuit is readily possible 2 if use is made of the
practical fact that over the range of operation the volt-ampere charac-
teristic of the regulator tube is linear. A typical characteristic has the
form shown in Fig. 14-17. This characteristic may be expressed by an
equation of the form
e = ail + b (14-43)
where and (14-44)

From the circuit diagram of Fig. 14-16 it follows that

Ei = (i l +i 2 )R 8 + Eb (14-45)
Also,
+b

l
Eb = Eo = Rd2 = ail (14-46)
from which it follows that
. =Eo
'h --- -b
a
and (14-47)
. Eo
'/.2 = -
Rz
Combining these equations yields

Ei = (Eo;; b +~;)R8 + Eo
from which it follows that
E _ aR1Ei + bR1R. (14-48)
o - R1Rs + aRs + aR I
This is the expression for the regulated voltage as a function of the supply-
voltage and circuit parameters.
In order to obtain an expression for the variation of the output voltage
as the input voltage varies, Eq. (14-48) is differentiated. This gives for
the ratio dEi/dE o, which is known as the stabilization ratio, the expression
G == dEi = R1(R. + a) + aRB (14-49)
dE o aR I
Combining this with the expression for Eo gives
G = aEi + bRa (14-50)
aEo
This expression shows that for perfect regulation, i.e., infinite stabiliza-
tion ratio, the fraction dEi/dE o should be zero. For best stabilization
features, both Eo and a should be small; and b, E i, and R. should be large.
Using typical values with a VR-75 tube, one finds
Ei = 250 volts. R = 00 R. = 32.5k fl.Ei = 20 fl.Eo = 0.15
= 31< = 5.851< = 20 = 0.60
308 ELECTRON-TUBE CIRCUITS [CHAP. 14

Such simple gas-tube regulators operate quite satisfactorily but are

seriously limited in their usefulness owing to their limited flexibility, both
because of their fixed potential ratings
T
and the relatively low current-carrying
+t capacity.
I
I
I 14-8. Simple Vacuum-tube Regu-
load Eo
I
I
lator. One may employ the variable
I
~ __ ~~~~ ________~--L
I beam-resistance characteristic of a
vacuum tube to maintain the output
FIG. 14-18. A simple vacuum-tube
voltage regulator. voltage from a power supply at a sub-
stantially constant level. The beam
resistance, as previously defined, is rb = E b/ h and is quite different from
the plate resistance rp of the tube. The circuit of such a simple voltage
regulator is given in Fig. 14-18.
Assume that the voltage across the load is at the desired value. Under
this condition, the cathode is positive relative to ground by a voltage Eo.
The grid may be made positive relative to ground by a voltage Eg which
is less than Eo. The potentiometer R2 is adjusted until the bias on the
tube is such that the tube will pass load current. With this bias, the
resistance of T is established at the desired value to reduce the rectifier
output to the desired load voltage.
If the rectifier output voltage
increases for 'whatever reason, the
voltage at the cathode of T tends
to increase. As Eo increases, the
bias on the tube increases and the
effective beam resistance of the
FIG. 14-19. A simple voltage-regulator
tu be b ecomes grea t er. C onse- circuit.
quently the voltage drop across
the tube becomes greater. If the circuit is properly designed, the
increased voltage across T is approximately equal to the increase of volt-
age and the output potential remains substantially constant.
The practical form of the circuit will replace the battery by a glow tube.
Such a circuit is shown in Fig. 14-19.
The output voltage from this regulator is not absolutely constant, since,
for an increased input to the circuit, the voltage at the cathode of T must
rise slightly if the regulator is to function. However, if the characteristics
of tube T are carefully chosen, the rise of load voltage is not large.
It would be expected, of course, that vacuum tubes in which the beam
resistance varied rapidly with small changes in bias would be most desir-
able for service in such regulators. Tubes possessing such a characteristic
eould probably be designed if there were no alternative approaeh. Aetu-
SEC. 14-9] RECTIFIER FILTERS AND REGULATORS 309

ally such special tubes are not necessary, as it is quite possible to achieve
the same ends by including a d-c amplifier in the circuit in such a way
that slight changes in output potential are amplified before being applied
degeneratively to T. These circuits are used extensively and a detailed
analysis will be given below.
14-9. Electronic Voltage Regulators-Basic Considerations. The
design of electronically regulated power supplies has become fairly well
standardized. The elements of such a circuit are given in Fig. 14-20.
As discussed above, the operation of the circuit is essentially the fol-
lowing; The source of unregulated current from the rectifier and filter is
applied across the input terminals of the regulator. The unregulated d-c
current is fed through the series control tube, now designated T2 in the
diagram, to the output circuit. T2
If the current requirements are
too high for a single tube, a num-
ber of such tubes may be con-
nected in parallel. A power tri-
ode is frequently used for this
purpose, although a pentode or
beam tetrode connected as a triode
will serve equally well. Popular FIG. 14-20. A basic electronic regulator
circuit.
series control tubes include
6AS7G, 6B4, 6L6, 6V6, and 6 Y6G. These tubes will pass approximately
75 ma without seriously exceeding the plate dissipation of the tube. The
regulating action is obtained by comparing a fixed fraction of the output
voltage with a standard voltage source, such as a battery or a VR gas
tube. Any difference between the two is applied degeneratively after
amplification by a high-gain d-c amplifier to the control grid of the series
current-control t;ube. The correction may be made nearly perfect by
using a d-c amplifier of sufficient gain.
The design of a voltage regulator requires a knowledge of the charac-
teristic curves of the control tube and also of the d-c amplifier tube.
However, since small changes in voltage and current are ordinarily
involved, the circuit operation may be analyzed in terms of the slope of
the tube characteristics-specifically, in terms of the mutual conductance
and the internal resistance of the tubes at the operating points. Several
cases of interest will be examined separately.
a. Varying Input Voltage-R3 Connected to the Input Side of the Regu-
lator. Suppose that the output load remains constant but that the input
voltage to the regulator varies, either because of the poor regulation of
the input a-c supply voltage to the rectifier or because of the ripple in
the output of the rectifier due to inadequate filtering. It follows from
310 ELECTRON -TUBE CIRCUITS [CHAP. 14

an inspection of the circuit of Fig. 14-20 that the input voltage to the d-c
amplifier is related to the output voltage by the relation
R2
Eg = Rl + R2 Eo = (3Eo (14-51)

This assumes that the current to the input of the d-c amplifier is negligible.
This is a valid assumption, as the grid of the d-c amplifier is ordinarily
maintained negative with respect to the cathode by an amount that will
allow approximately normal tube current in the d-c amplifier.
Suppose now that owing to a change t.Ei in the input a small change
t.Eo appears in the output voltage. The corresponding change in the
voltage appearing at the input to the d-c amplifier is
(14-52)
Let Kl denote the gain of the d-c amplifier. The voltage appearing across
the plate resistor R s, which is also the plate-grid circuit of T2, may be
calculated from the equivalent circuit shown and
IS

R3
where (14-53)

However, when Rs is connected to the input side of the regulator, a

second voltage appears in the circuit across Rs. This arises from the
appearance of t.Ei across Rs and rr- (the tube). This voltage is evidently

where
t.E 2 = -+
rp
Rs E
R-s t. i

t.E 2 = (1 - 'Y)t.E i

'Y =
rp
-~-=-
l (14-54)
,tJEo
rp + Rs I
I
I
I
The total voltage across Rs is I

The grid-cathode voltage of T2 is then

t.E gk = t.E pk - t.Epg
= t.E pk + (3K 1 t:,.Eo - (1 - 'Y) t.Ei (14-55)
But by definition
SEC. 14-9) RECTIFIER FILTERS AND REGULATORS 311

Then
flEb = -,u2[flEb + {3Kl flEo - (1 - 'Y) flEil (14-56)
It follows from this that
flEb(l + ,(2) = -{3K 1,u2 flEo + (1 - r),u2 flEi
But as

then

The voltage-stabilization ratio G is given by

G = flEi = 1 + ,u2 - f3K ,u2
1 (14-58)
flEa 1 + "1M2
which approximates. under normal conditions, to
G == -f3K 1M2 (14-59)
1 + "1M2
The quantity G gives a measure of the effectiveness with which the
regulator compensates for changes in voltage in the input. For a regu-
lator with a single d-c amplifier stage, G may be of the order of 300 to
1,000. If an improved value for G is required in order to achieve an
almost ripple-free output, it is necessary that the gain of the d-c amplifier
be increased. This is most easily done by adding a second stage of d-c
amplification. Such circuits will be considered below. With such cir-
cuits, a value of G of 25,000 is possible.
b. Varying Input Voltage-Ra Connected to the Output Side of the
Regulator. Somewhat improved results are theoretically possible if the
plate resistor Ra of the d-c amplifier stage Tl is connected to the output
side of the regulator instead of the input side. That an improvement is
possible follows from the fact that one may now assume that the plate
voltage to which the amplifier T1 is connected is substantially constant
and that the output from this stage, viz.,
flE = -f3K 1 flEo
is the cathode-grid voltage applied to the series control tube T2. Conse-
quently the net change in voltage appearing in-the output is approximately
flEb = flEi - flEo = - ,u2(f3K 1 flEo)
and the voltage-stabilization ratio is now given by the relation
G = 1 - ,u2f3K 1 == =,u2f3Kl (14-60)
Owing to the denominator that appears in Eq. (14-59), the value of G
appears to be higher in Eq. (14-60). Actually, however, the d-c amplifier
312 ELECTRON -TUBE CIRCUITS [CHAP. 14

T1 operates in a more linear manner under condition a, with somewhat

higher gain K 1. Under normal circumstances the two connections yield
about equal results, and both are used.
c. Varying Load. Suppose now that the change in output voltage
results from a change in the load current because of the internal resistance
of the supply. If this change in voltage is again denoted by AE o, then,
as before, the voltage appearing at the grid of the current control tube is
AE = -{3K1 AEo

If gm denotes the mutual transconductance of the control tube, the result-

ing change in current through this tube due to a change in potential AE2
at the grid is

which is
(14-61)
The ratio of the change in output voltage to the change in output current
is denoted by Ro and is the effective internal resistance of the regulated
power supply. This is given by
Ro ~ AEo = 1 (14-62)
Alp -g",{3K 1

In a typical case, the effective internal resistance of the regulated supply

can be as low as 0.5 ohm.
This calculation assumes, of course, that there is no change in the input
applied potential with changes in output current or, equivalently, that
the internal impedance of the unregulated power supply is low. This
condition is not often met in practice, and due account must be taken of
this factor. If the internal resistance of the unregulated driving source
is denoted by r, then by Eqs. (14-60) and (14-62) the total effective internal
resistance of the regulated power supply is given by

R = 1_...
__
-gm{3K 1
+ !. =
G
1
-gm{3K 1
(1 +-.'1"...)
rp2
(14-63)

For a typical case r = 500 ohms, and G = 1,000, so that the added resist-
ance due to the regulation of the input source may be only about 0.5 ohm.
14-10. Design Considerations. If AEo denotes the change in output
voltage, without regard to its cause, the change in current through the
control tube is again given by Eq. (14-61). If the total resistance acrosS
the output side of the control circuit is denoted by R z, the net change in
output voltage is given by
(gmf3K1Rz - 1) AEo lunreg = AEo Ireg (14-64)
When voltage regulation occurs, the two quantities on the left just balanee
SEC. 14-10] RECTIFIER FILTERS AND REGULATORS 313

each other and the initial change in voltage across the load is exactly
compensated by the change in plate current. If the difference is positive,
the initial change in output vdtage is overcompensated; if the expression
is negative, the circuit is under compensated.
In view of the several adjustable factors that appear in this expression,
they may ordinarily be so proportioned that the expression on the left
becomes zero. The practical problem in the design of the circuit becomes
that of determining the amplification factor }1, which appears in the
expression for the gain K 1, and in the determination of gm of tube T2.
After this, the resistance values R 1, R 2 , and Ra may be chosen to satisfy
the relation
(14-65)

The value of }1 and gm may be determined from the characteristic curves

of the appropriate tubes when the applied grid and plate potentials are
known. In the case of the d-c amplifier tube, the value of }1 is that at
which the grid potential is
Eo = «(jED - E.)
and the plate potential is
Eb = E. - E. - JpRa
where J p is the average tube current.
The value of the mutual conductance gm of the control tube T2 is
obtained in the same general way from the appropriate set of static-
6Y6G
I
+
I I
I I
I I
I
I
1 6SJ7 :
Unr~gulafed Reg;/afed
mput output
450 V .lOp'"
I
I
I
I
I

FIG. 14-21. An electronically regulated power supply.

characteristic curves of the tube used, where the total applied voltage is
Ei - Eo and the applied grid voltage is Ei - Eo - JpR a, which depends
on the operation of the d-c amplifier tube.
A typical circuit that yields satisf~ctory results over a wide range of
input voltage and over a wide range of load current is given in Fig. 14-21.
Although the diagram of Fig. 14-20 shows a triode as the d-c amplifier,
it is found more desirable to use a pentode in this position and this has
314 ELECTRON -TUBE CIRC UITS [CHAP. It

been done in the circuit of Fig. 14-21. The reason for this is that it
frequently happens that the d-c regulation is not as good as one would
expect when a triode is used, primarily because the grid impedance of a
high-gain triode is quite low. The grid impedance should be high, espe-
cially if the full gain capabilities are to be realized. A 6SJ7 tube it-;

,
,
I

I
I
I

O~fpuf
,,
,,
:

FIG. 14-22. An electronic voltage regulator employing a two-stage d-c amplifier.

superior in this respect and hence is frequently used. The 6Y6G pentode
called for in Fig. 14-21 gives satisfactory results. The newer 6AS7G
tube possesses some advantages over this, sin~e it has a relatively high
tube current rating (125 rna) with a relatively low tube drop. Also, the

,I
I
I
I

/6",1",'f

O~fPuf
,
:
I
I
,I
I

FIG. 14-23. A voltage regulator employing a· difference amplifier.

heater-cathode insulation is sufficiently good to avoid the need for a

separate filament heating transformer for this tube. The tube does have
a rather low value of J.I, (= 2.1) which requires a rather large control
voltage.
If a voltage regulator is required which is to provide a practically
ripple-free output and an almost perfect regulation, it is necessary that
the gain of the d-c amplifier be increased. This is most easily done by
adding a second stage of d-c amplification to the regulator circuit. A
variety of such circuits is possible, and several types are illustrated below.
Figure 14-22 shows a simple two-stage resistance-coupled amplifier;
Fig. 14-23 utilizes a difference amplifier; and Fig. 14-24 uses what is
called a "cascode" amplifier.
SEC. 14-11] RECTIFIER FILTERS AND REGULATORS 315

II I
I
I I
I
I I
I
I
I I
I
I I
I
lnpuf I
I
I
Olifplif
I
I II
I
I I
I I
I I
I I
I I

FIG. 14-24. A voltage regulator employing a "cascode" amplifier.

14-11. Special Precautions. 3 Although the principles of operation of

the regulated power supply are straightforward, it frequently happens
that the maximum performance will not be realized in practice. It is
well to consider some of the reasons for this. The important factors to
be examined closely are the degenerative d-c amplifier loop and the vari-
ous sources of hum.
a. The Degenerative D-C Amplifier Loop. For satisfactory operation,
the d-c amplifier must be degenerative at all frequencies for which the
loop gain is greater than unity. If this condition is not met, the system
will oscillate (refer to Sec. 5-9) and the regulation properties will be greatly
affected. One of the best ways to ensure that the power supply will not
break into oscillation is to limit the h-f response of the amplifier. This is
best done by including a large capacitor across the output (8 }J.f or 16 Mf).
This will usually provide the necessary h-f cutoff and will still keep the
power-supply impedance low at high frequencies. The O.l-}J.f capacitor
from the grid of Tl to B+ serves to prevent a phase lag at the grid of the
d-c amplifier and also compensates somewhat for additional phase shift
in the amplifier and increases Kl for frequencies above a cycle or two.
Sources of Hum. Among the sources of hum which give rise to a
higher ripple in the output than is expected, and their possible cures, are
the following:
1. Ripple from a-c heated filaments in the d-c amplifier. By grounding
the center tap of the heater transformer and by choosing tubes with low
heater-hum characteristics, the hum in the output voltage can be reduced
to 4 or 5 mv rms or less.
2. Ripple from common leads. This may arise from coupling between
the d-c supply and some a-c source, such as a filament supply. The use
of the chassis as a common ground ,,·ith grounds to various parts of the
chassis may introduce this hum voltage. This effect is ordinarily small,
perhaps several millivolts rms, except when the common coupling appears
in the input of the d-c amplifier in the regulator, in which case it may be
316 ELECTRON -TUBE CIRC UITS [CHAP. 14

appreciable. To avoid this difficulty, grounds should be separately

returned to a single point.
3. Ripple from supply voltages. The screen voltage to the d-c ampli-
fier must be ripple-free. This may require a filter at the screen terminal
at the tube base.
4. Ripple in the comparison voltage source. It might be necessary to
include a filter in the CR circuit for this purpose, in addition to the
O.OOOl-j.!f condenser shown (which is to prevent any effects that might
arise from the h-f plasma oscillations in the VR tube).
5. Induction loops. If coupling occurs between circuits by electro-
static or electromagnetic induction, it may be necessary to include a
simple RC filter in the input circuit to the d-c amplifier and in the com-
parison-voltage circuit.
Heater Supply. When the heater of the d-c amplifier is fed from an
unregulated source, changes in output with heater-voltage changes may
be quite noticeable. This can be eliminated by operating the heaters
from the regulated d-c supply.
REFERENCES
1. Overbeck, W. P., Proc. IRE, 27, 655 (1939).
2. Berg, W. R., Electronics, 20, 136 (October, 1947).
3. Lawson, J. L., M.I.T. Radiation Lab. Internal Rept., unpublished.
As general references, see
Hunt, F. V., and R. W. Hickman, Rev. Sci. Instruments, 10, 6 (1939).
Mautner, L., Elec. Eng., 66, 894 (1947).
Millman, J., and S. Seely, "Electronics," Chap. XIV, McGraw-Hill Book Com-
pany, Inc., New York, 1941.
PROBLEMS
14-1. It is planned to use a type 83 gas tube in a single-phase full-wave rectifier
circuit with capacitor filter. The transformer voltage is 350 volts rms to center
tap. The load consists of a 16-j.!f condenser in parallel with a 2,500-ohm resist-
ance. The tube drop and the transformer resistance and leakage reactance may
be neglected.
a. Calculate the cutout angle.
b. Determine the cutin point.
c. Calculate the peak tube current. Should the 83 tube be used? Compare
the peak current per plate with that given (1 amp) in the tube manual.
14-2. Given two 20-henry chokes, and two 16-j.!f capacitors. Calculate the
output voltage and ripple factors under the following conditions:
a. The two chokes are connected in series with the load.
b. The two capacitors are connected across the load.
c. A single L filter, consisting of the two chokes in series and the two capacitors
in parallel, is used.
d. A double L filter, consisting of two sections, each of one choke and one
capacitor, is used.
 RECTIFIER FILTERS AND REGULATORS 317

The load is 2,000 ohms, and a 375-0-375 transformer is used in a full-wave

circuit. Assume a 25-volt drop occurs in the tubes.
14-3. A power supply has the form shown in the diagram.
a. Determine the approximate secondary voltage of the power transformer.

]
b. What would be the ripple voltage if the power frequency is 60 cps; 400 cps?
14-4. In the power supply shown in the figure,

tlOOml?f

a. What is the output d-c voltage?

b. What is the ripple voltage in the output?
c. What is the minimum load current below which current cutout in the filter
occurs? What is the corresponding load voltage?
Note: Make allowance for tube drop, but assume a perfect transformer.
14-5. A typical circuit for the high-voltage supply for a cathode-ray tube is
shown in the diagram. Estimate the output ripple voltage.

] a25
20'00 v
0.25 2SC?pa

14-6. The circuit used (see the figure) is to supply two different voltages. If
the transformer is 375-0-375, what are the output voltages?
318 ELECTRON -TUBE CIRCUITS [CHAP. 14

14-7. Given the electronically regulated power supply shown in the diagram.
a. What must be the gain of the d-c amplifier to reduce the output ripple to
8 mv?
2volls
ripple IOOma
Recf/fi"er, --
and ~ :
f'//ler '500.n. :
' - -_ _ _. ...J 420

b. What must be the gain of the d-c amplifier to reduce the output impedance
to 1 ohm?
14-8. The addition of a resistor R to the electronic voltage regulator, as shown
in the figure, improves the voltage-stabilization ratio. Express the required

I
I
,,
:
I
I
,
I
I I
lnpuf Oufpuf
I
I
I
:
I

.
I
I
I
I
,
I
I

value of R in terms of the circuit parameters in order to achieve perfect

stabilization.
14-9. Given the basic cascode circuit shown in the diagram, show that the
total current through the tubes is
• Ji2Cf.!.! + 1)e o + f.!.!eol
2
~ = rpl + (f.!.l + 1)rp2 + Rz
Show that the over-all gain of this amplifier, when the grids are tied together

in so far as the a-c signal is concerned, eol = eo2 = eo> is

K _ + 2)R z
f.!.(f.!.
- (f.!. + 2)rp + Rz
 RECTIFIER FILTERS AND REGULATORS 319

14-10. ,Calculate the 'ripple voltage, and the per cent regulation in the regula-
tors of Figs. 14-21 to 14-24. A load of 75 ma at 300 volts is supplied. The
rectifier and filter system supplies an input voltage of 460 volts to the regulator.
A II filter, consisting of two 8-p,f capacitors and one 12-henry choke, is used.
The effective output resistance of the filter is 650 ohms.
14-11. Given the voltage regulator shown in the figure.

I
I
I
I
I
I
I
I
I
I
300
II
I
I
I
I
I
•
I
I

a. Specify the values of all circuit elements in the circuit, and give reasons
for your choice. The load current is 75 mao
b. Indicate on the diagram the d-c potentials at all points in the circuit.
C. There is a 3-volt rms ripple at the output of the rectifier and filter. The
effective internal resistance of the source is 530 ohms. Calculate (1) the
voltage stabilization ratio of the regulator; (2) the output ripple voltage;
(3) the output resistance of the system.
14-12. Repeat Prob. 14-11 for the electronic voltage regulator shown in the
accompanying diagram. Note that from practical considerations, this is not a
desirable circuit.

,
I
I
I
I
,,
I

,
I
I R3 :
I
450 JOO
I
I
,
!,
I
I
.I
I

14-13. Design a regulated power supply that will supply 150 ma at 250 volts
from a 450-volt rectified and filtered source, with a ripple not to exceed 0.1 per
cent and with a stabilization ratio of 15,000. What is the rms ripple voltage in
the output of this regulated source if a II filter consisting of two 8-JLf capacitors
and a 12-henry choke is used? The effective output resistance is 650 ohms.
 CHAPTER 15

AMPLITUDE MODULATION

T HERE are two major reasons for transmitting intelligenceatarelatively

high frequency level, (1) transmission by radiation is practicable at the
high frequencies, and (2) it is possible to transmit a number of messages
simultaneously without interference if the frequency level is different for
each message. Of course, any complete system of conveying intelligence
from one point to another must be capable of reproducmg the intelligence,
as represented by the amplitude or loudness, and the frequency, after
transmission. The process of altering the frequency level of the intelli-
gence is broadly known as modulation. The inverse process, in which
the intelligence is extracted from the radiated wave, is known as demodu-
lation or detection.
By definition 1 modulation is the process of producing a wave some
characteristic of which varies as a function of the instantaneous value of
another wave called the modulating wave. The modulating wave is
usually the signal, the modulated wave being the h-f carrier wave, which
has been altered in a manner to carry the intelligence.
Consider a wave which may be represented analytically by the expres-
sion
e = A cos (wt + (j) (15-1)

where t is the time. If either A, w, or (j is varied according to some func-

tion of the instantaneous value of a modulating wave, then this expression
will represent the modulated wave. It is possible, in fact, to produce a
wave in which all three parameters vary simultaneously. However, in
each of the modulating methods that are important practically, only one
of these parameters is varied, and in commercial transmitters great care
is taken to avoid the use of more than one type of modulation.
In amplitude modulation (a-m), the amplitude A is varied in accordance
with the modulating wave, while wand (j remain constant. In frequency
modulation (f-m) , the frequency w is varied, and both A and (j remain
constant. In phase modulation (p-m), the phase (j is varied, while A and
w remain constant. It should perhaps be mentioned that phase modula-
tion is not of much practical importance in itself, but as will be shown
320
SEC. 15-1] AMPLITUDE MODULATION 321
later, it may be used as an intermediate step in achieving frequency
modulation.
15-1. Characteristics of Amplitude Modulation. As indicated, ampli-
tude modulation is produced by varying the magnitude of the carrier in
accordance with the amplitude and frequency of the modulating source.
Let the signal voltage be designated as
(15-2)
and let the unmodulated carrier be written as
ee = Ee cos (Wet + (J) (15-3)
The carrier frequency We is usually much greater than the signal frequency
Wm and is chosen at the designated frequency level desired for the trans-
mission. The resulting modulated wave has the form
(15-4)
The amplitude factor Ee + kaEm cos Wmt expresses the sinusoidal varia-
tion of the amplitude of the wave, where the proportionality factor ka
determines the maximum variation in amplitude for a given modulating
signal Em. In this expression the arbitrary constant phase (J has been
chosen as zero, since it plays no part in the modulating process.
In examining this wave in detail, the expression is written in the form
e = Ee(1 + ma cos wmt) cos wet (15-5)
which is then expanded to the form

e = E e cos wet + maEe ( ) maEe (

-2~ cos We + Wm t + -2~ cos We ~ wm)t (15-6)

The factor ma is known as the modulation index

kaEm
ma=-~ (15-7)
Ee
and lOOm a is the percentage modulation. A sketch of Eq. (15-5) has the
form shown in Fig. 15-1.

FIG. 15-1. The modulating signal and a modulated carrier.

322 ELECTRON-TUBE CIRCUITS [CHAP. 15

The expanded expression of Eq. (15-6) indicates the frequency spectrum

of the modulated \vave. The first term is of carrier frequency We; the
second has the frequency We + Wm and is called the upper side band. Its
frequency is equal to the sum of the carrier and the signal frequencies.
The third term has the frequency
Relative We - W m , which is equal to the dif-
ampldude Carner Upper
Lower T side ference between. the carrier and
side , ; ---Lband
band "': IJ1 /2 the signal frequency. This com-
,a '" ponent is known as the lower side
"'m band. A plot of the frequency
spectrum of the modulated wave
2 F requency spectrum 0 f a smu- is illustrated in Fig. 15-2.
F IG. 15-. ·
soidally modulated wave. It should be emphasized that
the foregoing is not a mathemat-
ical fiction, as it is possible by means of appropriate filters to extract the
frequencies in the spectrum. In fact, the features of transmission of
intelligence with one or more of the frequencies in the spectrum sup-
pressed will be examined below.
A simple sinor representation of Eq. (15-6) by means of a pair of
rotating conjugate sinors in the complex plane is readily possible. A
little thought will convince the reader that the three sinors in Fig. 15-3
represent the three terms of the
equation, and that the resultant
sinor does exhibit the properties
of the a-m wave.
In general, the modulating sig-
nal is not sinusoidal but is a com-
plex wave. Since, however, this
complex wave may be represented FIG. 15-3. The sinor representation of an
a-m wave.
by a Fourier series if the wave is
periodic or by a Fourier integral if it is nonperiodic, the modulated car-
rier wave possesses a frequency spectrum which is more complex than
that illustrated. But each frequency in the modulating signal produces
a pair of side frequencies in the frequency spectrum. Then a signal
with frequencies in the band g(w) will yield a frequency spectrum with a
band of side bands Y2g(wc + w) and Y2g(w e - w), symmetrically disposed
about the carrier We. Such a spectrum would have the form illustrated
in Fig. 15-4.
The frequency-shifting property of modulation is not limited to shifting
an a-f wave to a higher position in the frequency spectrum. It is possible
to shift an h-f signal up or down in the frequency scale, and both processes
are important. Since such frequency shifting or frequency changing does
SEC. 15-2] AMPLITUDE MODULATION 323
not involve directly the intelligence to be transmitted, frequency changing
is classed, not as a modulation process, but rather as a detection process.

RelotHve
response

FIG. 15-4. The frequency spectrum of a complex wave.

15-2. Square-law or Small-signal Modulation. Amplitude modula-

tion may be produced by impressing two sinusoidal voltages of different
frequencies in a nonlinear circuit. The van der Bijl modulator is one
of the earliest methods, although it is seldom used at the present time.
This modulator depends for its operation on the curvature of the transfer

Q
characteristic. A circuit of this modulator is given in Fig. 15-5. The

IT Madu/d/ed ,",pol
~~Ec(j,..macoswm t)cosUJc ~

r! Modulafinq
enz=EmcosUJm t
FIG. 15-5. The circuit of the van der Bijl modulator.

carrier and modulation-frequency voltages are both applied in the grid

circuit of a triode or multigrid tube, and the modulated wave appears
across the tank circuit in the plate circuit of the tube. The tank is tuned
to the carrier frequency and must be sufficiently broad to include all the
important side bands, usually about ± 10 kc for normal broadcast pur-
poses; otherwise a distortion known as side-band clipping occurs.
The operation of the modulator is made clear in the sketches of Fig.
15-6.
To examine the modulation process analytically, it is supposed that the
transfer curve is parabolic over the range of operation, so that the a-c
plate current may be related to the input grid potential by the first two
terms of the series expansion
(15-8)
where al and a2 are constants. The excitation voltage eg is of the form
(15-9)
324 ELECTRON -TUBE CIRCUITS [CHAP. 15

and it follows, by combining this expression with Eq. (15-8), that the
plate current is
ip = alEm cos Wmt + alEc cos wet + a 2E';,. cos wmt + a2E';, cos 2 Wet
2

+ a2EeEm cos (We + wm)t + a2EmEe cos (We - wm)t

vrld c;rcu,r

PIRfe current
ib l-h

Lowfrequenc;y Modulakcl output

componenf(.;uppressed

FIG. 15-6. The wave shapes at various points in the van der Bijl modulator.

This may be written as

ip = + a2EmEe cos (We + wm)t + a2EmEe cos (We - wm)t
alEc cos wet
a2E';,. + --2- a2E~ + 2
a 2E';,. cos 2wmt + -2- a2E~ cos 2wet (15-"0)
+ al E m cos Wm t + -2- .L

Assuming that We» Wm and also that those frequencies which are not in
the neighborhood of We are eliminated by the use of tuned circuits or
filters, the only voltages which appear across the output are
 SEC. 15-3] AMPLITUDE MODULATION 325
i p = alEe cos wet + a2EmEe cos (We + wm)t
+ a2EmEe cos (We - wm)t (15-11)
This may be written in the form

(15-12)

from which it is seen that the modulation index is given by

2a2Em
m=-- (15-13)
al
It should be noted that a circuit of the present type will operate satis-
factorily if the signal voltages are applied in the plate and grid circuit,
AS tabulated.
Grid circuit Plate circuit
em + eo

The amount of modulated output available without appreciable dis-

tortion in such a modulator as here considered is not great, and the
efficiency is low.

FIG. 15-7. A balanced modulator.

Owing to the fact that the plate circuit contains a parallel tuned rather
than a pure resistance load, the foregoing analysis is not completely cor-
rect-this despite the fact that the load is purely resistive at or near
resonance. By taking this matter into account, Carson 2 has shown that
a substantial increase in output occurs with the substitution of the reso-
nant load. Hmvever, other methods provide better modulation charac-
teristics and have displaced the low-level method here discussed.
15-3. Balanced Modulators. The use of a balanced modulator, 3
which possesses an appearance somewhat like the push-pull amplifier,
automatically eliminates either. the carrier or the modulating frequency,
as well as many of the intermodulation frequencies. The circuit of the
balanced modulator is given in Fig. 15-7.
It is assumed that the two tubes are identical and that the circuit is
symmetrical. The tube characteristics may be expressed by equations
326 ELECTRON-TUBE CIRCUITS [CHAP. 15

of the form
i p1 = aleul + a2e~1 } (15-14)
i p2 = ale u2 + a2e~2

But the input potentials have the form

eul = Ee cos Wet + Em cos wmt }
(15-15)
eu2 = - Ee cos wet + Em cos wmt
Then the currents in the plate circuits become
i pl = alEc cos wet + alEm cos wmt + a2E~ cos 2
wet
+ a2E;' cos 2
wmt + a2EeEm cos (We + wm)t
+ a2EeEm cos (We - wm)t (15-16)
i p2 = - alEc cos wet + alEm cos Wmt + a2E~ cos 2 wet
+ a2E;' cos 2 wmt - a2EeEm cos (We + wm)t
- a2EeEm cos (we - wm)t
But the voltage induced in the secondary of the coupling network is given
approximately by
e == jwM(ipl - i p2) (15-17)
Then the output is of the form
e == 2jweMalEe cos wet + 2jweMa2EeEm[cos (We + wm)t + cos (we - wm)tJ
which may be written as

e == 2weM alEc (1 + Em :: sin wmt) sin wet (15-18)

In certain applications it is found advantageous to use a balanced

modulator in such a manner as to eliminate the carrier. This is readily
accomplished by interchanging the sources fm and fe in the diagram.
When this is done, the output contains frequencies We Wm and We - wm, +

FIG. 15-8. Balanced modulators for producing suppressed carrier modulation.

with the carrier term We missing. Such modulated waves with carrier
missing can be transmitted. The demodulation problem is more com-
plicated than when the We term is present. This will be discussed in
Chap. 16. Two circuits of balanced modulators which suppress the
carrier are illustrated. In these circuits, the LC circuit is approximately
SEC. 15-5] AMPLITUDE MODULATION 327
in resonance for all frequencies in the neighborhood of the carrier fre-
quency We.
15-4. Linear Modulation. A somewhat different view of amplitude
modulation from that above is possible. This point of view is very
important if the carrier and modulating voltages are large or if the
operation occurs near cutoff. Such a situation exists when a class C
amplifier is modulated, whether this modulation is applied in the plate
circuit, the grid circuit, the cathode circuit, or the suppressor circuit,
if a tetrode or pentode is used. If a carrier voltage were introduced into
the grid of a class C amplifier and if the modulating voltage were intro-
duced into the plate circuit, then because We» Wm the variation in poten-
tial in the plate circuit caused by the signal could be considered to be the

Gdulafing
cource

FIG. 15-9. A plate-modulated class C amplifier.

equivalent of a relatively slow change in plate supply potential. If the

relationship between the output tank current is a linear function of the
plate potential, for constant grid excitation, the output potential (which
is a linear function of the tank current) would have the desired modulated
characteristics.
With the proper design and adjustment, the modulation characteristic
of a class C amplifier is such that 100 per cent modulation with distortion
as low as 2 per cent in a plate-modulated amplifier and as low as 5 per
cent in a linear grid amplifier may be attained. The distortion can be
reduced below these values by the use of inverse feedback in the circuit.
15-5. Plate-modulated Class C Amplifier. The basic circuit of a plate-
modulated class C amplifier is given in Fig. 15-9. It will be observed that
it is essentially the circuit of the class C amplifier except for the introduc-
tion of the source of modulating voltage in the plate circuit. The modula-
tion characteristic of such a plate-modulated amplifier is the plot of the
tank current as a function of the plate supply voltage. The ideal plate-
modulation characteristic is illustrated in Fig. 15-10a. Generally the
situation is more nearly like that illustrated in Fig. 15-lOb, which shows
the effect of tank impedance. The more nearly straight the lines, the
less the modulation distortion. A high value of tank impedance yields
the smaller distortion, but the power output is also smaller.
328 ELECTRON -TUBE CIRCUITS [CHAP. 15

If the frequency of the modulating signal is low compared with the

carrier frequency, the impedance of the tank circuit at the modulating
frequency will be negligible. Consequently the properties of the circuit
are not appreciably affected by the modulating frequency, and the plate
and tank currents will follow the characteristics shown even when ebb
varies at the modulation frequency.
Ip,Ih RoJ
Ip Ro2
Ro3
Ia R04

"'""---_____. . . . e"
FIG. 15-10. (a) The ideal plate-modulation characteristic. (b) The modulation
characteristics, showing the effects of load impedance.

Ir Ir(rms)

-1---
-t- L - -
I I I Wc f

I I
I I
I

FIG. 15-11. The conditions during plate modulation.

Suppose therefore that the carrrier voltage of angular frequency Weis

modulated by a modulating voltage of frequency wm • The behavior of
the circuit is that illustrated in Fig. 15-11.
To analyze the modulation process, it will be supposed that we» Wm.
The grid-circuit potentials are, respectively,
eo = Eom cos wet }
(15-19)
and ee = Eom cos wet + Eee
If the modulating voltage in the plate circuit is written in the form
(15-20)
the resulting relatively slowly varying plate potential has the form
(15-21 )
This may be written in the form
ebb = Ebb(1 + m cos wmt) (15-22)
 SEC. 15-5] AMPLITUDE MODULATION 329
where m, the modulation index, is given by the ratio

(15-23)

It should be noted from the curves of Fig. 15-10 that the rms tank
v current and the d-c plate current are related to the plate potential by
expressions of the form
(15-24)

By combining Eqs. (15-24) with Eq. (15-72), it follows that

IT = kTEbb(1 + m cos wmt) }
(15-25)
h = kbEbb(1 + m cos wmt)
Also, corresponding to the rms tank current will be chosen the instan-
taneous tank current iT to be of the form
iT = .y2 IT sin wet (15-26)
This may be expressed in the form
iT = V2 k TE bb (1 + m cos wmt) sin wet (15-27)
But the output voltage eL that appears across the tank circuit is given
with good approximation by
EL = jXdT = XdT/90° (15-28)
Therefore
eL = V2 XdT sin (wet + 90)
= V2 X Lk T E bb (1 + m cos wmt) cos wet (15-29)
The corresponding plate-cathode potential of the tube is

which is
eb = E bb (1 +m cos wmt) (1 - .y2 XLkT cos wet) (15-30)
Also, from the fact that the plate current is given by
I - EL _ • X d T _ . /T _ IT /900
pi - Ro - J R;; - J Q- Q __
the instantaneous-plate current has the form

ipl = 'i kTEbb (1 + m cos wmt) cos wet (15-31)

Likewise, it follows from the curve of Fig. 15-10 that

ib = h(1 + m cos wmt) (15-32)
330 ELECTRON -TUBE CIRCUITS [CHAP. 15

The above information may be used to analyze the performance of the

modulated amplifier.
The average power supplied by the d-c plate power source is

(15-33)
which is
P bb = T1 J(T Ebbh(1 + m cos wmt) dt
0

This integrates to
hb = Ebb1b = kbEk, (15-34)
The power input by the modulating source is
1 (T .
Pm = T J0 emZb dt
This becomes
Pm = ~ loT Emm COS Wmt h(1 + m COS wmt) dt
which integrates to the form
(15-35)

Clearly, for 100 per cent sinusoidal modulation, the modulating source
must deliver one-half as much power as the d-c plate power supply.
This requires, of course, that the modulating source must be an amplifier
of large power capacity for a large power output.
The a-c power output at the tank circuit is given by the expression

PL =
I
T J(T
0
.
eLZpl
d
t =
1
T J(T
0
R . d
2
OZPl t (15-36)
This becomes

= T1 Jo(T 2Ro (kTEbb)

2
PL Q (1 + m cos wmt)2 cos 2
wet dt

which becomes, by performing the indicated integrations,

PL = Ro eT~bby (1 + ~2 + ~2)
This may be reduced to the form

PL = ~02~: P bb (1 + ~2) (15-37)

It may be concluded from this that the d-c plate power supply furnishes
SEC. 15-5] AMPLITUDE MODULATION 331
the power to produce the carrier wave and the modulated amplifier
furnishes the power to produce the side bands in the output.
The plate-circuit efficiency of the modulated amplifier is given by the
expression
(15-38)
which becomes
Rok~ Pbb(l + m 2/2)
'YJ = Q2kb P bb (l + m2/2)
or
Rok~
'YJ = Q2k b (15-39)

This result shows that the plate-circuit efficiency is independent of the

degree of modulation. Therefore one may calculate the efficiency from
considerations of the unmodulated amplifier as a simple class C device.
To find an expression for the power dissipated in the plate of the tube, it is
evident that
P p = P bb Pm - P L +
which may be written in the form

P p = P bb (1 - 7J) (1 + n;2) (15-40)

Consider the results when the modulation index m is zero. The fore-
going become, in this case,
P bb = kbE~b
Pm = 0
P L = 7JP bb
(15-41)
Pp = (1 - 7J)Pbb
Rok~
7J = Q2k b

By comparing these expressions with the corresponding expressions when

m is not zero, it is observed that the addition of modulation increases the
plate dissipation. This requires that a given tube in being operated
under modulated conditions must be operated with a reduced plate volt-
age and current if a specified maximum allowable plate dissipation is not
to be exceeded.
In order to design the modulating amplifier, a knowledge of the effective
impedance across the secondary terminals of the output transformer of
this amplifier is needed. This will permit a specification of the turns ratio
of the modulating transformer in order to reflect the optimum value of
impedance into the plate circuit of the modulating tubes. This imped-
332 ELECTRON-TUBE CIRCUITS [CHAP. 15

ance is readily obtained by observing that the plate impedance is sub-

stantially resistive and must be given by

R = E;" = E;"m/2 = Ebb = ~ = Rb (15-42)

m Pm E mm h(m/2) h kb
This shows that the effective impedance is independent of the modulation.
The plate-modulated amplifier is used extensively in radio transmitters.
It has the advantage that modulation without excessive distortion is
possible in practice by reasonably
simple methods. Also, it operates
at high efficiency and is relatively
easy of adjustment. It has the
---- - -ecc ----- disadvantage that a compara-
--'->OL>ULf- --.111---4----1
~ ./!lee tively large amount of power at
the modulating frequency is re-
FIG. 15-12. A grid-bias modulated class C quired. The resulting cost of the
amplifier.
heavy and bulky modulating
equipment is sometimes greater than that of other methods. It might
be of interest to know that the two modulating transformers at the 500-
kw radio station WLW weigh approximately 19 tons each and that an
audio choke weighs 12 tons. 4

LOIrge grid drive

eo ec eo
FIG. 15-13. Typical linearity curves of a grid-bias modulated class C amplifier under
different conditions.

15-6. Grid-bias Modulation. Amplitude modulation may be accom-

plished by connecting the modulating source in series with the grid instead
of in the plate circuit. The basic circuit of such a grid-bias modulated
amplifier is illustrated in Fig. 15-12. Typical linearity curves of such
an amplifier for several different conditions are illustrated in the curves
of Fig. 15-13. The general character of the operation is illustrated
graphically in Fig. 15-14.
To analyze the operation, the procedure is essentially parallel to that of
Sec. 15-5 for the plate-modulated amplifier. It is assumed that the
modulation characteristic is linear and that we» wm. The carrier signal
is chosen of the form
(15-43)
SEC. 15-6] AMPLITUDE MODULATION 333

and the modulating voltage is of the form

(15-44)
The total grid-cathode potential has the form
Ce = Egm cos wet + Emm cos wmt + Eeo (15-45)
and the slowly varying grid supply source is given by
Cee = Emm cos wmt + Eee (15-46)

FIG. 15-14. Conditions for 100 per cent modulation in a grid-bias modulated class C
amplifier.

Over the linear range of operation, the analytic form for the rms tank
current is
IT = kT(Cee - co) (15-47)
Also, if the instantaneous value of the tank current is chosen as
iT = V2 IT sin wet
this may be written as
iT = V2 kT(C ee - Co) sin wet
or
iT = V2 kT(Emm cos wmt + Eec - co) sin wet (15-48)
But when the modulation is zero, the tank current has the form
iT = vI2 kT(E ce - Co) sin wet
which may be written as
iT = I Tm sin wet }
where (15-49)
I Tm = V2 kTCEec - Co)
Note that, when the modulation exists, the value of the tank current
at the peak of the modulating cycle is
(15-50)
334 ELECTRON-TUBE CIRCUITS [CHAP. 15

The conditions are best examined graphically as in the accompanying

figure. Clearly, the degree of modulation is seen to be

(15-51)

which may be written as

(15-52)

By combining Eq. (15-52) with Eq. (15-48), the instantaneous tank

current assumes the form
iT = y2 kT[(Eee - eo)m cos wmt + (Eee - eo)] sin wet
which is
(15-53)
In an entirely parallel way, it is possible to find for the plate current the
expression
(15-54)
which may be written as
Ib = kb(Eee - eo) (1 + m cos wmt) (15-55)
The voltage across the tank circuit is
EL = jXdT
which is
(15-56)
Therefore the instantaneous potential across the tank circuit is given by
eL = V2 XLkT(Eee - eo)(1 +m cos wmt) sin (wet + 90)
or
eL = V2 XLkT(Eec - eo)(1 + m cos wmt) cos wet (15-57)
It is now possible to complete the analysis of the circuit.
SEC. 15-6] AMPLITUDE MODULATION 335
Expressions for the various important values of power in the circuit
are readily obtained. The power input to the plate circuit by the d-c
plate supply is
1 (T .
P bb = T Ebb~b dt Jo
This becomes, by Eq. (15-55),

P bb = ~'IoT Ebbkb(Eee - eo)(l + m cos wmt) dt

which integrates to
(15-58)
Since this result shows no dependence on the modulation index m, it
follows that the plate supply power is independen~ of the degree of
modula tion.
The power output at the tank circuit is given by

PL = ~ loT eLi pl dt = ~ loT Roi~l dt

which becomes, by Eq. (15-53),

PL = ~ loT 2~o:~ (Eec - eo)2(1 + m cos Wt)2 sin

m
2
wet dt

This integrates to

P = RQ~~
L (Ecc _ eo)2 (1 + ~2) = I~R (1 + ~2) (15-59)

This shows that the r-f power increases with percentage modulation.
The plate-circuit efficiency of the modulated amplifier is given by
PL (RokUQ2)(Eec - eo)2 (1 + m 2j2)
7J = P bb = Ebbkb(Ecc -.eo)
which becomes
Rk2 (m2)
7J = kbE~b (Eec - eo) 1 + 2'" (15-60)

This expression shows that the plate efficiency increases as the modula-
tion index increases. To realize the highest efficiency for a given modu-
lation index (= 1) it is necessary that the quantity Eec - eo be made as
large as possible. This requires that the load impedance of the modu-
lated amplifier be so adjusted that the peak amplitude of the output wave
is only slightly less than the d-c plate supply.
The plate dissipation of the tube is given by the expression
Pp = P bb - P L
336 ELECTRON -TUBE CIRCUITS [CHAP. 15

which is
pp = kbEbb(Ecc - co) - Rk~(Ecc - co)2 (1 + ~2) (15-61)
This may be written in the form

Pp = Ebbh - I~R (1 + ~2) (15-62)

It should be noted that the plate dissipation decreases as the percentage

modulation increases. Clearly, therefore, the plate dissipation is a
maximum when the amplifier is unmodulated.
Grid-bias modulation has the advantage that only a small amount of
l-f modulating power is required. However, the carrier power that is
obtainable from the amplifier is approximately one-quarter of that from
the same tube when operating as a simple class C amplifier. This is so
because the peak power of the modulated amplifier corresponds to class
C operation, and with a completely modulated wave the peak current is
twice the unmodulated current and the corresponding peak power is four
times the unmodulated, or carrier, power. Also, the plate efficiency
during the unmodulated intervals is approximately one-half of the
efficiency obtained with simple class C operation. This results from the
fact that, if the amplifier is so adjusted that the plate voltage is small at
the crest of the modulation cycle, then, when there is no modulation, the
voltage across the load is halved. This results in a high voltage across
the tube, with a corresponding large plate loss. As a result, the plate-
circuit efficiency in the absence of modulation is of the order of 34 per
cent. During 100 per cent modulation by a sinusoidal signal, the r-f
power output increases by a factor of Yz, and the plate-circuit efficiency
increases to approximately 51 per cent.
It is interesting to compare the operating features of a plate and a grid-
bias modulated amplifier.. These are
Grid-bias modulated amplifier:
Relatively low plate efficiency.
Low power output in proportion to the capabilities of the modulated
tube.
Low grid-modulating power.
Plate-modulated amplifier:
Large power output in proportion to the power capabilities of the
modulated tube.
Large modulator power.
In consequence, the over-all efficiency, considering both the modulating
and the modulated tube capabilities, is roughly the same. The choice
SEC. 15-7] AMPLITUDE MODULATION 337
between the two methods of modulation is largely one of convenience,
since both methods of modulation will give sensibly 100 per cent modu-
lated waves with essentially negligible distortion. The circuit adjust-
ments are more difficult with grid-bias modulation, as they are sensitive
to variations in the carrier exciting voltage, the plate supply voltage, and
the magnitude of the tuned load impedance.
15-7. Other Methods of Modulation of a Class C Amplifier. A
modulated output wave is produced if the modulating voltage is intro-
duced in the cathode of the amplifier. Since in such a circuit the modu-
lating voltage is effectively in both the plate and the grid circuits, the
characteristics are a combination of those of the plate-modulated and
the grid-bias modulated amplifiers. The plate efficiency and the modu-
lating power requirements are intermediate between the corresponding
requirements for the plate- and the grid-modulated amplifiers.
If a pentode is employed, the modulation voltage may be applied in the
suppressor-grid circuit. The general characteristics of such a suppressor-
grid modulated class C amplifier are similar to those for the control-grid
modulation. However, adjustment of the amplifier is simpler. Fairly
linear modulation up to 100 per cent may be obtained.
Modulation may be accomplished by injecting the modulation in the
screen circuit of a tetrode or pentode. Some power is required from the
modulating source, and 100 per cent modulation usually cannot be
achieved without some distortion.
REFERENCES
As general references, see
M.LT. Sta~, "Applied Electronics," Sec. 12-4 to 12-6, John Wiley & Sons, Inc.,
New York, 1943.
Croft Laboratory, War Training Staff, "Electronic Circuit and Tubes," Chap.
XX, Secs. 15-18, 15-19, McGraw-Hill Book Company, Inc., New York, 1947.
1. Standards on Transmitters and Antennas, Institute of Radio Engineers, 1933.
2. Carson, J. R., Proc. IRE, 9,243 (1921).
3. Peterson, E., and C. R. Keith, Bell System Tech. J., 7, 131 (1928).
4. Chambers, J. A., L. F. Jones, G. W. Fyler, R. H. Williamson, E. A. Leach, and
J. A. Hutcheson, Proc. IRE, 22, 1151 (1934).
PROBLEMS
15-1. The equation of a modulated wave is
e = (15 + 10 sin 3,OOOt - 8 cos lO,OOOt) cos 21r X l06t
a. What frequencies are contained in the modulated wave?
b. What is the amplitude of each?

15-2. Carry out the analysis to show that amplitude modulation results in a
square-law circuit when the carrier voltage is applied in the grid circuit and the
modulating voltage is applied in the plate circuit.
338 ELECTRON-TUBE CIRCUITS [CHAp. 15

15-3. Prove that the carrier term is missing in the balanced modulator of Fig.
15-8.
15-4. The balanced modulator of Fig. 15-8a is used for carrier suppression.
In this circuit, the potentials applied are the following:
e, = E, cos w,t
em = Em cos Wmt
with w, » Wm and with Em = 0.5E,. Assume that the transconductance gm
varies linearly with grid voltage.
a. Obtain an expression for the output voltage from the modulator.
b. Plot the envelope of this potential.
15-5. Suppose that a band-pass filter is connected in the output of Prob. 15-4
of such characteristics that the lower side bands are eliminated.
a. What is the expression for the resulting output?
b. Plot the envelope of this wave.
15-6. An ideal diode for which r p = 1,000 ohms in the forward direction and
rp = 00 in the inverse direction is used as a modulator. There is applied to this
circuit the two voltages
e = E, cos w,t +Em cos Wmt
with w, » Wm and with Em = 0.5E,.
a. Determine the amplitude of the component of current of angular frequency
(w, - w m ).
b. Repeat for the component of frequency (w, + w m ).
15-7. Repeat Prob. 15-6 for the case where the single diode is replaced by
four diodes connected in a bridge circuit.
15-8. A type 851 power triode operates with a bias of -300 yo Its and with a
peak r-f signal of 525 volts. The load impedance Ro = 1,500 ohms. Determine
and plot the a-c plate potential across the tank as a function of the d-c plate
supply voltage, for the following values of plate voltage: Ebb = 500, 1,500,2,500,
3,500 volts.
15-9. Repeat Prob. 15-8 for Ro = 1,000 ohms.
15-10. An 851 triode is used in a plate-modulated class C amplifier. It oper-
ates with a bias of -300 volts, a peak r-f signal of 525 volts, and a plate supply
Ebb = 2,000 volts. When ebb = Ebb, the peak plate swing is 1,750 volts. Deter-
mine the plate-current wave-forms at the values of ebb = 1,000,2,000,3,000 volts.
15-11. A type 891 r-f power triode has the following ratings as a class C oscil-
lator for telegraphy:
D-c plate voltage ........... . 10,000 volts
D-c grid voltage. . . . . . . .......... . -2,000 volts
D-c plate current. . . . . . . . . . ........... . 1.45 amp
D-c grid current.. . ............................. . 0.105 amp
Grid driving power ................................ . 310 watts
Power output. . . . . ........ . ......... . 10 kw
Peak r-f grid voltage .... 2,900 volts
 AMPLITUDE MODULATION 339
If the plate dissipation is the only limiting factor, determine the corresponding
ratings of the tube for class C telephony, allowing for 100 per cent plate modula-
tion. When plate-modulated 100 per cent, determine
a. The audio power required.
b. The impedance offered to the audio source.
c. Power output.
d. Plate-circuit efficiency.

15-12. An 852 transmitting triode has the following ratings as a plate-modu-

lated r-f amplifier under carrier conditions that allow for 100 per cent modulation:
D-c plate voltage ........................ . 2,000 volts
D-c grid voltage. . . . ...... . -500 volts
D-c plate current. . ............................. . 67ma
D-c grid current. . ............................. . 30 rna
Peak r-f grid voltage. . . . . ................... . 750 volts
Grid driving power .. . 23 watts
Power output ......... . 75 watts
The amplifier is sinusoidally plate-modulated 75 per cent. Determine
a. Audio power required.
b. Impedance offered to the audio source.
c. Plate efficiency.
d. Average plate dissipation.
e. Grid dissipation at the tube terminal.
If the amplifier were unmodulated, what would be the maximum allowable
r-f power output, assuming that the plate dissipation is the limiting factor,
and that the plate-circuit efficiency remains constant?
15-13. The results on a plate-modulated class C amplifier are given in the
figure. Suppose that this modulated amplifier is operated at 1,600 volts d-c,
10
q ./
Type 810
8 f-- Ecc=-200 /
7T
Egm=J70
~1 f-- Q =/2.6 /
E
15,6 / 0.6 III
1: // 0.
t
:::I
5
L/
0.5 ~
"'4 ../ 0.41:'
~ V / Ib ~
~3 0.3 5
/ ~ ......
2 0.2 '"
~

1
V ./'
i-"'" IS
0.1 "li
~ I-' I,)

00 o6
400 800 1200 1600 2000 2400 2800 3200 3600
with 75 per cent modulation. Assume a constant plate efficiency of 60 per cent.
Calculate the following:
340 ELECTRON-TUBE CIRCUITS [CHAP. 15

a. The power supplied by the d-c plate source.

b. The power supplied by the audio source.
c. The r-f carrier power.
d. The r-f side-band power
e. The resistance of the class C amplifier to the modulating source.
I. The maximum and minimum instantaneous voltage between plate and
cathode when the carrier is unmodulated.
g. Repeat f. for m = 0.75.
15-14. The essential elements of a plate-modulated class C amplifier are

J50

illustrated in the diagram. The tubes are operated under the following
conditions:
843 tube: 6N7 tube:
Ie = 1.2 megacycles Ebb = 300 volts d-c

Ebb = 350 volts d-c Effective plate-plate resistance 8,000 ohms

Ib = 30 ma d-c
Impedance of 843 tank circuit at 1.2 megacycles is 10,000 ohms.
Effective Q of 843 tank is 15.
The carrier output is to be 5 watts 100 per cent modulated.
Calculate the following:
a. The plate-circuit efficiency of the 843.
b. The power required from the 6N7.
c. The load impedance presented by the 843 to the secondary of the modulating
transformer.
d. The transformer ratio that should be used.
e. If the output is to be down 1 db 5,000 cycles off resonance, what must. the
effective Q of the 843 tank circuit be?
I. The value of C.
g. The approximate value of the r-f choke RFC.
h. With a loaded Q of 15, calculate the value of Land C of the 843 tank circuit.
t. If the actual resistance in the 843 tank is 5 ohms, how much resistance is
coupled into this tank from its load?
j. What is the efficiency of power transfer fJ'nffi the tank to the load?
 AMPLITUDE MODULATION 341
15-15. An 806 tube is to be used as the plate-modulated class C power amplifier
of a transmitter. It operates from a 2,500-volt power supply. Carry out the
design of this and the associated transformer-coupled class B modulator. The
design must fulfill the following specifications:

a. The plate dissipation is not to exceed 150 watts.

b. The transmitter is to be plate-modulated 70 per cent.
c. The distortion in the modulating envelope is not to exceed 10 per cent.
d. The transmitter frequency is 2 megacycles.
e. The grid bias may be obtained with grid leak, fixed bias, or a combination of
both.
f. Specify the plate supply voltage, grid bias, and excitation voltage.
g. The output is to feed a 76-ohm antenna.
h. Specify the primary inductance, secondary inductance, tuning condenser,
and mutual inductance. Assume that the unloaded Q of the coils is 300
and that the loaded Q is 12.
~. Give the modulation-tranE[former requirements.

Be sure that any assumptions that are made are clearly stated.
15-16. The characteristics of an 802 tube when used as a grid-bias modulated

1.0
-50 r------ A k f
Gne~ vol'S
a-I. -----1+50

0.9
: /
:I VI T
0.8
aE 0.1 I I
~0.6
II 60
I
/ ~h
r:
t 0.5 5
:>
~ 0.4 l /
s:
~ 0.3 I: 1/
.:t J
3

0.2 2

0.1 I. /1 I
o-zoo V 1 - -Ic 1 06
o

-115 -150 -125 -too -15 -50

InsJOIntomeous a-f gri£l volts
Control-grid modulation characteristic-type 802:
Ebb = 500 Eoc = - 130
Ecc2 = 200 Peak r-f grid volts = 145

class C amplifier are given in the sketch. Calculate for m = 1.0 the following:
a. The power supplied by the d-c source.
b. The r-f carrier power.
c. The side-band power.
d. The plate efficiency.
342 ELECTRON-TUBE CIRCUITS [CHAP. 15

e. The plate dissipation.

/. The maximum and minimum instantaneous voltage between cathode and
plate when the carrier is unmodulated.
g. Repeatj.form = 0.75.
h. What is the amplitude of the a-f signal to achieve the desired degree of
modulation?
i. What is the a-f driving power?
j. What is the grid-cathode power loss?
Choose RL = 12.6 ohms; Q = 26.5.
15-17. Consider a class C amplifier that is modulated 100 per cent. Assume
the output power to be 1,000 watts. Assume reasonable values of plate-circuit
efficiency for each of the tubes involved, and calculate (1) the total power required
by the modulated tube, (2) the total power required by the modulator tubes.
and (3) the over-all plate-circuit efficiency when the amplifier is
a. plate-modulated, using a class A modulator.
b. plate-modulated, using a class B modulator..
c. grid-bias modulated.
Repeat the calculations for zero per cent modulation.
 CHAPTER 16

DEMODULATION

W HEN the radiated modulated carrier signal reaches the receiving point,
the signal or intelligence must be extracted therefrom. The process by
which the signal is recovered from the radiated wave is broadly known
as demodulation or detection.
It is important that the processes of modulation and demodulation be
clearly understood. As already discussed, in the process of modulation
the signal frequencies, which are centered about the zero-frequency
reference level, are shifted upward on the frequency scale so as to be
centered about the carrier frequency We. This frequency shifting is
accomplished by mixing the signal frequencies with the carrier frequency
in appropriate circuits. In the process of demodulation the signal spec-
trum, which is centered about We, is shifted downward on the frequency
scale so as to be centered about the zero frequency, thus returning it to
its original frequency position. This frequency shifting is accomplished
by mixing the signal-frequency group centered about We with the carrier
frequency We in appropriate circuits. Clearly, both the modulating and
the demodulating processes involve frequency shifting, and both shifts
are by an amount We. However, in the modulating process the carrier is
generated in one circuit, and this is combined with the audio signal in the
modulator. In demodulation the required carrier wave is ordinarily
contained in the incoming modulated carrier, and no separate carrier-
generating circuit is generally necessary. Such a separate carrier-
generating circuit is required in suppressed carrier transmission.
It is also customary to consider the elimination of the original carrier
from the modulated signal and the substitution for it of a new carrier, as
in frequency changing, as demodulation. In essence, therefore, frequency
shifting in which the signal frequency does not playa direct part would
be considered as demodulation.
16-1. Square-law Detection. Detection is possible when a modulated
voltage is applied to the grid of a tube which is biased to the nonlinear
portion of its transfer characteristic. This is made evident by an exam-
ination of Fig. 16-1. The output curve clearly shows the presence in the
output of a component that varies at the modulating-frequency rate.
To examine the operation of the circuit analytically, it will be supposed
343
344 ELECTRON-TUBE CIRCUITS [CHAP. 16

that the transfer curve is a simple square-law characteristic of the form

(16-1)
Suppose that the signal is an a-m wave of the form
eg = Ee(1 + m cos wmt) cos Wet (16-2)
Then the output current will contain the terms
ip = + m cos wmt)2 cos wet
aE~(l 2

aE2
= T (1 + m cos wmt)2(1 + cos 2wet)
1 m2 m2
= aE~ [2 + m cos Wmt +T +T cos 2wm t

+ 21 cos 2wet + "2m cos (2we + wm)t + "2m cos (2we - wm)t

(16-3)

FIG. 16-1. Sketch showing the operation of a square-law detector.

Clearly, there will be included in the output a number of steady com-

ponents, a term of modulating frequency, a number of components of
frequencies equal to the sums and differences of the carrier and side fre-
quencies, and a number of components of twice the carrier and side
frequencies.
Suppose that a selective network is used which attenuates all com-
ponents except those in the neighborhood of the modulating frequency
wm • The terms that appear in the output will then be

i 1 ='= amE~ (cos wmt + ~ cos 2wmt) (16-4)

SEC. 16-2] DEMODULATION 345
which consists of the desired term plus one of second harmonic of this
frequency. If the second-harmonic amplitude is to be kept smaller than,
say, 10 per cent of the fundamental, it is necessary that the modulation
index be less than 0.4 for sinusoidal modulation. Despite this limitation,
such detectors have been used extensively with generally satisfactory
results, probably because the average modulation of the ordinary radio
program is of the order of 40 per cent.
It is of interest and of significance in subsequent work to examine those
terms in the neighborhood of the second harmonic of the carrier 2w c,
although these results are not of importance at this particular point.
The terms are

i2 = a:~ [ + ~2)
(1 cos 2wct
2
+ m cos (2wc + wm)t + m cos (2wc
2
- wm)t

+ 4m cos (2wc + 2wm)t + 4m cos (2wc - 2wm)t ] (16-5)

If m is small, this equation can be written as

i2 == a:~ [cos 2wct + m cos (2wc + Wm)t + m cos (2wc - Wm)t]

which is
i2 = a:~ (1 + 2m cos wmt) cos 2wct (16-6)

Therefore, if one were to use a tuned circuit at the output of a square-law

circuit which is tuned to 2wc, the output would be of second harmonic of
the carrier but the modulation frequency would be unchanged. Fre-
quency doublers of this type find extensive use in b-f operations.
16-2. Diode Detection. Diode detectors have almost completely sup-
planted other types of detectors in home radio-receiver use. They
operate very satisfactorily, although the modulated input to the detector
must be large-in excess of about 2 volts. Otherwise excessive distor-
tion may result because of the nonlinear dynamic curve of the diode
circuit. Since the diode detector possesses characteristics in common
with ordinary rectification, it is possible to present a qualitative discus-
sion of its operation on this basis.
There are two important types of operation of a diode detector. One
is known as average detection, and the other is envelope detection. The
circuit for average detection is illustrated in Fig. 16-2. Also illustrated
in this figure is the character of input and output waves. It will be
observed that this is just a simple diode rectifier without a filter in the
output. The application of a modulated wave to the circuit yields an
output the average value of which contains the modulating frequency.
346 ELECTRON -TUBE CIRCUITS [CHAP. 16

It follows from simple rectifier theory that the output voltage is of the form
R
Em = 7r(R + rp) Ec(1 +m cos wmt) (16-7)

The use of appropriate circuits permits the modulating-frequency com-

ponent to be extracted from the output..

FIG. 16-2. The circliit and the operational characteristics of an average diode
detector.

FIG. 16-3. Two circuits for envelope detection.

The circuit for envelope detection is illustrated in Fig. 16-3. It will

be observed that these are essentially the circuits of a simple diode
rectifier with a capacitor filter.
The operation of this circuit is sub-
stantially the same as that for the
simple rectifier, with the additional
fact that the a-c input voltage va-
ries in amplitude. With the proper
choice of the filter capacitor, the
wave forms of the input voltage, the
output voltage, and the tube cur-
FIG. 16-4. Wave forms of the input
voltage, output voltage, and plate cur-
rent are roughly those shown in Fig.
rent in a diode-detector circuit. 16-4. Since the carrier frequency
is much greater than the modulat-
ing frequency, the jagged appearance of the above curves is considerably
worse than is actually the case. In effect, therefore, the output voltage
SEC. 16-3] DEMODULATION 347
of the detector follows closely the envelope of the modulated input wave
and has the wave form of the modulating voltage.
16-3. Analysis of Diode Detection. 1 An analytic expression that
gives certain of the properties of the diode detector is possible if several
reasonable assumptions are made. In particular, it will be assumed that
the d-c potential across the diode load, which arises from the tube current
during the portion of the cycle when the input voltage exceeds the voltage
across the output capacitor, will remain constant over the carrier-fre-
quency cycle. It will also be assumed that the static characteristic of the
diode is linear. The conditions that apply during one cycle of the
impressed modulated r-f signal wave are as illustrated in Fig. 16-5.

FIG. 16-5. The approximate action of a diode detector.

The equation of the assumed characteristic is

. eb
~b = -rp = gpeb
(16-8)
ib = 0
But as the instantaneous voltage applied to the diode is given by the
equation
eb = EcC1 + m cos wmt) cos wet - Ea
= E' cos wet - Ea (16-9)
the plate current will be of the form
ib = gp(E' cos wet - Ea)
(16-10)
ib = 0
The shift from one equation to the other occurs at the angle (J, which is
such that
E' cos wetO = E' cos (J = Ea (16-11)
348 ELECTRON-TUBE CIRCUITS [CHAP. 16

The instantaneous current ib will contain a number of harmonic com-

ponents, but those of interest are the d-c or average component and the
a-c term at the driving frequency. The d-c component of ib is given by

Ib = -1 10 2
71"
~b
•
dO
271" 0

which is, because of the symmetry that exists,

h =- 110° ~b. dO
71" 0

This may be written as

h = - 11°
71" 0
gp(E' cos 8 - Ea) dO (16-12)
which becomes
h = gp [E' sin 0 - EaO] (16-13)
71"

This becomes, by Eq. (16-11)

h = gp E'(sin 0 - 0 cos 0) (16-14)

71"

Also, it is initially assumed that

Ea = IbR (16-15)
By combining Eqs. (16-14) and (16-15), there results
I~ = Ea = E' cos 0 = gpE' (sin 0 - 0 cos 0)
R R 71"

from which
r 1
-A = ;;: (tan 0 - 8) (16-16)

This expression shows that there is a direct functional relationship

between the operating angle 0 and the ratio of the plate resistance of the
diode to the external resistance. An explicit expression for (j as a func-
tion of rp/R is very difficult, but the information may be given graphically.
The detection efficiency of a detector is defined as the ratio of the actual
output voltage of modulation frequency Em to the maximum possible
value of this modulating voltage, i.e., to the value of the envelope of the
carrier. Analytically the detection efficiency is defined by the relation
(16-17)
To find an expression for 7], it is noted that, when no capacitor is used
SEC. 16-3] DEMODULATION 349
(average detection), the output voltage is
R
Em = 7r(R + rp) Ec(1 + m cos wmt)
which is, for R > rp
Em == Ec (1
7r
+ m cos wmt) (16-18)

Evidently the maximum possible value would occur with C shunting R

and would be
(16-19)
However, with C present, the output is actually given by Eq. (16-11),
which is
Ea = E' cos (I = Ec(l + m cos Wmt) cos (I (16-20)
Hence it follows, since

that
7] = cos (I (16-21)
A plot of 7] as a function of rp/R is contained in Fig. 16-6. It should be
noted that, if R is large compared with r p, then 7] is practically independent
(0 1.0
I ..-1-
8 0.8

6~ 0.6
'f1.
>~ ~'~
!tl.. ~
4 ·.0.4

2
J::-.

0.2
fJ. V
o 0 0.(
I- ~K
1.0 (0 100 1000
R/rp
FIG. 16-6. Various important terms in the analysis of diode detectors.
of rp. Consequently the detection efficiency of a diode is not appreciably
influenced by the curvature of the characteristic.
Another quantity of importance in detector operation is the power
absorbed by the detector, or the power loss in the diode circuit. To
evaluate this requires a knowledge of the fundamental-frequency com-
ponent of the current. The maximum value of the current is given by
the Fourier coefficient
2
I p1m = -110 ...
~b cos wt d(wt)
7r 0
350 ELECTRON -TUBE CIRCUITS [CHAP. 16

This becomes
8
I plm = -210 gp(E' cos (J - Ea) cos (J d(J (16-22)
7r 0
which integrates to
I plm = -2gpE'
7r-"2
((J + 4;1 sm
. 2
(J - ·
sm (J cos (J)
or
E'
I p1m = ~ «(J - sin (J cos (J) (16-23)
7r

The power input to the diode and its load is

T
p=-110 mbdt
.
T 0
E'I E'2
P = ~ = ~ «(J - sin (J cos (J) (16-24)
2 2
The effective resistance of the diode circuit is defined by the relation
_ E'2 _ 7r
R. = 2P - gp«(J - sin (J cos (J) (16-25)
which may be written as

Re = rp;3 }
(16-26)
;3 == «(} _ si:(} cos (})
That is, the effective resistance in parallel with the capacitor due to
the loss in the diode circuit is equal to ;3rp. A plot of ;3 is also contained
in Fig. 16-6.
By combining Eqs. (16-26) with (16-16), an expression for the equiva-
lent resistance that shunts the diode input circuit is possible. This is
Re Re rp tan (} - (}
R = r p R = (} - sin (} cos (}
This expression is plotted as a function of Rjrp in Fig. 16-6.
For the case when 'Y} is high, the equivalent input resistanc p -3duces to
a simple form. Noting that, for 'Y} high, (} is small,
• (} • (} (}3 (}2
SIn = -- cos (} == 1 - -
6 2
and
Re 1, (sin (} - (} cos (}) . 1
(16-27)
R = cos (} (} - sin (} cos (} = 2'Y}

Hence the effective input resistance becomes Rj2'Y} for large 'Y} and hence
is slightly greater than one-half the load resistance.
SEC. 16-4] DEMODULATION 351

16-4. Distortion in Diode Detectors. There are two sources of distor-

tion in a simple diode detector. One results from the curvature in the
tube characteristic, making the efficiency of rectification vary according
to the amplitude of the envelope. This source of distortion may be
minimized by making the load resistance large compared with the diode
plate resistance and by making the amplitude of the carrier amplitude
applied to the diode reasonably large. Under practical conditions, when
the detection efficiency exceeds 80 per cent, the distortion from this source
is of the order of 2 per cent for a CI{lJplnq awinq !rough
completely modulated wave. With of" modulated wave
small signals the distortion may reach
as high as 25 per cent for a com-
pletely modulated wave when the
signal voltage is a fraction of a volt.
The second source of distortion
arises from the fact that the voltage FIG. 16-7. Diagonal clipping in a
across the capacitor in the output diode detector when the load-circuit
can die away only as fast as the time constant is too large.
charge can leak off through the load resistor. Hence, unless the time
constant of this circuit is properly chosen, clipping may result during the
troughs of the modulated signal.
If the h-f variations of the output voltage are to be small, the time
constant of the load circuit RC must be large compared with the period
of the carrier-frequency cycle. However, if this value is made too large,
the output voltage cannot decay as rapidly as the envelope decreases, and
clipping occurs. The conditions discussed are illustrated in Fig. 16-7.
To ascertain the maximum allowable value of the time constant, it
should be noted that this value must be such as to permit the capacitor
to discharge at the same rate as the decrease of the modulation envelope.
This may be estimated in the following manner:2 The most unfavorable
condition occurs at the highest modulation frequency Wm that the detector
is designed to handle and is that for which the equation of the envelope is
e = Ec(1 + m cos wmt) (16-28)
At any particular time t = to, the value and the slope of the modulation
envelope are
e = Ec(1 + m cos wmt o) }
(~i)to= -wmmEc sin wmto
(16-29)

If the voltage across the capacitor equals the modulation voltage at the
time t = to,
ea = Ec(1 + m cos wto) (16-30)
352 ELECTRON -TUBE CIRCUITS [CHAP. 16

and it decays thereafter according to the exponential expression

(t-to)
ca = CaOC -----s:c- (16-31)
The initial rate of change is

(~~~\ = - RIC Ca = - Xc (1 + m cos wmto) (16-32)

If the capacitor voltage is to be less than the value of the envelope for
times t > to, the slope of Ca must be less than that of the envelope at
t = to. This requires that

or
-:c (1 + m cos wmto) ~ -wmmEc sin wmto

1 >
RC - Wm 1
(m+ sin wmtO )
m cos wmto (16-33)

For the initial rate of decay of the capacitor voltage to be greater than
the rate of decay of the envelope voltage, it is necessary that

1
->W ( m sin wmto )
1 + m cos wmto
m
RC
But the most severe condition on the RC constant is that for which the
fraction is a maximum. To find this, consider the expression

d ( m sin wmto ) _ 0
dt 1 + m cos wmto -c_
This yields
cos wmto = -m
sin wmto = V 1-:---m-:::2
from which
1 m
~>W (16-34)
RC m VI _ m2
If this equation is satisfied, the output voltage follows the wave form of
the envelope. According to this equation, as the modulation approaches
100 per cent, the required time constant approaches zero. Consequently,
at 100 per cent modulation, the output voltage contains the carrier as
well as the modulating frequency.
By taking into account a number of factors that were neglected in the
above analysis, such as the impedance of the source supplying the modu-
lated voltage, the results must be modified somewhat. Experimentally
it has been found that the amount of harmonic generation is not excessive
SEC. 16-6] DEMODULATION 353
for sound reproduction if
(16-35)

16-5. Diodes with Complex Load Impedance. It is generally found

necessary to provide a filter to prevent any r-f voltage from reaching the
output. A circuit that accomplishes this is illustrated in Fig. 16-8 and
makes use of a simple IT-type resistance-capacitance filter. It is also
common practice to use the d-c voltage that is developed across the diode
load for automatic volume control. This necessitates the addition of
impedance elements, as will be dis-
cussed below. The resulting mod-
ifications cause the impedance that
the load offers to the modulation-
frequency component of the out- FIG. 16-8. Diode-detector circuit.
put voltage to differ from, and in
general be less than, the resistance that the diode circuit offers to the recti-
fied d-c current. This results in a modification of the properties of the
diode rectifier, and the results of Eq. (16-34) must be reexamined.
The conditional relation given in Eq. (16-34) is rearranged to the form

This may be expanded to

or

from which
X2 _ X2R2/(R2 + X2)
R2 + X2 - R2
or
m < XR/VR2 + X2 = I impedance of load to Wm I (16-36)
- R resistance of load to d-c
This shows the maximum degree of modulation that can be rectified with-
out distortion. When the degree of modulation exceeds the value given
by this expression, the negative peaks will be clipped.
16-6. Rectification Characteristics. It is clear from the discussion in
Sec. 16-4 that the signal output from the diode detector consists of a d-c
potential E a , which is the average value of the rectified carrier signal and
appears across the capacitor C, and an a-c term of modulating frequency,
this being a measure of the .amplitude of the envelope of the incoming
354 ELECTRON-TUBE CIRCUITS [CHAP. 16

modulated carrier. Thus one might consider C as an effective by-pass

for carrier-frequency currents, with two voltages across the output, a d-c
term and an a-c term of modulation frequency. The quantitative rela-
tionships among the amplitude of the applied r-f voltage, the average
rectified current, and the average anode voltage Ea are contained in the

Carner .---+---'r---~---.----,600·
vo/fage

H---j----lSOO

1--=1-----1400 ~
I-
Q)
0..

fh----l300
e~
u
'E
rt~~~~~~---P~~---l200u
a

_~C:::_j30~==_iW==~_I~O~~OO
Ea(rj-c volts)
FIG. 16-9. The rectification characteristic of a 6H6 diode, and the circuit used for
its determination.

rectification characteristic of the diode. A circuit for obtaining these

curves and the results on a 6H6 are illustrated in Fig. 16-9. In the cir-
cuit shown, the a-c voltage E is maintained constant, while the bias volt-
age E a , which simulates the drop across the load resistor, is varied, the
rectified current being read by the microammeter.
Also indicated on this characteristic are the load lines from which one
obtains the output voltage as the carrier voltage varies. Thus, by
plotting the current as a function of time as the carrier amplitude varies
because of the modulation, it is possible to determine both the output
and the distortion.
In addition to the several static load lines, there is also shown an a-c
load line, which represents a 0.25M static load that is shunted by a second
0.25 M load. The effective value of the unmodulated signal is 10 volts.
The resultant is the effective dynamic load into which the tube is work-
ing. In such a case, the load line will not in general pass through the
point (0,0), but may result in a curve such as that illustrated on the
rectification characteristic. Consequently, if the operation is carried to
the region of no current, severe distortion may result. The curves of
 SEC. 16-7] DEMODULATION 355

Fig. 16-10 show the character of the variation of distortion with per cent
modulation.
16-7. Automatic Volume Control. The average amplitude of the
modulated carrier wave that reaches the detector stage will depend upon
a number of factors, including the
field strength at the receiver of the
station to which the receiver is tuned
and the propagation conditions be-
tween the transmitter and receiver.
It is desirable therefore to incorpo-
rate some means within the receiver
for maintaining the average modu-
lated carrier amplitude at the detec- 20 40 60 80 100
tor at a constant level so as to avoid Per cent modulVltion
FIG. 16-10. Variation of distortion
the effects of fading. Such an
with per cent modulation. (From
automatic-volume-control (avc) cir- F. Langford Smith, "Radiotron De-
cuit will automatically vary the gain signer's Handbook," Chap. 18.)
. of the r-f or i-f stages to yield a sub-
stantially constant level at the detector. Figure 16-11 shows a block
diagram of a receiver incorporating automatic volume control.

FIG. 16-11. Block diagram of a receiver incorporating automatic volume control.

A-f.slqnal

SOOk

FIG. 16-12. A diode detector and automatic-volume-control circuit.

The details of a circuit in which a diode is used both as a linear detector

and also to supply a d-c voltage for automatic-volume-control purposes is
ill~strated in Fig. 16-12. In this circuit the Use of a separate isolating
reSIstor and capacitor filter permits the extraction of a d-c potential that
356 ELECTRON-TUBE CIRCUITS (CHAP. 16

is proportional to the average modulated carrier level. Also, the d-c

voltage across the diode load resistor is blocked off by the use of a resistor
and capacitor, the a-f component being made available for the following
a-f amplification stages.
The time constant of the automatic-volume-control network ReI is
made long enough to average out the variations in carrier amplitude
corresponding to the modulation, but short enough so that the automatic-
volume-control voltage varies with the average amplitude of the carrier,
dropping off as the carrier fades and increasing as the carrier becomes
stronger. This voltage is used to vary the bias, and hence the trans-
conductance of the r-f and i-f amplifier tubes, to maintain a substantially

0.05
t-----tf--:.

Delayed Ave
SOOk

FIG. 16-13. A diode detector and delayed automatic-volume-control circuit.

constant level. In this way, fluctuations in the average amplitude of

the modulated carrier delivered to the diode are greatly reduced. The r-f
and i-f amplifier tubes that have automatic volume control applied to
them should be of the remote-cutoff or variable-mu types; otherwise the
system is very critical of adjustment, owing to the very marked sensi-
tivity of transconductance with bias of the sharp-cutoff type of tubes.
If it is desired to have the automatic volume control operate only after
the carrier strength reaches a specified minimum, so that the reception
of weak signals will not be affected by the automatic-volume-control
system, a delayed automatic-volume-control circuit may be used. A cir-
cuit with delayed automatic volume control is shown in Fig. 16-13. In
this circuit the automatic-volume-control voltage is obtained from a
separate diode. Also, the automatic-volume-control voltage is obtained
from the output of the previous stage in order that a voltage-that across
th~ cathode resistor of the triode element of the duo-diode triode-may
be used as the reference level, below which no automatic volume control
is applied.
SEC. 16-91 DEMODULATION 357
16-8. Single-side-band and Suppressed-carrier Demodulation. It
was pointed out in Sec. 15-3 that the output from a balanced modulator
which was arranged to suppress the carrier contained the side bands of an
amplitude-modulated carrier. The transmission of suppressed carrier
signals is accomplished at high efficiency since the greater part of the
power of an amplitude-modulated wave is in its carrier, and with its
suppression a saving in power is effected. In the process of demodula-
tion, it is required that the carrier be resupplied at negligible amplitude
in the receiver. The requirement of generating a suitable carrier com-
plicates the design of the receiver. As a result, suppressed-carrier trans-
mission is not well adapted to radiobroadcasting.
The most difficult problem involved in resupplying the carrier is that
of maintaining the proper phase relation between it and the side bands.
Thus, unless the phase of the carrier is maintained at the zero difference
required relative to the side bands [see Eq. (15-6)], it can be shown that
the two signal components are out of phase with respect to each other,
their difference in phase being a function of the difference between the
phase of the original carrier and that resupplied by the receiver. The
phase difference in the signal components will give rise to a serious distor-
tion of the output.
If single-side-band demodulation is effected, and this requires only that
one of the side bands be eliminated by a suitable filter network in the
receiver before the carrier is resupplied, the problem is greatly simplified.
Now the output from the demodulator depends only on the supplied
carrier and the remaining side band and is independent of the phase of
the carrier, thus eliminating the phase effects discussed above. The
frequency of the resupplied carrier should be very nearly that of the
original carrier; otherwise the signal frequency will not be translated to the
proper zero position. However, a translation of several cycles from the
zero is not particularly objectionable.
If the unwanted side band is eliminated at the transmitter rather than
in the receiver, single-side-band transmission has the added advantage
over conventional amplitude modulation in effecting an additional saving
of radiated power and also a saving in band width required in the r-f
spectrum. Single-side-band transmission makes it possible, without a
sacrifice of audio fidelity, to space r-f channels twice as close as with
standard amplitude modulation. Because of these advantages, this
type of transmission is used extensively in point-to-point commercial
communication. 3
16-9. Mixers and Converters. A superheterodyne receiver incorpo-
rates a mixing element, a device in which the incoming modulated signal
is combined with the signal from a local oscillator in order to shift the
358 ELECTRON -TUBE CIRCUITS [CHAP. 16

carrier level from one frequency to another. The use of such a device
permits the carrier level of any signal to be shifted to a preset i-f value
and then to provide i-f amplification to bring the voltage level to the 10
volts or So desired at the input of the diode detector. The use of the
superheterodyne is widespread because it provides a higher selectivity
than a tuned r-f circuit, primarily because of the use of double-tuned
amplifiers.
If a separate local or beating oscillator tube is used, the tube in which
the combining is accomplished is called a mixer. If a multi grid tube is
used to serve both as the local oscillator and the mixing element simul-
taneously, it is referred to as a converter. In both cases the effectiveness
with which mixing is accomplished, i.e., the ratio of the i-f current in the
output to the signal-voltage input to the circuit, is an important quantity.
This quantity is called the conversion transconductance and is, by definition,
ii.f aib(i-f)
g =- =-- (16-37)
e - E' aee(,-f)

It should be noted that this quantity is quite different from the mutual
conductance of the tube.
16-10. Square-law Conversion. To examine certain of the aspects of
the conversion process, it will be supposed that the output from the local
oscillator is combined with the modulated carrier voltage, and this com-
bined voltage is supposed impressed directly on the grid of a square-law
tube. If the tube characteristic is represented by an expression of the
form
(16-38)
then with the application of the voltage
eo = Ec(l + m cos wmt) sin wet + Eo sin wot (16-39)
the following terms will appear in the output:

i = a [ E~ sin 2 wet + E6 sin2 wot + m:E2 sin 2 (We + wm)t

+ m:E~ sin2 (We - wm)t + 2EcEo sin wet sin wot
+ mE~ sin wet sin (We + wm)t + mE~ sin wet sin (We - wm)t
+ mEeEo sin wot sin (We + wm)t + mEeEo sin wot sin (we - wm)t
+ m;E; sin (We + wm)t sin (We - wm)tJ (16-40)

By the use of selective circuits, all terms will be eliminated except those
having frequencies in the neighborhood of Wo - We = Wi. Thus there will
remain in the output the following:
SEC. 16-11] DEMODULATION 359
i = a[EcEo cos Wit + YzmEcE o COS (Wi - wm)t
+ ~~mEcEo COS (Wi + wm)t] (16-41)
which may be written as
i = aEcEo(l + m cos wmt) cos Wit (16-42)
That is, the only signal that can get through the tuning coils which have
been tuned to the i-f frequency and which have a band width sufficiently
wide to accommodate the a-f spread, is essentially' the modulation
amplitude at the i-f frequency.
Because the tube which converts to the intermediate frequency often
operates under square-law fixed-bias conditions, it is frequently called the
"first detector." Actually it is a detector only in the sense that it permits
obtaining output frequencies that are different from the input frequencies
or, rather, that it shifts the frequency from the r-f level to the i-f level.
The problem of keeping the difference between the local oscillator and
the input r-f frequencips constant as one varies the antenna tuning from
a position corresponding to one end of the band (say 550 kc) to one
corresponding to the other end of the band (say 1,600 kc) is not a simple
matter if one wishes to adjust a single control. It requires careful con-
struction of the variable capacitors and the choice of constants so that
they "track" together. In general, perfect tracking is not possible over
the entire band. Ordinarily provision is possible for ensuring perfect
tracking at only three specific points. The errors over the intervening
ranges are not great enough to throw the beat frequency out of the i-f
pass band.
16-11. Generalized Conversion Theory.4 In the foregoing discussion
it was assumed that both the broadcast signal and the local oscillator sig-
nal were impressed on the grid of a square-law amplifier. Under these
circumstances the conversion is distortionless. Owing to the interaction
between the two circuits, electron
coupling is ordinarily employed in
modern practice, thus reducing
this interaction. Consequently
the only coupling is that through
the electron stream.
Under the assumption that the
signal voltage is small and that FIG. 16-14. Illustrating the variation of
the local oscillator voltage is plate-grid transconductance of a converter
tube with a large oscillator voltage on the
large, the signal electrode trans- grid.
conductance may be considered
as a function of the oscillator voltage only. Then the signal electrode-
plate transconductance gm may be considered as varying periodically at
the oscillator frequency. The situation is then somewhat as illustrated
360 ELECTRON -TUBE CIRCUITS [CHAP. 16

in Fig. 16-14. Because of the periodic variation that occurs in gm, this
may be represented by a Fourier series of the form
bo
gm = 2" + bi cos wot + b2 cos 2 wot + . . . (16-43)

When a small signal is applied to the tube, the resulting a-c plate current
has the form
(16-44)
This may be written in the form

i p = E' cos wet (~+ 2:

n
bn cos nwot)

or
bo E' cos wet
Zp. = 2"

+ E' (bi cos Wet cos Wot + 2: b cos nWot cos wet)
n (16-45)
n
nr'l

For a circuit tuned to the frequency Wo - We, the i-f frequency, the output
IS

.
2p(i-f)
bi
= 2" bi E' cos Wit
E' cos (Wo - We )t = 2" (16-46)

whence the conversion transconductance is

(16-47)

Upon combining this with the known form for bl , there results
1 (27r
gc = 271" J0 gm cos wot d(wot) (16-48)

The value of gc thus depends upon

the magnitude of the mutual conduct-
ance at the operating point and the way
in which this conductance is varied by
the local oscillator. Typical variations
for different types of tubes are illus-
FIG. 16-15. Typical transconduc- trated in Fig. 16-15. With the oscil-
tance curves of different types of lator bias near cutoff and with an oscil-
tubes.
lator amplitude to yield a high value of
instantaneous gm, the value of gc is approximately in the range from 0.25
to O.3gmmax , the maximum value of gm reached during the oscillator cycle.
 DEMODULATION 361
If an exact determination· of gc for a specific tube is required, this can
be accomplished by a numerical evaluation of the Fourier coefficient [Eq.
(16-48)].
REFERENCES
1. Everitt, W. L., "Communication Engineering," 2d ed., pp. 427-433, McGraw-
Hill Book Company, Inc., New York, 1937.
2. Terman, F. E., and N. R. Morgan, Proc. IRE, 18,2160 (1930).
3. Nichols, H. W., Elec. Commun., 2, 11 (1923).
Oswald, A. A., Proc. IRE, 26, 1431 (1938).
4. Peterson, E., and F. B. Llewellyn, Proc. IRE, 18,38 (1930).
Herold, E. W., Proc. IRE, 30,84 (1942).
Peterson, L. C., and F. B. Llewellyn, Proc. IRE, 33, 458 (1945).
PROBLEMS
16-1. A sinusoidal voltage is applied to the average detector of Fig. 16-2, with
Rl = lOOk. Determine the variation of the rectified output voltage with applied
signal.
16-2. The envelope detector of Fig. 16-3 has Rl = lOOk and C = 200 ppJ.
The impressed voltage is
e = 12 cos 271" X 1.5 X 10 6t + 1 cos 271" X 1.51 X 10 6t
Write an expression for the instantaneous voltage across C. Neglect tube drop
in the diode.
lS-3. A carrier is sinusoidally modulated 80 per cent by a 7.5-kc signal. The
signal input to the envelope detector of Fig. 16-3 with Rl = lOOk has a carrier
component of 12 volts peak. What should be the maximum value of C for no
distortion? Neglect the tube resistance.
lS-4. The input to the detector of Prob. 16-3 with C = 100 ppJ, Rz = lOOk,

IOOp.,sec

1..----------IOOO,asec------ - --:J,
Rg = 1M , C' = 00 is a pulsed r-f signal, the carrier frequency being 5 megacycles.
Sketch the wave form of the voltage across R g •
lS-5. In the circuit of Fig. 16-3, the diode efficiency is 0.90. Calculate and
plot as a function of the modulating frequency Wm the maximum degree of
modulation of the input signal e for distortionless rectification.
lS-S. Suppose that the d-c load resistance of a peak diode detector is 250 k
and 11 = 0.90.

a. Calculate and plot the maximum degree of modulation of a signal without

negative peak clipping as a function of the ratio of a-c to d-c load impedances
for ratios between 1 and 0.5.
b. If the signal is 100 per cent modulated, plot the approximate distortion
under these conditions.
362 ELECTRON -TUBE CIRCUITS [CHAP. 16

16-7. Determine from the rectification characteristics of a 6Ro diode the

largest permissible modulation without clipping if the d-c impedance of the load
is lOOk and the modulation frequency impedance is 50k• If the effective modula-
tion is 70 per cent with a carrier of 10 volts rms, determine the approximate
percentage distortion.
16-8. A superheterodyne as indicated in the diagram in block form gives for an
unmodulated carrier-frequency signal of 5 /l-V input to the r-f amplifier a detector

R-f amp Converkr I-Tamp

output of 5 volts. If automatic volume control is used on all tubes, what must
the carrier level be to give a 5-volt output; a 10-volt output? Assume that the
full d-c output voltage is used for automatic volume control.
16-9. A receiver is provided with a square-law detector, which is represented
by the expression
i = ke 2
Calculate the form of the output, and specify whether or not an intelligible a-f
signal results when the following are applied to the input:
a. One side band is eliminated from the transmitted wave.
b. The carrier is eliminated from the transmitted wave.
c. One side band and the carrier are eliminated from the transmitted wave.
16-10. A single-side-band transmitted wave is applied to a square-law detector
for which
i p = 0.5(6 + eg )2 ma
If the input consists of a supplied carrier of 2.5 volts peak and the side-band
amplitude is 1 volt peak, calculate the signal current in the output.
16-11. A 6L7 pentagrid tube is used as a converter. A plot of the curve of
plate-grid 1 transconductance as a function of grid 3 bias is sketched in the
,----,----,-----,1500

901
1---+.~-+---l500

~--=--__:_=----;! 0
-20 -10 0
EC3

diagram for this problem. A beating oscillator signal of 24 volts peak, which
causes a bias of -20 volts to be developed, is applied to grid 3. The incoming
signal of 100 jJ,V is applied to grid 1, which is maintained at a bias of -6 volts.
Calculate the conversion transconductance under these conditions.
 CHAPTER 17

FREQUENCY MODULATION AND DETECTION

CHAPTER 15 was confined to a discussion of amplitude modulation

(a-m), in which the amplitude of the transmitted wave is altered in a
manner dictated by the amplitude and frequency characterIstics of the
signal. It was there pointed out that, in addition to such amplitude
variations, it is possible to effect changes in the frequency or in the phase
of the transmitted signal. In particular, however, if intelligence is to be
transmitted, it is essential that the two features that characterize intelli-
gence, viz., amplitude or loudness, and frequency, must be available.
In frequency modulation (f-m) the transmitting frequency is varied by
an amount depending on the signal amplitude, and the signal frequency
determines the rate at which the variation takes place. In phase modula-
tion (p-m) the phase of the transmitted wave is shifted by an amount
that depends on the signal amplitude, and the rate at which this shift
occurs is made proportional to the signal frequency. In general, any
system that can transmit the two aspects of information required for the
intelligence could serve as an acceptable system of communication. A
variety of pulse systems have been devised which are satisfactory and
which possess certain advantages over the a-m, f-m, and p-m systems.
These are outside of the scope of this book and will not be discussed.
It should be particularly noted that the amplitude of the oscillations
in the f-m system is not involved in the actual process of transmitting
intelligence. Consequently, it is possible to make the system insensitive
to any amplitude-modulated disturbances. This is a particularly desir-
able feature since atmospheric and man-made disturbances are largely
amplitude-modulated. Owing to the difference in character between
a-m and f-m signals, it is possible to separate and extract the signal from
the interference.
17 -1. Basic Characteristics of Frequency Modulation. To examine
graphically the fundamental principles of frequency modulation, suppose
that a telegraph dot and dash are applied to an a-m and to an f-m system.
The results have the forms illustrated in Fig. 17-1. For the a-m system,
the frequency of oscillation remains constant, but the amplitude is zero
or a constant, depending upon the time in the cycle. In the f-m system,
363
364 ELECTRON-TUBE CIRCUITS [CHAP. 17

the amplitude remains constant, but the frequency changes from a value
f2 to a value h or fa when the signal is applied.
If the applied signal is sinusoidal and of frequency fm, the effect pro-
duced in an a-m system has the form illustrated in Fig. 17-2b and the
effect produced in an f-m system has the form illustrated in Fig. 17-2c.

,PM

FIG. 17-1. The primary features of a-m and f-m waves.

17 -2. Instantaneous Phase and Frequency. The general expression
for an unmodulated carrier wave is given by
Ce = Ee sin (wet + 0) (17-1)
In this expression the period of the wave is given by

)t

The quantity
+0=

~~~~~~
wet cp (17-2)
---'tttttttttItH+t-tttHtItt\i\illll¥'rf\-f\f-\ft-f+~~l'_~ t: is the total instantaneous" phase"
~1Li tff~ of the function. If the phase is
11 written as cp(t) , the value of the
voltage at any instant is represented
by the expression
1tI++I1t+++++tft#Ht1tll\------...-t Ce = Ee sin cp(t) (17-3)
But, clearly, the angular frequency
FIG. 17-2. The output of a sinusoidally is related to the phase by the ex-
modulated a-m and f-m transmitter.
pression
dcp
w = dt (17-4)
SEC. 17-3] FREQUENCY MODULATION AND DETECTION 365

This expression agrees with the usual definitions of frequency, and in the
unmodulated case
d
W = dt (Wet + 8) = We
17-3. Frequency Modulation (F-M). Frequency modulation is pro-
duced by varying the instantaneous frequency of a carrier by an amount
that is proportional to the amplitude of the modulating signal and at a
rate that is proportional to the frequency of the modulating source. The
amplitude of the carrier is assumed to remain constant in the process.
That is, if the modulating signal has the form
(17-5)
the f-m wave has an instantaneous frequency given by the expression
wet) = We + ktEm cos wmt (17-6)
The proportionality factor k i determines the maximum variation in fre-
quency for a given signal strength Em.
To determine the expression for the f-m wave, use is made of Eq. (17-4).
This requires t h a t ·
dcp
W = dt = We + ktEm cos Wmt
from which it follows that
cp(t) = lot W dt
which yields the expression
cp(t) = Wet + kt Wm
Em sin wmt + 8 0 (17-7)

The initial phase 80 is neglected in what follows, for it plays no part in the
modulating process. Thus, for the f-m wave,

e = Ee sin (wet + kt~: sin wmt) (17-8)

The instantaneous frequency of the f-m wave is

f = .; = fe + let t sin wmt (17-9)

which has a maximum value of

fmax = fe + kt ~; (17-10)
and a minimum value of
(17-11)

The maximum swing of the frequency from its mean value is called the
366 ELECTRON-TUBE CIRCUITS [CHAP. 17

frequency deviation and is denoted by h It is

fa == f-x - fe = fe - fInin = kf t (17-12)

By analogy with amplitude modulation, the modulation index is defined as

(17-13)

Also, the ratio of fd to the modulating frequency fm is called the deviation

ratio and has the value

(17-14)

In terms of these factors, the expression for the f-m wave assumes the form
(17-15)
17-4. Frequency Spectrum of F-M Wave. To examine the spectrum
of the f-m wave, it is necessary to expand the expression [Eq. (17-15)] that
represents the f-m wave. This is done as follows:
e = Ee[sin wet cos (0 sin wmt) + cos wet sin (0 sin wmt)] (17-16)
Use is now made of the following expansions,

cos C' ,in """t) ~ JoC:) + 2 .t, J,.(,) co, 2n""",!

(17-17)
sin (0 sin wmt) = 2 I
n=O
J 2n+ 1 (0) sin (2n + l)w mt
where the function J n( 0) is the Bessel function of the first kind and of
order n. The f-m spectrum then becomes
e = Ee sin wet[JO(o) +
2J 2 (0) cos 2wmt +
2J 4 (0) cos 4w mt + ... ]
+ Ee cos wct[2J 1 (0) sin wmt + 2J 3 (0) sin 3w mt + ... ] (17-18)
which may be written in the form
e = Jo(o)Ee sin wet
+ J 1 (0)Ee[sin (We + wm)t - sin (We - wm)tl
+ J 2(0)Eclsin (We + 2w m)t + sin (We - 2w m)tJ
+ J 3 (0)Ee[sin (We + 3w m)t - sin (We - 3wm)t]
+ ... (17-19)
where use has been made of the trigonometric expansions
sin x cos y = Yz[sin ex + y) + sin (x - y)] }
(17-20)
cos x sin y = Yz[sin (x + y) - sin (x - y)]
SEC. 17-4] FREQUENCY MODULATION AND DETECTION 367
The Bessel function J n( 0) is defined by the series
on [02 04
In(o) = 2nn! 1 - 2(2n + 2) + 2(4)(2n + 2)(2n + 4)

2(4) (6)(2n
~ + 4)(2n + 6) + ...
+ 2)(2n ] (17-21)

It follows from Eq. (17-19) that the spectrum of the f-m wave consists
of a carrier and an infinite number of side bands all of whose amplitudes
are various-order Bessel functions. Graphs of several of these functions
are contained in Fig. 17-3. It will be noticed that Jo(o) has a root at
1.0

0.8
'\
'(t=o BESSL FU~CTION~ OFT~E FIR~T KI~D I n 10)
0.6

V\ ~ ~ r.:!.
/ ) / ( 1\ l"- /" ~ r-..... r-..... - V- ~

V IV ~ V' K/
oV
L VK \ \ )
2\3 \ 5)\ 6~X 8/"
I
VIO"'~VK
"' - )

-0.2 /
)C -..-/ ~ p<....V "- -'" ' - - l-

-0.4 I\-.
-0.6
FIG. 17-3. Bessel functions of the first kind.
about 2.40. This means that the carrier will vanish when the frequency
deviation is equal to 2.4 times the audio frequency. This fact provides
a method for measuring the frequency deviation since the zero point of
the carrier can be observed by a selective radio receiver.
A list of the roots of various Bessel functions is given in Table 17-l.
TABLE 17-1
ROOTS OF In(a)

n=O 1 2 3 4 5

2.4048 3.832 5.135 6.379 7.586 8.780

5.520 7.016 8.417 9.760 11.064 12.339
8.654 10.173 11.620 13.017 14.373 15.700
11.792 13.323 14.796 16.224 17.616 18.982
14.931 16.470 17.960 19.410 20.827 22.220
18.071 19.616 21.117 22.583 24.018 25.431
21.212 22.760 24.270 25.749 27.200 28.628
368 ELECTRON-TUBE CIRCUITS [CHAP. 17

It is instructive to examine a particular situation in some detail. Con-

sider an r-f signal that is modulated ± 75 kc by an a-f signal of 7.5 kc.
The corresponding deviation ratio is 10.0. The important Bessel func-
tion that occurs in the expression for such an f-m wave is I n (10.0). A
plot of J n(10.0) as a function of n is given in Fig. 17-4. Notice that the
function falls off toward zero rapidly after the deviation 75 kc is passed
but that the amplitudes are significant out to about 14 X 7.5 = 105 kc.
In(/O)
0.3

0.2

0.1

8 10 12 16 n

-0.1

-0.3
FIG. 17-4. A plot of J,,(lO.O) as a function of n.

1
0.4

16
,'IIIIII.lllltllll.IIIIII',
12 8 4 -S L-w :In
4 8 12 16

FIG. 17-5. The spectral distribution in an f-m wave with ~ = 10.

A plot of the spectrum corresponding to Eq. (17-19) is given in Fig.

17-5. Notice that this is just a plot of the spectral lines, without regard
for sign, as dictated by the Bessel function plot of Fig. 17-4. The
separation between individual side bands is 7.5 kc, the modulating fre-
quency. It is evident from this plot that the total band width necessary
to include all significant side-bands is 210 kc.
Two similar plots are also included. In Fig. 17-6 is illustrated a series
of plots for constant modulating frequency fm but for various values of
frequency-deviation ratio 8. The incr-easing number of side bands with
SEC. 17-4] FREQUENCY MODULATION AND DETECTION 369

o is clearly seen. Likewise, the disappearance of the carrier is plainly

seen.
The plots of Fig. 17-7 show the spectra for a constant frequency devia-
tion but for various modulation frequencies. It should be observed that
the total band width required to include all significant side bands decreases
somewhat with increasing deviation ratio. For a given frequency devia-
tion jd, except for the very small value of 0, almost all significant side

I. 0= Wd =J. 0
UJm ."
IIIIIL

o.s

1.0
~-Uld-j

1.5 10.0 .,11111 11.1.1,1.11,111""

2.0 12.0 .,11111.11.11,1.1.11,11,1/111.

-~ ~-wd

2.5 1111.1111 16.~ 111111,11.11.1.111 11.1.11,11,1111111 .•

um-t-_ UJd---J
FIG. 17-6. The spectral distribution in an f-m wave for different values of 0, different
Wd, and fixed wm •

bands are contained within the range h The curve of Fig. 17-8 indicates
the number of significant side bands (those with amplitudes exceeding 1
per cent of the largest side-band component) in an f-m spectrum for
different values of nand o. In particular, if 0 = 5, then n must be about
8 for In(o) to be negligible compared with unity.
Some very important information is contained in Fig. 17-8. To
appreciate this, consider the present Federal Communications Commis-
sion (FCC) regulations on frequency. modulation. These regulations
specify
Maximum frequency deviation, jd = ± 75 kc.
Allowable band width, B = 200 kc (including a 25-kc band at each
end).
370 ELECTRON-TUBE CIRCUITS [CHAP. 17

Frequency stability of carrier, ±2 kc.

Remembering that the frequency deviation is related to the amplitude
and frequency of the modulating voltage according to the relation
Wd = kfEm
then for that amplitude which provides a frequency deviation of 75 kc
the band-width requirements increase with decreasingfm. But from Fig.

0.5

1.0

2.0

J.O

5.0 .III,ld,111.

IQO .. dill ILL! LlLlIIIr.

20.0 .• ,llIh.lh.h,I,II,"I.IIII,I.II.III.IIIIII ..

24.0 .•IIllIh.,iI.II,Ii.h.I.IJ,!.i.I,ILh,I,II,.Illh"
~-·aw-../
FIG. 17-7. The spectral distribution in an f-m wave for different values of 0, fixed Wdj

and different Wm.

17-8 it is clear that all of the significant side bands or energy is contained
withinfd for values of 0 = 10 or higher. For 0 lower than 10, the number
of side bands outside of fd begins to increase rather seriously. This occurs
for frequencies of 7.5 kc or higher. Note, however, that with ordinary
SEC. 17-5] FREQUENCY MODULATION AND DETECTION 371
broadcasts the amplitude of the frequencies in the range of 7.5 kc and
higher is considerably smaller than with the frequencies in the middle or
lower registers. Clearly, therefore, Em for these higher frequencies is
smaller than for the mid-frequency band, and hence the total active h is
materially less than the allowable 75 kc and the system is operating within
a narrower band width than might otherwise be required. In fact, the
band width is so much less than that allowed that it is found convenient
to incLude weighting, or preemphasis, networks in the transmitter. These
1000

500

/
100 1/
50
c5 I/,
.-':
/0/
10
1//
5
V
/
// V
V- I
2 3 4 ~ 10 2030 50 100 500 1000
n
FIG. 17-8. Values of n for a given value of IJ to make In(lJ) < 0.01. Note that n may
assume only integral values. (From L. Mautner, "Mathematics for Radio Engineers,"
Pitman Publishing Corp., New York, 1947.)

have such properties that they tend to accentuate the higher frequencies.
In this way, the higher frequency components, which ordinarily do not
contain much energy, are expanded beyond their natural level. This
proves to be a desirable practice in that it tends to give an improved
signal-noise ratio, for the noise generated within the tube circuits is
uniformly distributed over the frequency band. Of course, for fidelity in
reproduction, the receiver must hate networks which deemphasize the
incoming signal to yield the proper energy distribution. More will be
said about this when the circuits for effecting preemphasis and deempha-
sis are discussed.
17-5. Phase Modulation (P-M). Phase modulation is produced by
varying the instantaneous phase of the carrier at a rate that is propor-
tional to the modulating frequency and by an amount that is proportional
to the strength of this modulating signal. The amplitude of the carrier
372 ELECTRON-TUBE CIRCUITS [CHAP. 17

remains unaltered in this process. If the modulating signal has the form

the instantaneous phase of the wave is given by the expression

6 = 60 + kpEm sin wmt (17-22)
where k p is a proportionality factor that determines the maximum varia-
tion in phase for a given signal strength.
To find the analytic expression for the p-m wave, use is made ·of the
fact that the instantaneous phase is given by
(17-23)
Equation (17-3) for the present case, when written in full, becomes
(17-24)
The constant phase 60 is taken as zero, as it plays no part in the modulat-
ing process. This expression is written in the form
e = Ee sin (Wet + 6d sin wmt) (17-25)
where the maximum deviation in phase is
6d == kpEm (17-26)
A comparison of this expression with Eq. (17-15) for the f-m wave indi-
cates that the two forms are identical. Consequently the entire discus-
sion of the spectral distribution of the energy contained in an f-m wave
can be extended to p-m waves. Therefore the frequency spectrum of a
p-m wave having a maximum phase deviation of, say, 10 rad will be
identical in form with the frequency spectrum of an f-m wave having a
deviation ratio of 10.
There is one very significant difference between the f-m and the p-m
waves, however. This difference is contained in the form of the phase
deviation 6d and the deviation ratio 0 that appears in Eqs. (17-25) and
(17-15), respectively. The differences lie in the definitions of 6d and 0,
namely
6d = kpEm for p-m waves
o = kfEm for f-m waves
Wm
Clearly, for p-m waves the phase deviation depends only on the amplitude
of the modulating signal, and all modulating frequencies of equal Em will
possess equal values of 6d , independently of the frequency wm • As a
result, the spectral distribution will be the same in each case, although
the separation of the spectral lines will depend on the modulating fre-
quency. In particular, if it is supposed that the maximum phase devia-
SEC. 17-6] FREQUENCY MODULATION AND DETECTION 373
tion of a particular wave is 5 rad, there will then be approximately 8
significant side-band components present. If the modulating frequency
is 5 kc, then the band width is 2 X 8 X 5,000 = 80,000 cps. If the
modulating frequency for the 5-rad maximum deviation is 50 cps, then
the band width is 2 X 8 X 50 = 800 cps.
In the case of frequency modulation, if the value of Em is such that the
deviation ratio is 0 = 5 for a modulating frequency of 5 kc, then for an
equal Em at 50 cps the corresponding deviation ratio is 500. The result-
ing spectral distribution in these two cases will be altogether differ-
ent, there being 8 significant side bands for 0 = 5, and there being in

Modulafing
signal

FIG. 17-9. A simple reactance-tube f-m transmitter.

excess of 500 significant side bands for 0 = 500. The band width i::;
2 X 8 X 5,000 = 80 kc under the first conditions and is approximately
2 X 500 X 50 = 50 kc in the second case.
Owing to the simple difference in form between Bd and 0, it might appear
that it should be possible to use p-m waves to produce f-m waves. This
would be possible if one could arrange, by means of appropriate circuits,
to causp the apparent phase deviation to vary inversely as the modulating
frequency. Such circuits are possible, and the Armstrong method of
producing frequency modulation operates on this principle. The details
of this method will be discussed below.
17-6. F-M Transmitters-Reactance-tube Types. A variety of
methods for the production of frequency modulation exist, although they
do not all enjoy very great flexibility. In principle at least, the most
direct way of producing an f-m wave is to alter the capacitance in the tank
circuit of an oscillator. This might conceivably be done by incorporating
a capacitor microphone as part of the tank capacitor in an oscillator cir-
cuit. A considerably more satisfactory method, and one which accom-
plishes the same result in sUbstantially the same way, is to incorporate a
reactance tube in the tank circuit. Such an electronic circuit produces a
reactance, either inductive or capacitive depending upon its manner of
connection, which may be varied by varying the potential on one grid
of the tube.
The block diagram of a simple reactance-tube f-m transmitter is shown
374 ELECTRON-TUBE CIRCUITS [CHAP. 17

in Fig. 17-9. The essential features of certain of the elements of the cir-
cuit are examined in some detail below. In particular, the operation of
the reactance-tube circuit and the preemphasis circuit will be discussed.
17-7. The Reactance Tube. A simple reactance tube has features in
common with the circuit of Fig. 8-18. A schematic diagram and its
equivalent circuit are given in Fig. 17-10. It is desired to find the output

I
IZ
-r-
I
I

FIG. 17-10. The reactance tube and its equivalent circuit.

impedance of this circuit. This is readily accomplished. Note from the

equivalent circuit the following relations:

I
c = R1 -EPk!
jXc
with (17-27)
Xc=-
wC
Also, the grid potential is
E = RI = RE pk (17-28)
g C R - jXc
The plate current is
Ip = E pk + ILE gk = E pk + ILEa
rp rp
which may be expressed as
Ip = E pk +.!!:. RE~k (17-29)
rp rp R - JXc
The total current is then
I = Ip + Ie = Ep~ + E pk + (1m RE~k (17-30)
R -JXc rp R -JX c
The output admittance of the circuit is given by the relation

y = I = ~= 1. + lrp + (1m R R. (17-31)

Z E pk R - JXc - JXc
which may be expressed in the form

y= 1. +l+ 1 (17-32)
R - JXc rp 1 . 1
--J--
(1m (lmRCw
SEC. 17-7] FREQUENCY MODULATION AND DETECTION 375
This expression indicates that, in so far as the output circuit of the
reactance tube is concerned, it may be represented by the circuit of Fig.
17-11. Clearly, if the impedances rp and R - jXc are large compared
with (l/gm) - j(l/gmRCw) and if l/wCR is large compared with unity,
the output admittance becomes purely capacitive and arises from an
equivalent capacitor gmRC. ____ ____ ____-+____T _I_
~ ~ ~

Since the output impedance of the J.. I

I

reactance-tube circuit above may be rp Epk

made to appear as a pure capacitance l
gmRC, then if the modulating sig-
nal is made to vary the gm of the FIG.17-11. The equivalent output cir-
tube, and this is readily accomplished cuit of a capacitive reactance tube.
by applying the modulating signal to the grid of the tube, the effective
capacitance will then change with changes in grid potential. The circuit
of such a reactance-tube f-m oscillator is given in Fig. 17-12. Also

19'.m
I
6SJ7
({1m: JOOO+SOOec )
2000~
1500 b
1000'E~
500 ~
---'C-...I!-~--'---'O
c- Reactcmce tube Tuned plate osc.
o
FIG. 17-12. A reactance-tube f-m oscillator, and the curve showing the variation of
gm with grid potential.

G R
I
IZ
Ie
~ R9 C
I pEgJr
1
+
- 1
K
FIG. 17-13. An inductive reactance tube, and its equivalent circuit.

included is a curve showing the variation of gm of the tube with changes

in grid potential. .
A reactance tube may be connected to yield an effective inductance,
rather than an effective capacitance, across the output terminals. Such
a circuit, with its electrical equivalent, is given in Fig. 17-13. By pro-
ceeding in the same general manner as for the capacitive reactance-tube
376 ELECTRON-TUBE CIRCUITS [CHAP. 17

circuit, it can be shown that the equivalent output admittance has the
form
y = 1 (17-33)
R + Ry/jwC
Rg + l/jwC
for which an equivalent circuit exists. This equivalent circuit has the
r-- _ _ _. . -_ _ _-.--,----I--=I'--.- form illustrated in Fig. 17-14.
R
t By the proper choice of the vari-
IIZ i
: ous elements in the circuit, the cir-
.,-- 4k cuit reduces to the simple form,
Rg c CR: i comprising only an equivalent in-
'--_~--~_-+--""gm!!L.---I __---'--t ductance CR / gm.
FIG. 17-14. The equivalent output circuit A reactance-tube oscillator that
of an inductive reactance tube. incorporates an inductive react-
ance tube circuit as part of a Hartley oscillator is illustrated in Fig. 17-15.
Also included is a graph showing the variation of gm with changes in grid
3 potential.

6L7
(g;}IJ70+'l5Ec3 )

L- Re.,ct~nce tube H~rtley osci II ~tor

FIG. 17-15. An inductive-reactance-tube f-m oscillator, and the curve showing the
variation of (/m with grid 3 potential of the reactance tube.

An approximate expression for the variation of the frequency of a

reactance-tube oscillator as the voltage on the control electrode is varied
is readily possible. The transconductance may be expressed analytically
as a function of the potential of the control electrode. Note from the
curve that
Go
gm = Eo ec +G 0
(17-34)
But since
ec = Ecc + em
= Ecc + Em cos wmt (17-35)
then
gm = Go Ecc
Eo + G + Eo
0
Go E m cos Wm t (17 - 36) --E.~o~---~IE~--L-~~
cc
SEC. 17-8J FHEQUENCY MODULATION AND DETECTION 377

For the capacitive reactance-tube circuit, the effective output capacitance

IS

Ce = gmCR = GoCR (1 + ~: + ~: cos wmt) (17-37)

If it is assumed that the frequency of oscillation of the oscillator is that of

the tank circuit alone, then
1
f= .... ~
271" ·,/L o( C0 +C e)
1
f = . (17-38)
271" ~LoCo + LoGoCR (1 + ~: + ~: cos wmt)
The carrier frequency is evidently the value of the frequency of the oscil-
lator when the modulating signal voltage is zero. This is

fe = _~======1=====- (17-39)
2~Lo rCo + GoCR (1 + ~:) ]
The frequency ratio f /fe is given by
f _ 1 (17-40)
lc _. 11 + LoGoCREm cos wmt
'\j Lo[Co + GoCR(E o + Ecc)]
By expanding this expression by the binomial theorem and retaining only
the first term in the expansion, since the total frequency shift is small, then
:[ = 1 _ ! Em cos wmt (17-41)
fe 2 (CoEo/GoCR) + Eee + Eo
This expression may be written in the form

f = fc(l + mf cos wmt)

where the modulation index is
1 Em
mf = - - ~-=~~c=---,---co~-c----=-
l (17-42)

2 (CoEo/GoCR) + Eee + Eo
17-8. Preemphasis Circuits. As discussed in Sec. 17-4, there is a
relatively small amount of energy contained in the h-f portion of the audio
spectrum. As a result, the deviation at these high frequencies is far less
than the maximum allowable value of 75 kc. The corresponding band-
width requirement is correspondingly less than the allowable 150-kc
total. In fact, the relative h-f amplitudes are so low that it is customary
to include preemphasis networks in the circuit to accentuate the h-f
378 ELECTRON -TUBE CIRCUITS [CHAP. 17

terms as against the lower frequency terms. In this way the relative
signal strength at these higher frequencies is improved relative to tube
and circuit noise, which has a uniform distribution over the entire audio
spectrum. Of course, corresponding deemphasis must be incorporated
in the receiver in order to bring the relative amplitudes of all frequencies
to their proper levels.
Preemphasis circuits are chosen to satisfy the equation
1
(17-43)

The value of Wl was originally chosen to be l/wl = 100 f.Lsec, but it is now
taken as 75 f.Lsec. With such a preemphasis circuit the amplitude of a
2,100-cps signal is increased in the ratio 0/1 over the normal level, and
the relative amplitude of a 21-kc signal is increased in the ratio lOll.
Either an RL or a CR circuit may be used to accomplish preemphasis.
Two different circuits are illustrated in Fig. 17-16. In the RL circuit,

FIG. 17-16. Two different preemphasis circuits.

the voltage ratio cclcl is readily found to be
Cc jwL R/ L)
C1 R + jwL vI + 1(R/wL)2 / tan -1 (
- w
(17-44)

By writing Wl = RIL, this becomes

ICo I 1 (17-45)
~ e;: = VI + (Wt/W)2
Similarly, for the CR circuit, the mathematical development becomes
CC R1 (17-46)
C1 R/jwC
Rl + R + l/jwC
which is
(17-47)
R
R1 + jwCR + 1

At the higher frequencies R» X G or R» . IC' Eq. (17-47) then becomes

JW
1
R 1
Rl + JW
. CR 1 1+.JW CR 1
SEC. 17-9] FREQUENCY MODULATION AND DETECTION 379

By writing "'1 = 1iCR I

(17-48)

or

~
1 1= -yI'31=+c:=17("'=d=i=",=;=;)2 (17-49)

17-9. Frequency Stabilization of F-M Transmitters. Just as in the

case of the a-m transmitter, it is necessary that the average or carrier
frequency of an f-m transmitter be maintained very nearly constant, even
though the instantaneous frequency of the f-m transmitter varies with
the modulating signal. When a reactance-tube modulator is used to
modulate the carrier, the carrier cannot be crystal-controlled and the
average frequency will depend to some extent on the temperature, the
tube characteristics, and the various potentials. Slight drifts in the
operating characteristics of the reactance tube or slight changes in any of
the circuit elements will be accompanied by an appreciable change in the
average frequency. It is possible to minimize the effects of the drift in
the reactance-tube characteristics by employing two such tubes in a
balanced connection. Nevertheless the stability is not sufficient without
employing some type of stabilization to maintain the carrier frequency
within the 2-lm deviation specified by the FCC regulations.
Two basically different methods of stabilizing a reactance-tube modu-
lator are in present-day use. In both cases a standard reference fre-
quency is provided by a crystal-controlled oscillator, and the fundamental
or some subharmonic of the transmitter frequency is compared with this
reference frequency. Deviations between the two serve to actuate con-
trol circuits which operate in such a manner as to reduce these deviations.
The RCA and Federal schemes employ somewhat similar all-electronic
methods to effect the frequency stabilization, and the Bell Laboratories
method employs a frequency-sensitive servomechanism which drives a
small motor to which is geared the tuning capacitor, the direction of
rotation of the motor being determined by the relative frequency of the
transmitter and the reference standard.
The RCA method of stabilizing a reactance-tube f-m modulator 1 is
shown schematically in Fig. 17-17. In this circuit the frequency of the
reference crystal differs from the center frequency of the f-m transmitter
by some definite amount, say 1 megacycle. The two frequencies are
mixed, and the difference frequency is applied to a discriminator (the
operation of the discriminator will be discussed in Sec. 17-Hi). The d-c
output from the discriminator, which is a direct measure of the difference
frequency, is then applied to the grid of the reactance-tube modulator in
380 ELECTRON-TUBE CIRCUITS [CHAP. 17

such a manner as to make the net control voltage equal to zero when the
difference frequency is exactly 1 megacycle. Any frequency drifts can
then be appreciably reduced. Clearly, the method cannot yield perfect
stability, for unless there is a slight frequency difference, no control
voltage is applied to the modulator. However, the improvement that
results is sufficient to maintain the frequency stability within the FCC
regulations.
The Federal center frequency-stabilization system is shown schemati-
cally in Fig. 17-18. In this system the frequencies of the crystal oscillator

Oufpuf

FIG. 17-17. RCA method of stabilizing a reactance-tube frequency modulator.

Oufpuf

FIG. 17-18.

and the master oscillator are each divided to a common frequency and are
then combined in a balanced phase detector. The d-c output voltage,
which is a measure of the phase difference between the two oscillators, is
used to actuate the reactance-tube JIlodulator in a manner to lock the
oscillator mean frequency to that of the crystal reference frequency.
This system maintains the center frequency constant within about 1 kc.
The schematic diagram of the Bell Laboratories method of stabilizing
the frequency of an f-m transmitter2 is illustrated in Fig. 17-19. Here
the output is frequency-divided, and the resulting subharmonic is then
modulated by the output of the crystal-controlled oscillator in such a
manner as to produce two-phase beat currents. These currents are used
to operate a small synchronous motor to which is geared a tuning capaci-
tor, which is part of the f-m oscillator circuit. If the subharmonic
remains in synchronism with the crystal reference, the motor does not
move. If the carrier frequency drifts, the armature rotates, the direction
SEC. 17-10] FREQUENCY MODULATION AND DETECTION 381
of rotation being set so as to readjust the carrier to the correct frequency.
This method of stabilization proves to be very satisfactory and operates
over a very wide range of drifts to yield satisfactory results.
17 -10. F - M Transmitters-Armstrong System. The Armstrong phase-
shift method of obtaining frequency modulation incorporates a crystal-
controlled oscillator as a basic element of the system. Consequently this
system has an intrinsic stability as good as that of its crystal, and no
additional frequency stabilization is required.

FIG. 17-19. The Bell Laboratories method of stabilizing a reactance-tube frequency

modulator.

Before the specific features of this method of producing frequency

modulation are examined, it is pertinent to examine the spectrum of p-m
waves with small value of maximum phase deviation. For the particular
case where (Jd = 0.5, which is illustrated in Fig. 17-6, the significant terms
depend on the following values of Bessel functions,
J 0(0.5) = 0.9385
J 1 (0.5) = 0.2423
J 2(0.5) = 0.0306
I n (0.5) == 0 for n >2
and the modulated wave has the explicit form
e = Ee sin (Wet + 0.5 sin wmt) (17-50)
which is
e = 0.9385Ee sin wet +
0.2423E e[sin (We +
wm)t - sin (We - wm)t]
+
0.0306Ee[sin (we +
2w m)t +
sin (we - 2w m)t] (17-51)
Note, however, that the second side-band components are quite small and
that the expression may be written approximately as

e == 0.9385 {Ee sin wet + ~ [sin (We + wm)t - sin (We - wm)t]} . (17-52)

Clearly, this expression will be more accurate for values of (Jd less than 0.5.
382 ELECTRON -TUBE CIRCUITS [CHAP. 17

Now consider the corresponding a-m wave, having the form

(17-53)
which may be written as

e = Ee sin wet + ~a [- cos (We + wm)t + cos (We - wm)tj (17-54)

Note specifically that, if ma = (Jd, the only essential difference between the
a-m and the p-m waves is in the relative phase of the carrier and the side
bands. Evidently for small values of (Jd, if the side bands of the a-m
wave can be shifted by 90 deg with respect to the carrier, a p-m wave
results. It is immaterial, of course,
whether the phase of the carrier or the
phase of the side bands is shifted in
order to achieve the p-m waves.
The process here discussed can be
given graphically in a manner that is
quite illuminating. It was shown in
FIG. 17-20. The sinor representation Sec. 15-1 that amplitude modulation
of an a-m wave.
could be represented by means of a
sinor diagram. In this diagram, the carrier potential is represented by a
fixed sinor, and the side-band components are represented by two sinors
which rotate in opposite directions. This sinor representation is redrawn
for convenience in Fig. 17-20. The resultant vector e represents the a-m
wave at any instant.
A corresponding sinor representation of the process of p-m production
is possible. Here, as shown in
Eq. (17-52), the carrier must be
shifted in phase by 90 deg rela-
. tive to the side bands. The re-
sultant sinor diagram then has
the form shown in Fig. 17-21. It
is evident from this diagram that
a phase-modulated wave does re- FIG. 17-21. The sinor representation of a
suIt. Moreover, since (Jd is p-m wave.
chosen to be small, the amplitude
variations that result in this process are very small. These variations
can be eliminated by the use of amplitude limiters, although these are
not found necessary.
A block diagram of an Armstrong type f-m transmitter is given in Fig.
17-22. 'The essential features of the system are the following: A stabilized
200-kc primary-frequency oscillator is used to control the mean or carrier
SEC. 17-11] FREQUENCY MODULATION AND DETECTION 383

frequency of the radiated wave. Part of this 200-kc signal is mixed in a

balanced modulator with a signal representing a frequency-distorted
version of the audio signal, the predistorted signal being such that the

FIG. 17-22. Block diagram of an Armstrong f-m transmitter.

amplitude is made to vary inversely with its frequency. The output

from the balanced modulator is the amplitude-modulated side bands with
the carrier-frequency component miss-
200*
ing. The modulation products are
shifted through 90 deg in phase and are R
,_~f
tI
then combined with the carrier in the C ~C' R1 1M ~c
I
combining buffer amplifier. The re- ,;
sult is an f-m wave, which has been FIG. 17-23. A predistorter circuit.
achieved from the p-m wave, the phase
deviation of which has been made to vary inversely with the modulating
frequency. The resulting frequency modulation is multiplied in fre-
quency until it is brought to the desired frequency level for final amplifi-
cation and transmission.
17 -11. Pre-distorter Circuit. The function of the predistorter circuit
is to provide an output voltage the amplitude of which varies inversely
with the frequency of the input voltage. A circuit which achieves the
desired results is given in Fig. 17-23. The ratio of the output to the
input voltage is given by
384 ELECTRON-TUBE CIRCUITS [CHAP. 17

RI/jwC Rl
e. R l + l/jwC 1 +jwCR 1 1
(17-55)
el R + RI/jwC Rl 1 + R 1 + jwCR
Rl + l/jwC R + 1 +jwCR 1 Rl
or
ec 1
(17-56)
el (1 + R/Rl) + jwCR
This ·becomes, for the specific circuit
constants indicated on the diagram,
1
1.02 + j2.51f (17-57)

Note particularly that for all fre-

quencies in excess of 50 cps the re-
sults are given within 1 per cent by
the expression
ec == 1 (17-58)
el j2.51f
17-12. F-M Transmitters-The
General Electric Phasitron. A cut-
away sketch of the General Electric
GL-2H21 phasitron is shown in Fig.
17-24. It consists of a cathode, an
electrostatic focus and deflection
system, and an anode structure.
The electrons that are drawn from
the cathode surface to the anode as-
sembly are acted on by the focus ele-
ments to form a tapered, thin-edged
disk, whose axis is the cathode and
whose focus is at anode 1 of the
anode assembly. The deflection
system consists of 36 rigidly mounted
elements whose active portions lie in
a radial plane below the electron disk
and a solid neutral plane located
above the disk. Every third deflec-
tor is connected together and to one
FIG. 17-24. The General Electric 2H21 phase of a three-phase excitation
phasitron. source. The three-phase v 0 I tag e
source comprises a crystal oscillator and phase-splitting network.
~EC. 17-12] FREQUENCY" MODULATION AND DETECTION 385
The action of the deflection system is such that portions of the electron
disk are deflected above or below the normal plane by the magnetic fielq.
of the three-phase system to form a sinusoidal edge. The appearance of
the disk is clearly illustrated in Fig. 17-24. The disk may be considered
as rotating at a rate determined by the crystal oscillator.
Anode 1 is a cylinder with 24 holes punched alternately above and below
the normal plane of the disk. Electrons striking the surface of the
cylinder are collected by it, while those that pass through the holes are
collected by the solid anode 2. Figure 17-25 shows a developed portion

FIG. 17-25. A developed portion of anode 1.

S/qna/

I-phase fo
J-phase r-f't'lmp. xfal
nefwork o,sc//lafor

Mulfipl/ers Oufpuf
and I-----'::.=='-~

power amp.
FIG. 17-26. The General Electric phasitron f-m transmitter.

of anode 1. The solid sine curve represents the edge of the electron disk
at the time when the maximum number .of electrons pass through the
openings to anode 2. The dotted curve shows the situation one half cycle
later, and almost no electrons pass through the openings to anode 2. If,
therefore, the two anodes are connected to opposite ends of a resonant
circuit, the circuit will be excited at the crystal driving frequency and in
a time-phase sense that is determined by the phase of the anode-current
pulses.
Frequency modulation of the resonant anode circuit is produced by
phase modulation of the electron disk. This is accomplished by applying
the audio signal to a solenoid which surrounds the phasitron. The axial
magnetic field that is so produced causes the electron disk to be advanced
or retarded about its axis relative to its zero signal position. Conse-
quently the phase of the oscillator is shifted, with a resultant production
386 ELECTRON-TUBE CIRCUITS [CHAP. 17

of phase-modulated waves. Moreover, since the magnetic field is pro-

duced by a solenoid which is essentially a pure reactance at audio fre-
quencies, then, for a constant voltage input, the current, and hence the
magnetic field that is produced, will vary inversely with the frequency of
the impressed voltage. Clearly, therefore, the output from the oscillator
is an f-m wave.
A schematic diagram of the phasitron f-m transmitter is given in Fig.
17-26.
17 -13. F -M Receivers. The basic circuit of an f-m receiver is some-
what similar to that of an a-m receiver of the superheterodyne type.
However, there are a number of significant differences in the two receivers.
The required band width in the f-m receiver is larger than that for a-Ul
reception, which requires that the frequency converter and the r-f and

Input /-1'" amp AudIO

R-f" amp f!§:!2i Mixer ~ and !->- DISCTlininO/1 f"requency
8-I08 MC limiter ampl/fier
1~:t'3.7Mc
Beat
frequency
osc.

FIG. 17-27. Block diagram of a typical f-m receiver.

the i-f amplifiers must be designed for this broader band width. Also,
the last i-f stage of the f-m receiver is operated as a limiter, thus eliminat-
ing any fluctuations in the amplitude of the i-f carrier, however produced.
The other outstanding difference is in the circuit used to demodulate the
f-m carrier. The f-m discriminator that is used to convert from frequency
modulation to amplitude modulation does not appear in an a-m receiver,
and also the detector, while it uses conventional diode circuits, operates
somewhat differently in the f-m circuit. The operation of this will be
examined below.
A block diagram of a typical f-m receiver is shown in Fig. 17-27. Such
a receiver must provide a high r-f gain in order to permit high sensitivity
with amplitude limitation. Also, it is necessary to use a relatively high
i-f frequency in order to permit the necessary 225-kc band width. In
addition, the high intermediate frequency has the feature that the image
signals fall outside of the tuning range. In particular, in the block
diagram shown for use in the range from 88 to 108 megacycles, the image
frequencies lie in the band 115.4 to 135.4 megacycles.
17 -14. The Limiter. It is the function of the limiter to remove any
amplitude modulation that might exist in the signal. These fluctuations
in the amplitude of the i-f carrier might have been produced either by
variations in the transmitting conditions or by man-made or natural
SEC. 17-15] FREQUENCY MODULATION AND DETECTION 387

static. Such a circuit, which usually operates on the nonlinear portion

of its characteristic, provides an output voltage that is sensibly indepen-
dent of the amplitude of the input voltage. Such limiter action is easily
secured by operating the plate of a tube at a very low plate potential;
by using a high series-grid resistor, by using a low screen potential, or by
a combination of these three. Other limiters are discussed in Chap. 7.
As ordinarily used, the last i-f stage of the f-m receiver is usually
operated at low screen and low plate potential, in a circuit of the type
illustrated. The general character of the gain curve is also included.

( --
I
20 k JOk
1
B+(SO-90 V) 10 20 30 40
Input, rms
FIG. 17-28. The circuit of a limiter, and the general character of the results.

Such a circuit as that illustrated will saturate at about 10 volts input to

the grid, and at this point the stage has a gain of approximately unity.
17-15. The Discriminator. In demodulating an f-m wave, the method
generally used is first to convert from frequency modulation to amplitude
modulation and then demodulate the amplitude modulation by conven-
tional methods. The circuit that is used to make this conversion is known
as a discriminator. A variety of such circuits are possible, and several
will be considered in some detail.
The simplest form of discriminator comprises an ordinary resonant
circuit that is tuned to a frequency that differs somewhat from the average
frequency of the f-m carrier. The action is made clear by reference to the
diagram of Fig. 17-29. Clearly, as the carrier frequency fluctuates, the
current in the detuned circuit varies, increasing as the impressed fre-
quency approaches the resonant frequency of the circuit, and decreasing
as the impressed frequency departs from the resonant frequency. The
output from such a circuit is an a-m wave. However, since the side of
the simple resonance curve is not linear, the amplitude-modulated output
is distorted.
The linearity can be greatly improved by using two "off-tuned" or
"stagger-tuned" circuits instead of one and then choosing the difference
between the two outputs. Such a stagger-tuned discriminator circuit 3
is illustrated in Fig. 17-30, In this discriminator, the input A is tuned to
388 ELECTRON-TUBE CIRCUITS [CHAP. 17

the carrier frequency leI circuit B is tuned to a frequency that is somewhat

higher than fe, and circuit C is tuned to a frequency that is somewhat
lower thanfe. The a-m output from such a circuit is without appreciable
distortion, owing to the linear resulting characteristic around the point

e e
AM

-+-----~t -~--~~------~f

FIG. 17-29. A simple antiresonant circuit operating as a discriminator.

FM::n
rM .r--"1~-,
R2 ,i
~~--------~-4-~I·-b'
FIG. 17-30. A discriminator employing two stagger-tuned circuits.

fe. Such circuits suffer from the fact that reception is possible at three
points, corresponding to each outer portion of the resonant curves and
also to the center or desired linear operating region. The response from
such a circuit is illustrated in Fig. 17-31.
As seen in the diagram, the output from each circuit is passed through
a diode detector of the envelope or peak detection type. The capacitors
e C1 and C2 are equal and have neg-
ligible reactance at the carrier fre-
I ,c /~\B
I \ 1/ \ quency. The resistors R1 and R2
/
/ \)(
/1'
I \ "-
are equal and are quite large.
/' _ ~ I c.....1t;- . . . , The d-c potential across CIR I is a
-----Y-L--..,I'--'---P'-------__ f measure of the amplitude of the
output from circuit B, and the d-c
potential across C2R 2 is a measure
FIG. 17-3l. The a-m output from the dis- of the amplitude of the output
criminator of Fig. 17-30.
from circuit C. Also, the total
output across a'b' is then a measure of the difference between the outputs
from circuits Band C and has a form like the resultant curve CR of Fig.
17-31. By careful adjustment of the circuit constants, and jf the fre-
SEC. 17-15) FREQUENCY MODULATION AND DETECTION 389

quency deviation is limited to the range between h and II, the rectified
voltage is an approximately linear function of the impressed frequency.
Another commonly used type of discriminator circuit 4 is shown in Fig.
17-32. It is possible to show that this circuit is substantially a stagger-
Cj=SOppf

J o-,--+--.

Lf2m h
FIG. 17-32. A center-tuned discriminator circuit.

tuned pairs and that the results illustrated in Fig. 17-31 also apply for
this case.
A limited analytic solution of the operation of the center-tuned dis-
criminator circuit is possible. Consider the circuit comprising C1L sC4 •
At the i-f frequency, assumed to be 4.3 megacycles in this circuit,
10 12
X 01 = 211" X 4.3 X 10 6 X 50 = 800 ohms
X 04 = 400 ohms
X LS = 27r X 4.3 X 10 6 X 2 X 1O- s = 50k

The voltage across L3 is then .

EL == J'X L3 E I2 == E12 (17-59)

jX L3 - j(XOI + X 04 )
Consider now the mutually coupled circuit. If the mutual inductance
is small, the impedance coupled into the primary circuit is small and
approximately

which becomes, for high-Q coils,

I == E12 (17-60)
jX L1
The voltage induced in the secondary is then

E s• c = ±jwMI = ± ~ E12 (17-61)

If the loading effects of the diode rectifiers are neglected, then

390

where
E -
ELECTRON-TUBE CIRCUITS

-jXC2
ab - R2 +jX L2 - jX C2 -

. 1
_ - jX c2 (M/L 1) E
+ R2 +jX 2 12
I [CRAP. 17

(17-62)
X2 = WL2 --C
W 2

Note now that the output d-c potential Ea,o is proportional to the peak
of the envelope of E ao , and correspondingly the output voltage Eb,o is
proportional to the peak of the envelope of E bo. The total output d-c
potential is
Ea'b' = Ea,o + E ob, = Ea,o - Eb,o (17-63)

l
However, the a-c envelopes can be represented in terms of the potentials
Eab and E 12 , namely,

Eao = + EL =
Eac Eac+ E12 = ~ab + E12
-Eab 17-64)
Ebo = Ebc + EL = -Ecb + E 12 = -2- + E12

Consider the situation when the instantaneous frequency equals the

carrier frequency. At this frequency, the secondary circuit is resonant,
and the quantity X 2 is zero. The voltage Eat is then given by the fol-
lowing expression, obtained from Eqs. (17-62). The positive sign is
chosen arbitrarily.

Eab = j (~:2 ~) E12 = j C~) E12 YIO (17-65)

The voltages Eao and Ebo then have the amplitudes and phase somewhat
as illustrated in the accompanying sinor diagram. Note that since

~~--------~IEL=~2
I

and

then
Ea'b' = 0
When the instantaneous frequency is greater than the carrier frequency,
the secondary reactance X 2 is positive and Eqs. (17-62) may be written
SEC. 17-15] FREQUENCY MODULATION AND DETECTION 391

in the form
Eab = j C~L) E12 R2 ~ jX 2 = j C~) E12Yj-O (17-66)

The corresponding sinor diagram has the form shown.

Here, since

then

and it follows that

Ea'b' is positive
When the instantaneous frequency is less than the carrier frequency,
X 2 is negative and Eqs. (17-62) become '

Eab = j ( CML) E12 R

W21
1 'X = j (
2-)2
CML) E 12Y/O-
W21
(17-67)

The corresponding sinor diagram in this case is as shown. In this case

and

so that
Ea'b' is negative
The above gives a good representation of the action of such a center-
tuned discriminator over the linear range. Owing to the approximations
that have been made, it does not represent too well the action over the
entire range.
REFERENCES
1. Crosby, M. G., RCA Rev., 6, 89 (1940).
2. Morrison, J. F., Proc. IRE, 28, 444 (1940).
392 ELECTRON-TUBE CIRCUITS [CHAP. 17

3. Travis, C., Proc. IRE, 23, 1125 (1935).

4. Foster, D. F., and S. W. Seeley, Proc. IRE, 25, 289 (1937).
5. Arguimbau, L. B., "Vacuum Tube Circuits," pp. 486-494, John Wiley & Sons,
Inc., New York, 1948.
Sturley, K. R., Wireless Eng., 21, 72 (1944).
PROBLEMS
17-1. Determine and plot the instantaneous frequency corresponding to each
of the phase functions in the figure.

,./

17-2. A100-megacyclef-msignalismodulated ±75kcata400-cpsrate. Write

an expression for the instantaneous voltage if the signal amplitude is 10 volts and
if both the frequency and instantaneous magnitude are a maximum at t = 0 sec.
17-3. A wave is frequency-modulated at an audio rate of 5,000 cps. If the
frequency deviation fa is 75 kc, plot the spectrum of the wave, including all
significant side-band components.
17-4. li. The amplitude of a 15-kc wave causes a 75-kc frequency deviation of
an f-m wave. Plot the spectrum, and calculate the band width required to pass
all side bands of appreciable magnitude.
b. Suppose that the amplitude is altered to give deviation ratios of 3 and 1.
Repeat part a for these two cases.
c. From these results, estimate the value of the deviation ratio that may be
used to be within the FCC limitations of ±75-kc-frequency band spread.
17-5. Consider the reactance-tube circuit shown in the diagram. Show that
by choosing L properly Z is capacitive and is given by C/(}mR without approxi-
mation. Compare with Eq. (17-32).

I
I
+-z
I
I
 FREQUENCY MODULATION AND DETECTION 393

17-6. Given the reactance-tube circuit in the figure. Show that, if the
reactance of C is negligible at the operating frequency and if CJJL» R, the
effective input impedance results from an inductance L/YmR.

I
I

--+-z
I
I

17-7. Consider the reactance tube and oscillator shown in the diagram for this
problem. The oscillator is to operate at 5 megacycles with a frequency deviation

of ± 10 kc. What change in Ym is required to achieve the desired frequency

moduli tion ?
17-8. Assume that between the limits of -2 and -6 volts the ym-ecl charac-
teristic of a 6SJ7 tube may be represented by the expression
Ym = 3,000 + 500ec J.Lmhos
This tube is connected as a reactance modulator, as illustrated in the figure. It is
desired to have a center or carrier frequency of 5 megacycles and a frequency

deviation of 7.5 kc. Determine the correct setting of the oscillator tank capaci-
tor and the required modulating voltage.
17-9. Carry out the analysis to show that Eq. (17-33) does give the equivalent
output admittance of the inductive reactance circuit.
17-10. Obtain an approximate expression for the variation of the frequency of
an f-m oscillator as the voltage on the grid of an inductive reactance tube is
',J'
394 ELECTRON-TUBE CIRCUITS [CHAP. 17

varied. Proceed in a manner analogous to that employed in the text in obtaining

the corresponding expression for a capacitive reactance tube.
17-11. Calculate and plot the voltage ratio for the preemphasis or accentuator
circuit, shown in the diagram, as a function of frequency. Plot the curve on
semilog paper.
0.00/

17-12. Calculate and plot the voltage ratio for the deemphasis circuit, shown
in the accompanying figure, as a function of frequency. Plot the curve on semi-
log paper.
lOOk
"/WINv
~
,
I

e1
I
I
!-OOI
+,
I

eo
I
I I
t t
17-13. Determine the maximum frequency deviation possible with the Arm-
strong system of Fig. 17-22 if the distortion is to be less than 6 per cent. The
phase varies at a 60-cps rate.
17-14. What must be the ratio of side-band to carrier voltages in the Armstrong
system to produce a frequency deviation of ±12.2 cps at the audio frequency of
400 cps?
17-15. Show by an analysis similar to that which leads to Eq. (16-6) for the
a-m case that the output of a frequency doubler to which an f-m signal is applied
is the f-m signal centered about the second harmonic of the carrier.
17-16. An f-m wave of the form given by Eq. (17-15) is combined with a large
sine wave Eo sinwot. Show that, if both waves are applied to a rectifier, the output
will contain the f-m wave shifted in the frequency scale.
17-17. An antiresonant circuit consists of a capacitance of 65 J.IpJ and an
inductance of 0.4 mh and 16 ohms. It is to be used to receive a wave having a
frequency modulation of 1.5 kc. What should be the value of the carrier fre-
quency? Estimate the percentage modulation of the output.
17-18. Suppose that a discriminator as shown in Fig. 17-30 comprises two
circuits which have band widths of 200 kc and are tuned approximately to 4.7
megacycles. Plot the resultant discriminator characteristic for the following
separation of the resonant peaks: 150, 200, 250, 300 kc.

!.
 CHAPTER 13

RELAXATION OSCILLATORS

IN ADDITION to the feed-back and negative-resistance oscillators that are

discussed in Chap. 12, there is a class of oscilllittors that is referred to as
relaxation oscillators. This type of oscillator, like the feed-back oscillator,
is provided with a feed-back loop. However, in the case of the feed-back
oscillator the amount of feedback is usually so adjusted that a substan-
tially sinusoidal output voltage results. In the relaxation circuits, the
feed-back voltage is very large-so large, in fact, that the tube is driven

FIG. 18-1. The output-voltage wave forms in the multivibrator and in the blocking
oscillator.

beyond cutoff. The tube remains cut off for a time determined by the
time constant of the elements in the grid circuit, after which the grid
recovers control of the circuit. Because of this operation, a seriously
distorted output results. But as such a distorted wave is rich in har-
monics, it may be used as a harmonic source. More often, however,
these waves serve to provide wave shapes that possess direct applications.
Among the important relaxation oscillators are the multivibrator, a
device that provides a sensibly square wave in the output, and the block-
ing oscillator, a device that provides relatively narrow pulses. The
general character of the wave shapes at the output of these circuits is
illustrated in Fig. 18-I.
Owing to the nonsinusoidal wave forms present in the circuit and the
fact that cutoff exists for an appreciable portion of the cycle, the mathe-
matical analysis of the general feed-back circuit is no longer applicable to
a description of the operation of relaxation oscillators. In fact, owing to
the cutoff that occurs, a substantially transient analysis must be made.
Details of the operation of several types of circuits will be given below.
395
396 ELECTRON-TUBE CIRCUITS [CHAP. 18

18-1. Plate-Coupled Multivibrator. The multivibrator was first

described by Abraham and 'Bloch 1 in 1918. The circuit is shown in
Fig. 18-2. It will be observed that this circuit is essentially that of a
simple two-stage resistance-capacitance coupled amplifier, with the out-
put of the second stage coupled through a capacitor to the grid of the
first tube. As a result, any signal on the grid of thp first stage will be
amplified by the two-stage RC amplifier, and the output signal is in phase
with the input signal on the grid of the first stage. Because the output
of the second stage is of the proper polarity to reinforce the input signal,
positive feedback results and oscillations can take place.
The operation of the circuit is substantially the following. When the
anode current in one of the tubes, say Tl, is increasing because of a'
positive-going signal on the grid, a negative signal is being applied to the
grid of the second tube T2. As a result, the current in T2 is reduced, and
the output signal is positive. This positive signal, which is coupled back
to Tl, causes the grid of Tl to become more positive, thus increasing the
anode current of Tl still further. This effect is cumulative-the current
in Tl reaches a maximum, while the potential of the grid of T2 is driven
so far negatively, almost instantaneously, that the current in T2 is cut
off. With T2 still cut off, the
charge on capacitor C 2 leaks away
through the resistor R g2 , and at
some point the potential of the
r.-~tt---- ~1" Vy grid of T2 becomes such that the
< - '!. tube will again begin to conduct.

This results in a negative output

FIG. 18-2. A free-running plate-coupled
signal being applied to the grid of
multivibrator. Tl. The current in Tl decreases,
and the start of a cumulative
chain, which results in Tl being cut off and T2 reaching the state of maxi-
mum conduction, is initiated. Evidently there are two unstable limiting
conditions that occur. In one of these Tl is cut off and T2 is fully con-
ducting, and in the other the roles of Tl and T2 are interchanged.
18-2. Detailed Explanation of Operation. 2 Consider the circuit from
the instant when the cumulative action discussed above has caused the
grid of T2 to be driven beyond cutoff. The circuit of Fig. 18-2 is redrawn
for convenience. Just before the switching occurs, the voltage across
the capacitor C1 IS El and is given by
(18-1)
where Eb2 is the drop across tube T2 when it is conducting and EOl is the
cutoff potential of Tl and the grid potential of Tl at which current will
just begin to flow in Tl.
SEC. 18-2] RELAXATION OSCILLATORS 397

When T2 is cut off, this tube is effectively removed from the action of
the circuit and remains out of the circuit until the grid potential has
recovered to the cutoff value E o2 , when the tube will again begin to draw
current. During the cutoff period, there will be no plate current through
-
the plate resistor Rz;'---;iid-- thi~- plate will assume the potential of the
--

. . . . - t-----___ .r--=-----;--,.-
IiOJl '---'-IIf-+-.--f---- ti
62 j
Ebi T2 E62
,
,
,
,
I

I
,
,
,
I

~'_____L--------'-_--'---.L.L
FIG. 18-3. The plate-coupled multivibrator, with voltages and currents labeled.

I
I
I
eCl
I
I
I

tCIG. 18-4. The charging path, and the equivalent circuit for charging capacitor C 1•

B-suppl:y s<?llice E bh • Hence, during the cutoff period of T2, the capacitor
C1 begins to charge toward Ebb. Since the clw,rge on C1 cannot change
instantaneously, the sudden rise e,
in plate voltage will appear on
the grid of Tl. Thus T1 is
caused to conduct its maximum
plate current. But as the grid of i ct
T1 becomes positive with respect E6i;Eh2+EoJ
to the cathode, grid current will
flow. Thus C1 begins to charge
toward Ebb through the pat.h
shown in Fig. 18-4., In fact, since
the resistance !b. of the internal
grid-cathode path is smaller than FIG. 18-5. The wave shapes of the current
the--;esistor R g1 , capacitor C1 will through and the potential across capacitor
C 1 during its charging period.
charge mainly through this path.
Since rc of the internal grid-cathode path during conduction is small
compared with R g1 and, in fact, is also small compared with R Z2 , the charg-
ing time constrult-i§..almroximatelx CJ;W2. During the charging process
398 ELECTRON -TUBE CIRCUITS [CHAP. 18

the potential el, across the capacitor 0 1 will vary between -:- EI?J-~Ol
and Ebb, where Eb2 is the voltage across T2 when this tube is conductiiig.
The voltage and current conditions have the forms illustrated in Figs.
18-5 and are governed approximately by the equations
t
e1 == Ebb - (Ebb - Eb2 + E01)e - R12C. (18-2)
and

(18-3)

The process illustrated in Fig. 18-5 continues in the manner illustrated

until the switching action occurs. At this time T1 stops conducting, and
T2 begins to conduct. T~e discharge path of C1 and the equivalent elec-
trical circuit are given in Fig. 18-6. At the instant that discharge of C1

rI
I
ec1 RfI1
I,
LL--_---L_--+-_--'
FIG. 18-6. The discharge path, and the equivalent circuit for the discharge of capaci-
tor C,.

begins, the voltage across C1 is essentially Ebb. C 1 begins to discharge

toward Eb2 - E01 in the circuit shown. But in this circuit RZ2 is shunted
by rb2, the'beam resistance of the tube. Ordinarily rb2 is small compared
with R~-;;~ and the controlling equations for the current and voltage are
approximately
t
e1 == (Eb2 - E 01 ) + (Ebb - Eb2 + E01)e - Ru.C. (18-4)
and
(18-5)

Clearly, the current builds up until it reaches the value +Eor/Ru1, at

which time the potential on the grid of Tl is E01and the cumulative action
once again occurs. The cycle of operation is thus completed, and. the
circuit is ready to repeat its cycle.
The wave forms of the potentials and currents at various points of the
circuit are given in Fig. 18-7. The corresponding results for the second
tube are of the same form, but will be shifted t1 sec along the time axis.
18-3. Frequency of Oscillation. The period of oscillation of the multi-
vibrator is readily found. It follows from Eq. (18-5), since Tl is off for
8Ec.18-3] RELAXATION OSCILLATORS 399
the time required for el to increase from Ebb - Eb2 to EOl along the
!:-(
exponential curve, that ( '''1' '
t
E01 = - (Ebb - E b2 )e - Role,
_ ( :":.0·
{I." .... '

from which ;,

Similarly, for T2, the results are

Ebb - Eb1 Eb;Eol+--+---"'I----+-~t---
t2 = R y2 C 2 loge E (18-7)
- 02

The period of the complete oscil- iCJ

Ebh-EblEoJ
lation, neglecting the switching Rl2

time, is
T = t1 + t2 = RolC l loge
Ebb - Eb2 + R 02 C 2
I
oge
Ebb - Eb1
-E
- E 01 02 eCl

(18-8)
If the two tubes have identical
characteristics, then
Eb1 = Eb2 = E2
E01 = E02 = Eo
and Eq. (18-8) becomes
T = (R g1C1
,
+ R g2C2) log. Ebb_ -Eo Eb (18-9)
If, also,

FIG. 18-7. The potential and current wave

(18-9) reduc~s
Eq.-,----- ,,' to I forms at various points in a symmetrical
'T= 2R C log. Ebb ~ Eb I! (18-10)
\
g
- 0
multivibrator.

The expressions given by Eqs. (18-8) to (18-10) apply to multivibrators

having a low repetition frequency, since they do not take stray capaci-
tances into account. At the higher repetition frequencies, the following
more precise equation should be employed instead of Eq. (18-8):3

T = (C 1 + Ckg1)R g1 loge ( Ebb -E Eb2 C +C Ckg1

- 01 1
1 )

Ebb - Eb1 C2 )
+ (C + C) R 2 kg2 g2 loge ( -
E'02 C 2 Ckg2 + (18-11)
400 ELECTRON-TUBE CIRCUITS [CHAP. 18

This expression is also subject to limitations and should not be used for
extremely high repetition frequencies, except perhaps as an approximation.
An approximate value for the period of oscillation of the multivibrator
is frequently found in the literature. If one assumes that the period is
that determined only by the time constants of the discharge circuits of
C 1 and C 2 without regard for the potential levels between which the ditl-
charges occur, the period of the oscillation is approximately
(18-12)

This result is in error, of course, and frequently an empirical constant is

used to give the frequency of operation. The frequency is given by

(18-13)

where N, the correction factor given in Eq. (18-13), is about 2 for fre-
quency~peration below !iQQ,..£ps and rises to about -i at lo,qoo cps.
18-4. Biased Multivibrators. 3 If the time duration 'otthe output
wave of a multivibrator is important, switching of the nonconducting
tube, which determines the trail-
ing edge, is critical. Switching
occurs when the exponential grid
voltage discharge curve intersects
the tube cutoff curve. If this
intersection is sharp, the time at
, which the trailing edge occurs is
"--_ _ _ _ _ _ _+--_ _----1 --.-1.. correspondingly well defined. If
FIG. 18-8. A biased multivibrator to re- h .. d I h
duce the time jitt~r of the trailing edge.
t e intersectIOn IS gra ua, t e
point of intersection will depend
to a greater or lesser degree on the variations of tube constants and volt-
ages. To ensure a sharp intersection, it is usual to employ a positive
bias on the tubes. Such a circuit is illustrated in Fig. 18-8. The
expression for the period of such a modified circuit becomes
T - C R 1I
- 1 u
Ebb - Eb2
Oge - E 01 + E e
+
Ec +CR 2 u2
I
oge
Ebb - Eb1
E 02 +
+ Ec
L' (18-14)
- Dc

For normal operations, it is found that the frequency of oscillation may

be varied over wide limits in an almost linear manner by controlling Ee.
The linearity between frequency of oscillation and control voltage Ee can
be improved by including resistors Rk in each cathode,4 the value of these
being determined experimentally.
The effect of the application of the positive bias is best illustrated
graphically, and Fig. 18-9 shows the wave form eel on the grid of Tl.
SEC. 18-4] RELAXATION OSCILLATORS 401
The equivalent circuits for the charging of C1 and the discharging of
C2, corresponding to those shown in Figs. 18-4 and 18-6, now become as
shown in Fig. 18-10.
If it is desired to operate the bi-
ased multi vibrator at a fixed fre-
;'
quency, Ec should be made as high /'
/'

as possible, e.g., equal to Ebb. If

the values of Eb and Eo are negli-
gible in comparison with Ebb, the
period then approximates to _ (E6b-~2
T,: (C 1R g1 + C 2R g2 ) loge 2 c. FIG. 18-9. Waveform of ecdor the circuit
(18-15) of Fig. 18-8.

If the positive bias is applied to a single tube only, as in the circuit of

Fig. 18-11, the expression for the period becomes

T = C 1Rg1 loge
Ebb -
-
E
01
Eb2
+C 2
R
g2
1
oge
2Ebb -
E
bb -
E
Eb1
02
(18-16)

This shows that the period has changed because the time required for the

y -:, r,,
,
Cz
! ,,
RgJ i Rll
1(:1 eel E rh1 e
f2
Be
i,
I
..--i L
r, Ee---.l
I
I

0-chtlrqe Cz-clt:scharge
FIG. 18-10. Equivalent circuits for the charge of C t and the discharge of C 2 in Fig.
18-8.

grid recovery of T2 has been decreased, the recovery time for Tl remaining
unchanged. This means that the wave forms for the recovery of the two
tubes are not of equal time duration,
j
l and an unsymmetrical or unbalanced
Rgz condition results. Of course, an un-
symmetrical wave form is easily ob-
tained with the previously discussed
circuits by changing the grid-circuit
time constants of one circuit and not
FIG. 18-11. Multivibrator with one
grid returned to the plate supply of the other. One should not at-
voltage. tempt to achieve a markedly unbal-
anced condition of time in the multi-
vibrator, since if the time constant C2R g2 were made too different from
C 1R g1 , the trailing edge of the long-time-constant tube might become
unstable.
402 ELECTRON-TUBE CIRCUITS [CHAP. 18

The wave form of ec 2 of this circuit is given in Fig. 18-12. The equiva-
lent circuits governing the charge and discharge of the capacitors are
ec2 those of Fig. 18-10, with Ec in the
Ebb discharge circuit replaced by Ebb.
/ 18-5. Cathode-coupled Multi-
/ vibrator. The circuit of a cath-
/
ode-coupled multivibrator is given
in Fig. 18-13. The operation of
this circuit is somewhat different
from the plate-coupled circuit of
FIG. 18-12. Wave form of ee' for the cir- F· 18 2 T d d h
cuit of Fig. 18-11. 19. -. 0 un erst an t e ac-
tion of this circuit, suppose that it
is initially without plate voltage. There will be no charge on C, and the
grids of both Tl and T2 will be at ground potential. When the supply
potential is suddenly applied, both tubes will start to conduct and the plates
of Tl and T2 will begin to fall in potential. But since the voltage across C
cannot change instantaneously, the drop in potential that takes place at

~
:
I

:
~------~ __~__________~-1
FIG. 18-13. A cathode-coupled multi vibrator using direct coupling.

the plate of Tl is coupled to the grid of T2, tending to cause T2 to cut off.
This tendency is also accentuated by the tube currents which flow through
the cathode resistor R k , which raises the cathode potential of both tubes.
But, with the current i b2 tending to decrease, ibl will increase, resulting in
a larger negative potential to the grid
of T2 and also a larger positive po-
c
tential to the cathode of T2, and T2
will rapidly reach cutoff.
T2 is held beyond cutoff during the
time required for C to discharge along
an exponential curve and reach the
FIG. 18-14. The equivalent circuit
cutoff potential 0 f t h e tu b e. The for discharge of C of Fig. 18-13.
equivalent circuit of the discharge is
given in Fig. 18-14. When this cutoff potential is reached, T2 will begin
to conduct. This current through Rk will tend to raise the cathode poten-
tialof Tl, and the current ib1 will begin to decrease. As a result, the plate
SEC. 18-6] 'RELAXATION OSCILLATORS 403

potential of T1 will increase, resulting in a positive signal to the grid of

T2, and the cumulative cycle will continue until Tl is cut off and T2 is
conducting its maximum current.
C
-----r
t
--r
II I

,
I ,
I
I

ez ez
R9.l
I
I
~2 ,
I
Rk I =-Ebb I
I
I

--*-- ---'-
fa) (bJ
FIG. 18-15. The equivalent charging circuits of C: (a) applies during the time that
grid current is drawn; (b) applies for the remainder of the time interval.
The grid of T2 is driven highly positive, resulting in a large plate cur-
rent, which causes the voltage across Rk to rise quickly. However,
because grid current is drawn, the capacitor C charges relatively quickly
for a time, until the potential ec2 and ek are the same, and thereafter
charging continues at a slmver rate.
The equivalent charging circuits ar.e
shown in Fig. 18-15. As C charges,
the bias on T2 decreases, causing ib2
to decrease. This in turn causes
ek to decrease. The grid of Tl is
held constant at ground potential
so that this tube remains cut off as

FIG. 18-16. Wave forms in a cathode- FIG. 18-17. An electron-coupled

coupled multivibrator with direct cou- multivibrator.
pIing.
long as ek is positive relative to ground by more than cutoff voltage.
When ek drops to Eo!, T1 begins to conduct and the cycle reverses.
The wave forms at several points in the circuit (see Fig. 18-16) illus-
trate the operation.
18-6. Pentodes in Multivibrator Circuits. One may use pentodes in a
multivibrator circuit, as illustrated in Fig. 18-17. It will be observed
that the circuit is essentially a conventional plate-coupled multivibrator,
404 ELECTRON-TUBE CIRCUITS' [CHAP. ]8

the cathode, control grid, and screen grid serving as the triode for the
switching action. Since the load resistors are connected to the anodes,
which are shielded from the switching circuit by the suppressor grid,
load changes will not materially affect the oscillatory circuit and the fre··
quency is reasonably independent of the load.
18-7. Design Considerations. In addition to the choice of constants
to yield the desired repetition frequency, a number of other factors must
be considered in designing a multivibrator circuit. If the wave shape of
the output is important, it is necessary to examine the factors which affect
the steepness of the rapid shifts of potential. If a rapid rise of voltage is
desired, one may have to employ design factors that are common to the
design of video amplifiers. Thus it is necessary to employ low values of
plate resistance and to use tubes with low shunt capacitance. With
reasonable care in the choice of these values, it is not difficult to design a
multivibrator circuit which can reach its full rise of potential in 1 f.lsec,
using low-current receiving types of tubes. If
more rapid action is required, heavier current
tubes must be used. For example, the full rise
may be obtained in 0.2 f.lsec, using a 6AG7 tube.
18-8. The Blocking Oscillator. 4 Suppose that
the second stage of a free-running multivibrator
is replaced by a transformer which is so connected
that regenerative feedback results. Such a circuit
FIG. 18-18. A simple has the properties that the tube can be made to
blocking oscillator. conduct hard for a short period of time and then
turned off for a relatively long interval before it
goes through its cycle. The circuit of such an oscillator is given in Fig.
18-18.
The operation of the circuit is substantially the following: Suppose
that the grid is only slightly negative. The tube will conduct, and the
voltage of the anode will begin to fall. This changing potential, which
appears across the transformer in the plate circuit, will be accompanied
by a changing potential in the grid winding of the transformer. The
phase of the transformer connection is such that the potential of the grid
becomes positive, thus increasing the plate current. This is a regenera-
tive action, which continues until the grid draws current, thus charging
the capacitor C to a voltage ~ iT i g dt, where-.2:}~~~e durati()ll oLthe
charging time, the current through Rg being neglected. The charging
ceases when the plate potential falls so low that the plate circuit can no
longer drive the low impedance reflected from the grid circuit.
At this time there is no longer any voltage induced in the grid winding
SEC. 18-8] RELAXATION OSCILLATORS 405

of the transformer, and the voltage on C begins to discharge. The dis-

charge of C causes the grid-cathode voltage to decrease. This results in
a decreasing plate current and hence a rising plate potential. The
resulting voltage in the grid winding causes the grid to go more negative,
thus initiating the cumulative action that causes the tube to cut off.
Now, however, wIth the tube cut
off, the charge on the capacitor can eh
leak off only through the grid re-
- - - -==-"'--
sistor R g, resulting in an exponential
rise toward ground with a time con-
stant approximately equal to RuC,
This rise continues until the grid
potential reaches the cutoff poten-
tial of the tube, when the cycle will ec
repeat itself. -lI4--------/-+-----
The wave forms at various points Eo
in the circuit are shown in Fig. 18-
19.
The total repetition period of
such a blocking oscillator is seen to
be

where T is the duration of the pulse

and E is the potential to which the
capacitor is charged during the
pulse.
The transformer in the circuit f\
may be connected in a number of FIG. 18-19. The wave forms in a single-
ways. Thus, in addition to the swing blocking oscillator.
connection of Fig. 18-18, the transformer windings can be in the grid and
cathode line and also in the plate and cathode lines (see Prob. 18-6).
Precisely the same considerations concerning the possibility of jitter of
the trailing edge of the output pulse applies for the blocking oscillator as
for the multivibrator, and the biased blocking oscillator achieves the
same results in the same way as the biased multivibrator. Consequently,
if the jitter is to be kept to a minimum, a biased blocking oscillator should
be used. The diagram of such a circuit is given in Fig. 18-20. The
recurrence period of this device will be given by the expression

(18-18)
406 ELECTRON-TUBE CIRCUITS [CHAP. III

The duration of the pulse, i.e., the length of the conducting period T,
depends upon the capacitor C and upon the characteristics of the trans-
former. For a given transformer, C is the
most important circuit element, larger values
of C being accompanied by longer pulses.
I For example, a certain transformer, when
I
I
I used in the circuit of Fig. 18-20, yielded
&cI pulses which could be varied from approxi-
I
I mately 0.2 to 20 ,usec by changing the size of
I
I
I the capacitor C.
~L- __________ ~~ __
18-9. Van der Pol Relaxation Oscillator.
FIG. 18-20. A biased blocking One of the earliest forms of single-tube relax-
oscillator.
ation oscillators was described by van der Pol
in 1926. 5 The circuit of such an oscillator is given in Fig. 18-21. It con-
sists of a tetrode in which the control grid and the screen grid are capac-
itively coupled, the control grid being maintained positive by means of a
high resistance coupling to Ebb. However,
since these oscillators ordinarily incorporate a
pent ode rather than a tetrode, the pent ode-
type oscillator will be discussed here.
The circuit of the pent ode-type relaxation
oscillatot is given in Fig. 18-22. This circuit
depends for its operation on the fact that a FIG. 18-21. The van der
change of potential ,on the suppressor grid is Pol oscillator.
accompanied by an amplified change of poten-
tial without phase reversal on the screen grid. To see that this is so,
consider for the moment that capacitor C is removed, and suppose that
the tube is in a quiescent state. The situation is then as illustrated
in Fig. 18-23. Suppose that a negative
signal is applied to the suppressor grid of
R4 R2 such amplitude that the anode current
is interrupted. As a result, the total
space current will be collected by the
screen grid, with a consequent decrease in
screen potential, owing to the screen re-
sistor. In fact, if it is considered that the
control-grid potential establishes the total
FIG. 18-22. A pentode van der
Pol relaxation oscillator. space current, the effect of a signal on th6'
suppressor grid is to control, in a nonlinear
manner, the division of the space current between the anode and the
screen grid, although a more positive suppressor grid results in a
decreased screen current and hence a more positive screen-grid poten-
SEC. 18-9J RELAXATION OSCILLATORS 407
tial. That is, the suppressor-signal and the screen-potential variations
are in the same phase. Thus, if the gain of the circuit is greater
than unity and if feedback is arranged between the output circuit (the
screen) and the input circuit (the suppres-
sor), the device becomes an oscillator and
is almost equivalent to a multivibrator.
Refer now to the circuit of Fig. 18-22,
and suppose that, at the instant that the rr-----.. .
I
suppressor is positive, the anode is taking eC3
current. As a result, the anode potential LL-_---+_~___....L.__
is falling, and the screen potential is ris-
ing, O\ying to the reduced screen current. FIG. 18-23. To illustrate the
The rising screen potential, which is principle of operation of the pen-
tode type of van der Pol re-
coupled to the suppressor through the laxation oscillator.
capacitor C, is accompanied by a rising
suppressor-grid potential, since the capacitor voltage cannot change
instantaneously. This is a cumulative action that continues until the
maximum current is drawn by the anode. When this condition is
reached, a charging current immediately starts flowing into the capacitor
through the combination of R2 in parallel with the screen-cathode path
resistance and the combination of Rs in parallel with the suppressor-
cathode path resistance, assuming that a suppressor-grid current flows.
As the voltage across the capaci-
• tor rises, the suppressor voltage
falls. This results in a falling
screen potential, and at a critical
value the anode current begins to
fall, with a resulting increase in
screen current. This causes a
FIG. 18-24. Typical wave form of the sup- cumulative effect which continues
pressor voltage in a pentode van der Pol rapidly until the anode current is
relaxation oscillator.
cut off and maximum current
flows in the screen circuit. The capacitor charging current now flows
through Ra and the combination of R2 in parallel with the screen-cathode
path resistance, and the potential of the suppressor begins to rise. This
brings the action to the point at which the considerations began, and the
cycle repeats.
A typical wave form of the suppressor voltage is given in Fig. 18-24.
The wave is unsymmetrical because during one portion of the cycle the
r
time constant T is given by C (Ra + RR2 2 ) , where r2 is the average
2 + r2
screen-cathode resistance; and during the second portion of the cycle the
408 ELECTRON-TUBE CIRCUITS [CHAP. 18

time constant T' is given by C (R~~~ r~ + R~~~ r;), where r~ and r~ are
the average screen-cathode and the average suppressor-cathode resist-
ances during the charging period. Evidently, the larger that C and R2
are made, the lower is the frequency of oscillation.
18-10. Synchronized Relaxation Oscillators. Synchronization in mul-
tivibrators may be of two general classes according to the character of
the operation. The lightly synchronized multivibrator is free-running,
and its frequency is "corrected" by the application of a synchronizing

FIG. 18-25. Synchronizing a multivibrator with positive pulses.

voltage. The heavily synchronized multivibrator operates with one or

both tubes so heavily biased that the circuit is inactive until a compara-
tively strong synchronizing or triggering voltage is applied. Further-
more, the presence of the heavy bias causes the circuit to come to rest
after it has passed through one complete cycle. Such circuits are fre-
quently referred to as gate, one-s/tot, or univibrator circuits. Only the
lightly synchronized multivibrators will be discussed here, Chap. 19
being devoted to a study of gate circuits, among others.
In the lightly synchronized multivibrator, the inherent poor frequency
stability of the system is overcome by "driving" the multivibrator with
a synchronizing voltage. This forces the period of the multivibrator to
be exactly the same period as the synchronizing frequency or a multiple
or submultiple of i~. The synchronizing wave form may be of almost
any shape, although a pulse or a sine-wave potential is generally used.
Actually the discussion to follow applies for any of the free-running
relaxation oscillators and is not confined to multivibrator circuits.
18-11. Synchronization by Positive Pulses. Figure 18-25 illustrates
the effect of applying positive pulses to one grid of a free-running multi-
vibrator. This diagram shows the conditions before synchronization
occurs and the transition stages until synchronization becomes complete.
Note that the pulse A has no effect because it does not raise the grid
recovery voltage above the cutoff potential of the tube and that pulses
13 and C have no effect because they are applied to the grid of the conduct-
ing tube and grid clipping occurs. Pulse D is of sufficient amplitude to
SEC. 18-13] RELAXATION OSCILLATORS 409
raise the grid above the cutoff potential, and as a result the tubes are
switched as the current begins to flow in the tube that was formerly cut off.
Suppose that the frequency of the triggering pulses is considerably
higher than the natural frequency of the multivibrator. The situation is
then as illustrated in Fig. 18-26. Thus, if trigger A causes the multi-
vibrator to switch, triggers Band C occur at times when the tube is con-
ducting. Since the grid is already positive, these pulses have no effect

I I I I
-j-+- '-I-
I I I I I
I I I I I
I I I I I
I I I I

I I
I
I
I
I
II I I
I
I
:
I I
I I I I I I I I I I I I I I I I
ABCPEFGHIJKLMNOP
k k k ~ k k L.LU A ~.~ k k l
FIG. 18-26. Synchronization with pulses of higher frequency than the natural period
of the multivibrator.
on the conduction. Triggers D, E, F are applied to the nonconducting
tube, but they are not large enough to cause conduction. Trigger G does
carry the grid voltage above the cutoff point of the tube, and conduction
occurs. In this example, every sixth pulse switches the multivibrator,
and hence the repetition frequency of the multivibrator is one-sixth of
the frequency of the trigger pulses.
In multivibrators, if accurate frequency division is desired, the division
per stage should be relatively low, say 10 or less.. If accurate divisionis
not-requirea~asi:ngle stage may be used -for a division of 100 or more.
Also, for best results, the peak of one pulse should rise above the base of
the next pulse on the grid recovery curve by about 20 per cent of its
height.
18-12. Synchronization by Negative Pulses. Multivibrators may be
synchronized by means of negative pulses. The situation is then some-
what as shown in Fig. 18-27. To understand the operation in this case,
it is necessary to remember only that the application of a negative pulse
to the grid of the nonconducting tube will have no effect, as the grid is
already negative and beyond cutoff. However, a negative trigger applied
to the conducting tube will result in the application of a positive pulse to
the nonconducting tube, owing to the amphfier action of the conducting
tube. If the resulting positive pulse is large enough, this will initiate the
regenerative effect, with the subsequent switching.
18-13. Sine-wave Synchronization. The condition when a sine wave
is used for synchronization is not unlike that for synchronization by
410 ELECTRON-TUBE CIRCUITS [CHAP. 18

pulses, although the addition of a sine-wave synchronizing voltage to the

grid recovery does change the grid-voltage wave form, somewhat as
shown in Fig. 18-28.
The operation of the circuit is substantially the following: When T1 is
conducting during the time from A to B, grid limiting occurs and the
potential eel remains substantially constant. When the synchronizing
voltage falls below zero and starts to decrease the voltage on the grid, the
7nqgerTJ--~~---'r-----~----v-----~--__~___

FIG. 18-27. Synchronization with negative pulses.

FIG. 18-28. Synchronization with sine-wave voltage.

multivibrator regenerative action causes T1 to cut off. As this occurs at
the same point B, C, D, . . . in each cycle of the synchronizing voltage,
the multivibrator is forced to operate at the synchronizing frequency.
Evidently, however, this action exists only when the synchronizing fre-
quency is higher than the natural frequency of the multivibrator.
Synchronization by means of a sine-wave potential injected in the
cathode circuit of one tube is possible. It is essential that the internal
impedance of the synchronizing voltage source must be low in order that
the tube current through this impedance does not seriously alter the
shape of the wave. The operation of such a synchronized circuit is illus-
trated graphically in Fig. 18-29. If one supposes that the grid-ground
potential is not affected by the presence of the sine-wave synchronizing
SEC. I8-H] RELAXATION OSCILLATORS 411
voltage but that the grid-cathode potential contains this sinusoidal com-
ponent of voltage, then one may consider that the effective cutoff poten-
tial of the tube varies sinusoidally about the normal value in phase with
the synchronizing voltage.
At the instant when the synchronizing voltage is suddenly applied at
point A, it is supposed that Tl is conducting. The synchronizing voltage
causes the cathode to rise, so that conduction of Tl is decreased. This

L\ C\ C\

FIG. 18-29. Wave forms of a multivibrator synchronized by a sine wave in the

cathode circuit.

FIG. 18-30. The grid-voltage wave form of a blocking oscillator synchronized with
positive pulses.
leads to the regenerative action, and the tube is quickly cut off. This
switching action will cause the tube to stop conducting at a time earlier
than would be the case for the free-running multivibrator. The voltage
Cel will follow its normal exponential curve until it intersects the cutoff

curve, which has the assumed sinusoidal form, and switching takes place
at the point B. If it is assumed that during conduction the grid-cathode
potential remains constant at some small positive value, which is given,
in fact, by iore, the presence of added negative potential on the cathode
will cause the grid voltage to follow the cathode potential and the poten-
tial follows the curve CD. When the cathode voltage begins to rise,
conduction in Tl is decreased and the regenerative action again takes
place.
18-14. Synchronization of Blocking Oscillators. Blocking oscillators
may be synchronized with either sine waves or pulses applied to the grid,
412 ELECTRON-TUBE CIRCUITS [CHAP. 18

in much the same manner as for synchronizing multivibrators. For

example, Fig. 18-30 illustrates the grid-voltage wave form of a six-count
divider. As for the multivibrator synchronization, the trigger heights
should be such that the peak of one pulse rises above the base of the next
pulse by about 20 per cent of its height. A smaller trigger might allow
the divider to fire un synchronously between triggers. An excessively
large trigger may introduce instability due to variations in trigger heights.
Also, for accurate frequency division with a blocking oscillator, as for the
multivibrator, the division per stage should be relatively low, also 10
or less.
REFERENCES
1. Abraham, H., and E. Bloch, Ministere de la guerre Pub. 27, April, 1915.
Abraham, H., and E. Bloch, Ann. Phys., 12,237 (1919).
2. Vecchiacchi, F., Alta frequenza, 9, 745 (1940).
3. Puckle, O. S., "Time Bases," John Wiley & Sons, Inc., New York, 1943.
4. Benjamin, R., J.IEE, 93, 1159 (1946).
Cruft Laboratory, War Training Staff, "Electron Circuit and Tubes," pp. S04-
S09, McGraw-Hill Book Company, Inc., New York, 1947.
5. van der Pol, B., Phil. Mag., 2, 97S (1926).
As general references, see
Reich, H. J., "Theory and Application of Electron Tubes," New York, McGraw-
Hill Book Company, Inc., 2d ed., 1944.
M.l. T. Radar School Staff, "Principles of Radar," McGraw-Hill Book Company,
Inc., 2d ed., New York, 1947.

PROBLEMS
Note: In the following problems, if necessary, assume that rc for cach tube is
500 ohms. If it is necessary to determine the value of rb when ec = 0, refer to the
plate characteristics of the tube. The cutoff grid voltage of each tube may also
be obtained from the plate characteristics .
./ 18-1. Determine the frequency of oscillation of a plate-coupled multivibrator
using 6J5 tubes for which

Rut = R g2 = 50()k Rn = Rl2 = 3()k Ebb = 300-

18-2. Repeat Prob. 18-1 when the grids are returned to a positive supply.
Plot the frequency as Eoc is varied from 0 to Ebb.
18-3. Given the biased multivibrator shown in the diagram.
 RELAXATION OSCILLATORS 413
a. Calculate the discharge time constants of the capacitors.
b. Calculate the peak value of each grid voltage. Assume the grid voltage of
the conducting tube to be zero in these calculations. Ebb = 300 volts.
c. Calculate the cutoff periods of each section of the tube.
d. Calculate the repetition frequency of the multivibrator.
/18-4. Design a free-running balanced multivibrator to operate at 1,000 cps.
Assume the following: Rz = 10k , Rg = 500 k , 6J5 tubes, Ebb = 200 volts .
./ 18-5. Design a free-running unbalanced multivibrator to operate at 5,000 cps,
with one half cycle three times as long as the other. Use Rg = 50Qk, Rz = 10k ,
6J5 tubes, and Ebb == 250 volts.
18-6. Sketch blocking-oscillator circuits with the transformer windings in
a. Grid and cathode circuits.
b. Plate and cathode circuits.
Be sure to indicate the transformer winding directions with appropriately
placed dots.
18-7. Discuss the operation of the tetrode relaxation oSclllator shown in
the figure. *

18-8. A balanced multivibrator is operated at 1,000 cps. The constants are

those of Prob. 18-1 except for the values of C1 and C2 • It is to be synchronized by
injecting a positive pulse on the grid. What must be the amplitude. of the
synchronizing pulse if the frequency is 1,100 cps?
18-9. Repeat Prob. 18-8 if the pulse-recurrence frequency is 5,600 cps.
18-10. Sketch a curve of the grid potential of one tube of a balanced multi-
vibrator that is being synchronized by a sine-wave voltage injected in the grid
circuit. The synchronizing voltage is in the frequency ratio of 5: 1 with the
multivibrator frequency.
18-11. Discuss the possibility of synchronizing a multivibrator with a positive
pulse, the recurrence frequency of which is a submultiple of the free-running
frequency of the multivibrator.
*Black, D. H., Elec. Commun., 18, 50 (1939).
 CHAPTER 19

HEAVILY BIASED RELAXATION CIRCUITS

T HERE are two important heavily biased relaxation circuits. One type
is known as gate, one-shot, or univibrator circuits, the other general class
being known as trigger or flip-flop circuits. The gate circuit operates
with one or both tubes so heavily biased that the circuit is inactive until
a comparatively strong synchronizing or triggering voltage is applied.
This strong signal causes the circuit to change from its stable limiting
condition to an unstable limiting
condition, and after a definite
time interval that is determined
'l.r by the circuit constants the circuit
returns to the stable state, where
it will remain until it is again
triggered. The trigger, or flip-
FIG. 19-1. A gate circuit with two a-c
flop, circuit is quite like the gate
couplings. circuit except that for fixed values
'of applied voltage there are two
stable conditions of equilibrium. The currents and voltages in such cir-
cuits can be made to change abruptly from one set of stable values to a
second set of stable values, or back again, although independent disturb-
ances are required for each switching action.
19-1. A-C Coupled Gate. Refer to the diagram of Fig. 19-1, which is
essentially a multivibrator of the conventional plate-coupled type, but
with one tube biased to or beyond cutoff by the application of a fixed
positive potential to the cathode of tube T2. In this circuit T1 is nor-
mally conducting, and T2 is normally cut off, thus requiring that the
synchronizing or triggering pulse must be negative-going. Likewise, the
duration of this negative-going pulse should be less than the time of the
unstable portion of the cycle. If it is required to operate the circuit with
a positive pulse, it should be applied to the grid of T2, although in this
case the pulse amplitude must be larger than when it is applied to Tl.
The analysis of such a circuit follows the same general method as that
used in analyzing the free-running multivibrator. The result of such an
analysis leads to the following expression for the time duration of the
pulse:
414
SEC. 19-2] HEAVILY BIASED RELAXATION CIRCUITS 415

An alternative circuit to that of Fig. 19-1, but one which is normally

triggered with a positive pulse, is
given in Fig. 19-2. The time
duration of the output square .fL
wave is found to be L
Ebb - Ebl
t = C 2R g2 loge E (19-2)
02

19-2. D-C Coupled Gate. The

circuit of a d-c coupled gate is FIG. 19-2. A gate circuit that is roughly
comparable with that of Fig. 19-1 except
given in Fig. 19-3. It will be for the polarity of the triggering pulse.
noted that this circuit is only a
slight rearrangement of the circuit of Fig. 19-2. In this circuit tube T2 is
biased at or beyond cutoff by the application of a negative potential to the
grid. This fixed bias is obtained from the voltage divider comprising
R ll , R, and R g2 , which is connected between Ebb and Ecc. Also, in order to
ensure that T1 is normally in the
conducting stage, R g1 , which is
.r-::---t-- Ushown connected to ground, may
be connected to Ebb.
T The capacitor C2, which might
appear to be unnecessary at first
glance, actually serves a twofold
purpose. It serves to decrease
FIG. 19-3.
the switching time from T1 to T2,
and also it serves to increase the reliability of operation of the circuit.
To examine this matter, it is necessary to consider the effects of the
several interelectrode tube capacitances. Thus the tube capacitance
Cgk in T2 tends to prevent the grid-cathode voltage from changing
rapidly. The capacitance Cgp which couples the grid with the plate of
the tube tends to cause the grid voltage to vary in a direction to oppose
the effective switching from T1 to T2. The capacitance Ckp tends to pre-
vent the plate potential from changing.
Another factor to be considered arises from the voltage-divider action
of the coupling and biasing resistors Rand R g2 • Without C 2 in the cir-
cuit, the change in grid voltage is R g 2/ (R + R g2 ) of the change in the
plate potential of T1, and this ratio is normally about Y2. In fact, the
action of the interelectrode capacitances is to reduce the change in grid
voltage considerably below this value of ~f
The presence of C2, which is usually of the order of 50 p.p.f and which is
416 ELECTRON-TUBE CIRCUITS [CRAP. 19

therefore appreciably higher than the tube capacitances, forces the grid of

e<,t r ;,
T2 to respond more quickly to the changes in the Tl plate potential. This

Ie
hecau,", any change ;n poknt;al of the
plate results in a nearly equal instanta-
neous change of the voltage of the grid.

f-
(, Furthermore, with C 2 shunting the cou-

e
c
2
E bb -Eb2 J

-p=====~- --- pIing resistor R, the full value of the volt-

age appears on the grid, the resistor R
being effectively short-circuited for these
short-period changes.
The shapes of the grid potentials at
FIG. 19-4. The grid potentials each of the tubes are illustrated in Fig.
of the d-c coupled gate.
19-4. The time duration of the gate
from the circuit is given by
Ebb - Eb2
t = CIR gl loge E (19-3)
- 01

19-3. The Cathode-coupled Gate. The cathode-coupled gate is essen-

tially the cathode-coupled multivibrator which is so arranged that one
tube is normally biased beyond
cutoff. A comparison of the cir-
cuit of Fig. 19-5 with that of Fig. I 11
18-13 shows that these are almost J - -.--.-+
the same, except that R g2 in the
circuit of Fig. 19-5 is connected
from the grid to the cathode of T2
instead of from grid to ground. FIG. 19-5. A cathode-coupled gate.
In this circuit, Tl will be noncon-
ducting, and T2 will normally be in the conducting state. Consequently
a positive triggering pulse applied to the grid of Tl is required in order
to activate the circuit.
As ordinarily used, the grid of
Tl is returned to a positive po-
tential, rather than to ground.
With this change, it is found that
the pulse width t is directly pro-
portional to the d-c potential Ee
to a high degree of accuracy
FIG. 19-6. A cathode-coupled gate with ·d f. .
the grid of Tl tied to a positive bias E g • over a WI e range 0 VarIatIOn
of this d-c potential. Such a
modified circuit is illustrated in Fig. 19-6. A detailed analysis of the
operation of this circuit follows.
The d-c bias voltage Ee is kept low enough so that tube T2isnominally
SEC. 19-3] HEAVILY BIASED RELAXATION CIRCUITS 417

conducting. The current through this tube i~ given by the relation

Ebb
I b2 = (19-4)
+ rb2 + Rk
=-----:---,:--;0;-
Rl2
and the cathode-ground potential Ek2 is then simply
Ek2 = Ib2Rk (19-5)
Consequently, the voltage across the capacitor, assuming that the grid-
cathode potential of T2 during conduction is zero, is
E2 = Ebb - Ek2 (19-6)
When the circuit is triggered, T1 conducts and T2 is turned off. The
current through T1 is given by the relation
Ebb
Ibl = Rll + rbl + Rk (19-7)

and the cathode-ground potential Ekl is

(19-8)
Clearly, when the circuit is triggered, the plate of tube T1 falls from Ebb to
Ebb - Ib1R n , and the point K changes from Ek2 to E kl . Because of this,
the voltage acrOS8 the capacitor C will tend to change from the value
given in Eq. (19-6) to the value
EI = (Ebb - IblRn) - Ekl = ETl - Ekl = Ebl (19-9)
The capacitor potential changes by an amount
l1E = E2 - EI = Ebb - ETI - Ek2 + Ekl (19-10)
Of this, the fraction ec2 initially appears across the grid of T2, where

ec 2 = (Ebb - ETI - Ek2 + Ekl) R' ~2Rg2 (19-11)

and where R' is the parallel combination of rbl and Rn + R k. The

potential will then approach zero along
an exponential curve, and T2 will again c
begin to conduct when ec2 = E02 in the
I[,l
circuit of Fig. 19-7. This occurs at the
time when
t FIG. 19-7. The discharge circuit
-E02 = e 2e-cm;,
C (19-12) of the capacitor C.
approximately. This expression assumes that R g2 is large compared
with the parallel combination of rbl and Rll + R k. Solving for the time
t yields
418 ELECTRON-TUBE CIRCUITS [CHAP. 19

which is given with good approximation by

t --'- CR g2 I oge Ebb - Eb1 -
E
Ek2
- 02
(19-13 )

To reduce the jitter in the trailing

edge of the pulse, and also to permit
somewhat greater control of pulse
width with d-c bias voltage, the grid
FIG. 19-8. A cathode-coupled delay of T2 is returned to the plate supply
multivibrator. voltage through a large grid resistor,
as illustrated. Because of this change, the analysis given above must be
modified. Such an analysis leads to the following expression for the dis-
charge time of the capacitor and hence e c2
A
for the width of the gate:

t -'-
-
CR I 2E
g2 oge E
bb - ET1 -
E + E
Ek2
bb - k1 02
(19-14)
A sketch of the important wave shapes
of the circuit is given in Fig. 19-9.
This circuit will receiver further con-
sideration in Sec. 19-8.
19-4. Pentode Gate Circuits. The FIG. 19-9. The grid potential of T2
van der Pol relaxation oscillator shown and the plate potential of Tl in the
in Fig. 18-22 may be converted into cathode-coupled delay multivibrator
of Fig. 19-8.
a gate circuit by applying a negative
bias to the suppressor grid or by reducing the value of the resistor
R 2• As already discussed in Sec. 18-9, the principle of the circuit is
essentially the same as that of the two-tube multivibrator, except that
the necessary amplification without phase
reversal is obtained in a single tube. A
.02 circuit of this pentode gate is given in Fig.
19-10. Such a gate circuit may be trig-
gered either by positive or by negative
pulses, the general character of the wave
forms at various points in the circuit being
as illustrated in Fig. 19-11.
FIG. 19-10. A pentode gate cir- A th f f d .,
cuit. no er orm 0 pento e gate CIrCUIt
is illustrated in Fig. 19-12.1 In the
quiescent condition the tube T1 draws grid current, and the plate
potential is at a value E b• The application of a negative triggering pulse
SEC. 19-5] HEAVILY BIASED RELAXATION CIRCUITS 419
to the input will cause the tube to cut off, the time of cutoff being con-
trolled by C 1, R 2 , and Ebb. During the triggering pulse, the capacitor

_Jl..l- - - - - - - G r i d
r
n --~~--~----------Phk
u ~~---,----------,sc~en

~
[17 [17 Suppressor

FIG. 19-11. The wave forms of a penta de gate when triggered with negative and with
positive pulses.

C 1 is discharged through T2, and recharging takes place through R 2,

since at the end of the pulse T2 will be cut off. The grid and plate wave
forms are somewhat as illustrated Trigger 1/
in Fig. 19-13. r
19-5. Trigger or Flip-flop Cir- ,.-
/'
/
cuits. The trigger circuit is not ./
/
unlike the gate circuit except that, eel ""'_9------7',;,,_=_=_--_=_=_=_=_=_9--
for fixed values of applied voltage, I
I;
there are two stable conditions of .R2 0 Charginqloward Ebb
I·

iI
II
I)
Ebb I
II
II
II Eb
II-I---.....:;;..--_.....

FIG. 19-12. A positive-bias penta de FIG. 19-13. The grid and plate wave forms
gate circuit. in the penta de gate circuit of Fig. 19-12.

equilibrium. The currents and voltages in such a circuit can be made to

change abruptly from one set of stable values to a second set of stable
values, or back again, although independent disturbances are required for
each switching action.
Refer to the circuit of Fig. 19-14, which was first discussed by Eccles
and Jordan in 1919. 2 Observe that this circuit is like the multivibrator,
although direct coupling exists between the plate of T1 and the grid of
420 ELECTRON -TUBE CIRCUITS [CHAP. 19

T2 and between the plate of T2 and the grid of Tl. Likewise, the grids
are normally maintained negative by the voltage divider and the negative
C-supply voltage.
Suppose that Tl is conducting and T2 is cut off. The d-c potentials on
the grids of the tubes are then, respectively, approximately zero on Tl,
owing to grid clipping, and - Ecc + R2 !o~g2 (Ecc + E b1 ) on T2. These
must be such that eel is slightly positive and ec 2 is negative. If now a
positive signal is applied to both grids simultaneously, there will be no
appreciable action in Tl but T2 will begin to conduct. This will reduce
the grid potential on Tl, which re-
duces the tube current, setting off a
regenerative action that continues
until Tl ceases to conduct and T2 is
fully conducting. The transfer is
effected in precisely the same manner
as that discussed for other multi-
vibrator circuits. The circuit will
remain in this new stable condition
until another positive pulse is applied.
FIG. 19-14. The Eccles-Jordan trig-
ger circuit. Negative trigger pulses may also be
used to effect the switching. In this
case the negative trigger acts on the conducting tube, causing a sud-
den decrease in the plate current and a corresponding rise in the plate
voltage. This rise is passed to the tube that is cut off, resulting in a flow
of current in its plate circuit, which initiates the switching action.
A practical form of the trigger circuit is illustrated in Fig. 19-15. Note
the use of capacitors C1 and C2 across the coupling resistors Rl and R 2 •
These are used to overcome the ef-
fects of the presence of the tube ca-
pacitances and also to increase the
reliability of operation, precisely as for
the action of the capacitor in the d-c
coupled gate circuit of Fig. 19-3.
Since two pulses are required to cause
the circuit to complete its cycle, viz.,
for each tube to pass from the non- -
conducting state', to the conducting FIG. 19-15. A "scale-of-2" circuit.
state and then back again, the output from a differentiating circuit con-
nected to either plate will consist of a series of positive and negative pulses,
the rate of each being one-half that of the triggering group. Thus, if
either the positive or the negative output pulses are selected, there will be
SEC. 19-5J HEAVILY BIASED RELAXATION CIRCUITS 421

one-half as many as the primary triggering source. This accounts for

the name, scale-of-2 circuit. Circuits of this type are used very exten-
sively in scaling down the number of pulses produced in nuclear-physics
reactions. Ordinarily such scaling circuits are connected in cascade to
produce scale-of-4, scale-of-8,
scale-of-16, etc., outputs. This
permits a high-speed pulse source
to be measured on a low-speed
mechanical recorder.
A circuit of the Eccles-Jordan
type may be used to form gates,
.. 11 db FIG. 19-16. Eccles-Jordan type of trigger
one edge 0 f whlCh IS contro e y circuit using pentodes.
one circuit, the other edge of
which is controlled by a second circuit. In this case, the grids are fed
from separate sources, one source switching the circuit on, the other
switching it off.
A number of variants of this basic circuit have been published. Reich 3
replaced the triodes by pentodes. In this case the grids are free for the
application of the initiating signals, the suppressor grids serving the same
function as the triode grids. The circuit is shown in Fig. 19-16. Here
the application of a positive voltage to the control grid of the nonconduct-
ing tube cannot cause conduction to begin, owing to the high negative
voltage on the suppressor grid, which prevents plate current from flowing.
The application of a negative voltage to the control grid of the conducting
tube reduces its current and triggers the circuit. If a short-duration
negative pulse is applied to the control grids of both tubes simultaneously,
both tubes will be cut off. However, the coupling capacitors cause the
suppressor of the tube which was
conducting to be more negative
than that of the other tube, and
as a result the current transfers to
the other tube at the end of the
triggering pulse. The size of the
FIG. 19-17. An alternative form of trigger capacitors should be such that the
circuit using pentodes.
time taken for them to charge or
to discharge from one equilibrium value of voltage to the other values is
large compared with the duration of the triggering pulse but small com-
pared with the time between successive pulses.
An alternative circuit that operates satisfactorily was suggested by
Regener. 4 The circuit is shown in Fig. 19-17.
A cathode-coupled scale-of-2 circuit was suggested by Seal. 5 The
circuit is shown in Fig. 19-18. It will be observed that this circuit bears
422 ELECTRON-TUBE CIRCUITS [CHAP. 19

the Bame relation to the cathode-coupled multivibrator that the circuit

of Fig. 19-15 bears to the plate-coupled multivibrator.
19-6. Single-tube Pentode Trigger Circuit. It was first pointed out
by Reich 3 that, by replacing with a resistor the coupling capacitor
between the screen and the suppressor grids of the pentode relaxation
oscillator of Fig. 18-22, the result is a trigger circuit. This circuit is
illustrated in Fig. 19-19.

FIG. 19-18. A cathode-coupled scale-of- FIG. 19-19. A singlepentode trigger

2 circuit. circuit.

A physical explanation of the operation of the circuit is possible. Sup-

pose that the tube is initially in such a condition that the current to the
plate is zero when the current to the screen is a maximum. If the resis-
tors Rand R3 and the suppressor bias are properly chosen, the suppressor
potential is sufficiently low to maintain no plate current and the tube
remains in the zero-plate-current state.
Suppose now that a positive triggering pulse is applied to the control
grid which is sufficiently positive to cause plate current to flow. As a
result, the screen current will begin to fall, and the screen and also the
suppressor potential will begin to rise. This causes the plate potential
to tend to increase, resulting in a cumulative action which continues until
the maximum possible plate current flows. At this point, the screen is
almost at E bI" the suppressor potential is slightly positive, owing to the
suppressor-grid current when the potential is positive, and the plate con-
tinues to draw current. The tube thus remains in the conducting state.
To turn the tube off will require the application of a second positive
triggering pulse. With an increase of tube current, the screen will collect
more electrons, lowering the screen potential. This lowers the suppressor
potential, which then begins to control the anode current. With a
decreased anode current, the screen current increases and the cumulative
action begins, which ends when the anode current is zero.
19-7. Scaling Circuits. In addition to the scaling circuits which may
be constructed by forming a cascaded group of scale-of-2 6 circuits, the
result being scale-of-4, -8, -16, -32, -64, etc., circuits are possible to yield
scale-of-5,7 scale-of-6,8 scale of-lO,9 and others. Some of these scalers use
SEc.19-7J HEAVILY BIASED RELAXATION CIRCUITS 423

the basic trigger circuit but provide for forced resetting at the count of,
say, 10. Others make use of what is known as a ring circuit. Since
decade scalers are used extensively, it might be well to consider the opera-
tion of such typical circuits.

FIG. 19-20. A decade scaler consisting of conventional trigger circuits with forced
recycling at the count of 10.
A representative type of decade scaler which consists of four conven-
tional trigger circuits but which provides for forced recycling at the count
of 10 is illustrated in Fig. 19-20. 9 This circuit will operate in a conven-
tional manner for the first nine pulses of a pulse train, assuming that the
initial reset causes all even-numbered tubes, T2, T4, T6, T8, to be conduct-
ing. Table 19-1 shO\vs the sequence of tube operation with successive
pulses. If now provision is made after the ninth pulse to force T4 to

FIG. 19-21. Block diagram of a conventional ring counter.

remain conducting and also to cause T8 to become conducting, then the

scaler will have been recycled, with T2, T4, T6, T8 conducting after the
tenth pulse. It is the function of the feedback from a tap on Rz asso-
ciated ,vith T2 through capacitor C2- 7 to force T7 to go out and T8 to
become conducting. Also, in a similar manner, the feedback obtained
from Rz associated with T7 through capacitor C 7- 4 forces T4 to remain
conducting.
A block diagram showing the general features of a ring counter which is
capable of operating with as many as 50 stages is given in Fig. 19-21.
The pulses are fed continuously into each stage, which is some form of
424 ELECTRON-TUBE CIRCUITS [CHAP. 19

trigger circuit with two input channels. This tends to keep all the stages
in one of their two stable conditions, which might be termed the normal
state. If some pulse causes stage 1 to flip into its other state, the next
pulse will flop stage 1 back to normal. This change of state causes a
pulse which flips stage 2 to its odd state, to be returned to normal by the
next input pulse. Thus the odd state advances systematically down the
string of states, moving one stage with each regular input pulse.
TABLE 19-1
SEQUENCE OF TUBE OPERATIONS IN A DECADE SCALER

Conducting tubes
Pulse
'1'l 12 T3 T4 T5 T6 T7 T8
-~

0 x x x x

1 x x x x
---~
----
2 x x ~ x

3 x x x x

4 x x x x

5 x x x x
- - -- - -
6 x x x x

7 x x x x

8 x x x x

n x x x x

10 x x x x

Clearly, the pulse from each preceding stage, which serves to flip a
given stage over from its normal state, will arrive at the stage almost
simultaneously with the regular input pulse which tends to keep the stage
in its normal state. If the regular input pulses are short enough and if
the time constants of the stages are long enough, then the flipping-action
pulse may be delayed until the regular input pulse has passed and satis-
factory operation is possible. The diagram of a scale-of-5 ring circuit
employing pent odes is given in Fig. 19-22.7
19-8. Linear Delay Circuits. Certain radar applications among others
require circuits to generate pulses, the width of which is directly propor-
SEC. 19-8] REA VILY BIASED RELAXATION CIRCUITS 425

tional to a d-c potential, to a high degree of accuracy, and over a wide

range of variation of this d-c potential. Such a circuit as this will make
possible the generation of a marker or triggering pulse at an accurately
known instant after it was initiated. Although a tremendous effort was
expended in the investigations of circuits to fulfill these requirements, only
a few circuits have been found which are even moderately satisfactory.

FIG. 19-22. The General Electric scale-of-5 ring circuit.

Among the circuits to be studied which do possess the desired character-

istics are: (a) the linear delay multivibrator, (b) the sanatron, (c) the
phantastron, and (d) the linear-sweep delay circuit.
The essential element of the linear delay multivibrator is the cathode-
coupled delay multivibrator of Fig. 19-8. To see that the time duration
of the gate is directly proportional to the bias voltage Eo, Eq. (19-14) is
examined in some detail. This expression is rewritten for convenience.

t ..!.- CR 1 2Ebb - En - Ek2 (19-15)

- 02 oge E bb - E kl - E 02

First it is noted from the tube characteristics of a typical triode, say the
6SN7, that the plate current can be related to the plate and grid potentials
by a relation of the form
(19-16)

where Eb is the plate-cathode potential and Ec is the grid-cathode poten-

tial. That is, except in the region of very small currents, the linear set
of curves is a fair representation of the normal space-charge curves of the
tube. It follows from this expression that
(19-17)
426 ELECTRON-TUBE CIRCUITS [CHAP. 19

Also, it follows from Fig. 19-8 that

Eb + h(Rll + Rk) = Ebb (19-18)
By combining Eqs. (19-17) and (19-18) there results
Ib = Ebb + IJ-Ec (19-19)
Rll + Rk + k
Also from the figure
Eg = Ec + I~k (19-20)
It then follows from Eqs. (19-19) and (19-20) that
Ib = Ebb + IJ-Eg (19-21)
Rll + (IJ- + l)Rk + k
These expressions are combined with Eq. (19-15) to yield

t -- CR g2 I oge Ebb + hRl! - E k2

~-'---.;c~----=~
Ebb - hRk - E02
t - CR I {[Rl! + (IJ- + l)Rk + k](Ebb - E k2 ) + RllEbb + IJ-EgRll}
- g2 oge [Rll + (IJ- + l)Rk + k](Ebb - E 02 ) - RkEbb - IJ-EgRk
(19-22)
This may be written in the form
a + bEg
t = CR g2 loge c _ dEg (19-23)
where
a = [Rll + (IJ- + l)Rk + k](Ebb - E k2 ) + RllEbb ~
b = IJ-Rll
C = [Rll + (IJ- + l)Rk + k](Ebb - E 02) - RkEbb (19-24)
d = IJ-Rk
N ow expand Eq. (19-23) to the form

t = CR g2 [lOge ~ + loge (1 + ~ Eg) - loge (1 - ~ Eg) ] (19-25)

But over the range for which

a~E g < 1 ~E
c g < 1 (19-26)

the logarithm may be expanded and only the first term in the expansion
retained. The expression for t then becomes

t +a
== CRg2 ( loge -ac - Eg e
+ - Eo b d)
or finally

t = CR g2 [lOge~ +(~+~) Eo] (19-27)

SEC. 19-9] HEAVILY BIASED RELAXATION CIRCUITS 427

A circuit that was used extensively at the M.LT. Radiation Laboratory

for producing a marker pulse is illustrated in Fig. 19-23. This circuit
yields an output pulse the delay of which is directly proportional to the
d-c potential applied to the grid of Tl within ±O.25 per cent over the

Inpuf Blocking
diode oscil/cdor t
Sk
SOk

Posift'W'!
+--4""'"-1--_ oll/pllf
pliise

FIG. 19-23. A linear delay multi vibrator which triggers a blocking oscillator to
produce a sharp pulse.

range from about 8 to 150 J,tsec. Other characteristics of this circuit will
be given below.
19-9. Miller Integrating Delay Circuits. 1 The sanatron and phantas-
tron delay circuits are two of a variety of linear delay circuits which have
been designed around the so-called Miller integrating circuit. This cir-

FIG. 19-24. The basic Miller integrating FIG. 19-25. The suppressor and plate
circuit. potential of the Miller integrating
circuit of Fig. 19-24.

cuit, which is illustrated in Fig. 19-24, makes use of the fact that the
potential of the plate eb will fall linearly with time when the plate is made
conducting by applying a positive-going gate to the suppressor. The
general character of the output is illustrated in Fig. 19-25.
To examine the action of the circuit analytically, it is noted that the
428 ELECTRON-TUBE CIRCUITS [CHAP. 19

current ig is given by
(19-28)
and the output voltage is related to the grid potential by the expression,
(19-29)
where K is the gain of the tube. It is assumed that the grid current in the
tube remains substantially constant, and any changes occur across the
capacitor C. It then follows that

(19-30)

By combining Eqs. (19-28), (19-29), and (19-30), there is obtained

eb = eb _ Yg ~ (E _ eb)
, K Cp cc K
which is

or
(19-31)

By writing, for convenience,

b Yg (19-32)
- = C(K - 1)
then
(19-33)

This is the controlling differential equation that relates the plate potential
with the input potential.
The solution of this differential equation, subject to the initial condi-
tions
when t = 0
is the expression
(19-34)
Ordinarily the bias supply potential Ecc is the plate supply potential
Ebb, in which case the expression, Eq. (19-34), assumes the form
(19-35)
But for the conditions of operation this expression may be expanded and
only the lower-order terms retained. This yields
SEC. 19-9] HEAVILY BIASED HELAXATION CIHCUITS 429

eb = KE/;b(l - K)Ebb [ 1 - bt
(bt)2
+ 2! - ... J
or approximately
(19-36)
But
(1 - K) 1
(1 - K)b = - CRy(K - 1) + RyC
and Eq. (19-36) reduces to the form

eb = Ebb (1 - R~C t) (19-37)

Some measure of the linearity is possible by examining the ratio of the

second order to the first order term in the above expansion. The per-
centage deviation from the linear is

Per cent deviation = T(bt)

1L 2
100 = 50bt (19-38)

For a circuit for which the linear time is to extend for 100 J.lsec, the
deviation becomes
100 X 10-6 X 50 .
50bt = 3 X 10 6 X 100 X 10-12 X 200 = 0.1 %

The initial potential drop that appears in the plate potential does not
appear in these equations but represents the contribution to the initial
plate current by the charging current required to charge the tube and
wiring capacitances.
If the circuit is modified as illus-
trated in Fig. 19-26, a control is
possible to the upper and lower po- fi-'-"="-
tentiallimits between which the lin- ;;;.+-+---.1

ear saw tooth traverses. In this

circuit, if the potential of the plate
is higher than E 1, the diode T2 will FIG. 19-26. A Miller integrator with
controlled limits.
conduct. In this way the plate of
T1 is tied to E 1• Likewise, when the plate potential is falling with the
application of the gate to the suppressor grid of T1, diode T3 will be non-
conducting until the plate potential falls to E 2, when the diode T3 will fix
the lower potential of the plate fall to E 2•
The linear saw-tooth wave from the Miller integrating circuit is then
used in a comparator circuit. In this circuit a comparison is effected
between the potential, and hence the corresponding setting of a pot en-
430 ELECTRON-TUBE CIRCUITS [CHAP. 19

tiometer, with that of any desired position on the linear saw tooth. The
output, which appears as the movable edge of the gate, is directly propor-
tional to the potential on the linear potentiometer, and hence to the position
of the potentiometer shaft. Such a voltage comparator circuit is illustrated
in Fig. 19-27. Note that the saw-tooth voltage from the Miller integrator
is fed directly to the grid of one tube, the second tube being connected
to the movable arm of the linear potentiometer.
The circuit resembles the cathode-coupled gate, and its operation
depends on two cumulative actions taking place when certain conditions
are realized. Suppose that the potential at the slider of the ranging

FIG. 19-27. A voltage comparator circuit.

potentiometer is about >2E/ib. At the instant that the Miller integrator

is activated by the application of the positive gate to the suppressor grid,
the potential of the grid of Tl will be Ebb, the potential of the grid of T2
"\yill be Y2E br , and T1 will be conducting. This state of the comparator
circuit continues while the grid of Tl falls with the saw tooth, until it
approaches the potential of the grid of T2. The circuit now becomes
unstable, and a flip action takes place, with Tl becoming nonconducting
and T2 beginning to conduct. Clearly, as the potential of the slider is
reduced, the flip action takes place at an increasing time after the start
of the initial event. The time delay between them will be proportional
to the angular position of the shaft of the ranging potentiometer. The
corresponding action which returns the circuit to its initial state takes
place at the equivalent potential on the exponential rise when the .Miller
circuit capacitor is recharging.
The amplitudes of the voltage steps that appear at the anodes of Tl and
T2 depend on the amount of current that is transferred from Tl to T2 at
the instant of the flip action. This current depends on the setting of the
potentiometer slider, but the variation may be reduced considerably by
returning the cathode resistor to a negative supply. The two adjustable
resistors at each end of the ranging potentiometer are to permit the volt-
age across the ranging potentiometer to be adjusted during calibration.
19-10. The Sanatron. 1 A circuit which incorporates a pentode gate
circuit for producing the positive gate for starting the integrator, and the
SEC. 19-111 HEAVILY BIASED RELAXATION CIRCUITS 431
Miller integrator circuit, the output of which would then be used to feed
a voltage comparator circuit, is called a sanatron. The complete circuit

FIG. 19-28. The sanatron delay circuit.

is given in Fig. 19-28. The significant wave forms at several points in

this circuit are given in Fig. 19-29.
19-11. The Phantastron. The phantastron combines the Miller
r r integrator and the trigger properties
of the sanatron in a single tube. The
schematic diagram is given in Fig. 19-
30, and the significant wave forms at
several points in the circuit are given
in Fig. 19-31. Owing to the mutual
effects that exist during the course of
its operation, the phantastron in the

T2 e~1
1------'c,l-i-------r[::--
C1

II
FIG. 19-29. The potentials at various FIG. 19-30. The phantastron delay cir-
points of the sanatrOll. cuit.

form shmvn is somewhat inferior to the sanatron. It is for this reason

that during the Second W orld War the British, who devised both cir-
cuits, preferred the sanatron.
The American version of the phantastron is designed around the 6SA7
heptode, which, owing to the mutual shielding bet,Yeen significant ele-
ments, proved to be entirely satisfactory. One version of the circuit is
given in Fig. 19-32.
A second variation of the basic circuit is given in Fig. 19-33. This
circuit employs a cathode follower to restore the 6SA7 to the quiescent
432 ELECTRON-TUBE CIRCUITS [CHAP. 19

state quickly by helping to re5tore the plate voltage to the starting condi-
tion at the end of the operating period. In fact, the circuit of Fig. 19-33
yields an output pulse the delay of which is directly proportional to the
d-c potential applied to the control
Inpuf l
friqqer --"-_ _ _ _ _ _ _ _A _
l diode within ±0.1 per cent over
the range from about 8 to 150 f.1sec.
Plafe 19-12. The Linear-sweep De-
lay. The linear-sweep delay cir-
cuit incorporates a feed-back lin-
earizing circuit to achieve a linear

W------I
I saw-tooth wave. This particular
circuit will be discussed in some
emh"'" detail in Chap. 20. A comparison
is then effected between the poten-
I
I tial, and hence the setting of a po-
Screen I~------- tentiometer, with that of any de-
sired position on the linear saw
tooth. Although the voltage com-
parison is accomplished in a dif-
FIG. 19-31. The potentials at various ferent manner from that in the
points of the phantastron. sanatron, the net result is a com-
parable delay circuit.
The essential elements of the linear sweep may be discussed by reference
to the circuit of Fig. 19-34. In this circuit Tl is normally conducting,
and the potential across the capacitor is E bl • With T2 absent and with
the application of a negative gate to the grid of Tl, the capacitor will

Pulse
ou/puf

FIG. 19-32. One version of the phantastron built around the 6SA7.

begin to charge from Ebl toward Ebb along an exponential curve, with a
time constant Rne. With T2 connected as shown and with the applica-
tion of the negative gate to cut Tl off, then as the potential across the
eapacitor increases, the feedback through the cathode follower T2 which
is applied to the circuit through which the capacitor charges ,,"ill cause
SEC. 19-12] HEAVILY BIASED RELAXATION CIRCUITS 433
6SA7 6SN7

Positive
fnqger~~4----+~r
inpuf C1

Oufpuf
FIG. 19-33. A second version of the phantastron delay circuit.

the point A to increase. In effect, therefore, the capacitor, instead of

charging to a constant potential En, charges toward a continually increas-
ing potential. As a result, in-
stead of the charging curve being
exponential, it is very nearly lin- A+----,
ear, the extent of the linearity
being determined by how nearly
unity the gain of the cathode fol- l i
lower is.
The circuit of the linear-sweep
FIG 1934 A linear saw-tooth generator.
delay is shown in Fig. 19-35. In . - .
this circuit, T1 and T2 are connected as a simple cathode-coupled gate, the
negative square wave thus generated being applied to the grid of T3.
The time duration of this square wave is the maximum length of the

Gate Linearsweepgenerafor Amplifier

Resforinq Resforin" Pick-orr 2SK SOK
diode and diode diode 8/oclrin" -ISO
clamper .osci//afor
FIG. 19-35. A 150-llsec linear-sweep delay circuit.
434 ELECTRON-TUBE CIRCUITS [CHAP. 19

delay desired. This negative square wave is applied to the linear saw-
tooth circuit, comprising T3 and T6. T4 helps restore the sweep quickly
to the quiescent condition. The 2k resistor to ground causes a rapid rise
at the beginning of the saw tooth, making it a trapezoidal wave. This is
necessary in order to make the sweep start rapidly enough to overcome
circuit capacitance. Often this resistor may be omitted. The 400 k resistor
and T5 are placed in the circuit as an additional sweep correction to pro-
duce linearity. T7 is called the pick-off diode because it begins to conduct
at a point on the sweep determined by the selection of its cathode voltage
with the adjustment of the 35 k potentiometer. This is essentially the
voltage comparator in the circuit. When T7 begins to conduct, its output
is amplified by T8, whose plate output triggers T9, the blocking oscillator
tube.
Table 19-2, taken from M.I.T. Radiation Laboratory Report T-18, shows
a comparison of the features of the cathode-coupled delay multivibrator
(Fig. 19-23), the phantastron circuit (Fig. 19-33), and the linear-sweep
delay circuit (Fig. 19-35). These data are tentative. The percentage
change in range indicated is the maximum change of the maximum
observed at any point in the cycle.
TABLE 19-2
COMPARISON OF THREE DELAY CIRCUITS FOR 150 ILSEC MAXIMUM.
DURATION

Delay
Subject Phantastron Linear sweep
multivibrator

Duration vs. voltage .. .. 0.25 % from about 0.1 % from 8 to 150 0.1 % from 5 to 150
8 to 150 p'sec p'sec p'sec
10 % change in Ebb about ±0.5 % change in =+= 0.15 % change in ±0.15% change in
250 volts time duration duration duration
Temperature coefficient
(% change in duration
7°C) .... ........... . -0.005%rC -0.002%/oC ±0.003%;oC
Number of tube enve-
lopes ................ 2% 3 5
Max over-all sensitivity
to all tubes .......... ±10% change ±5% ±1%

19-13. Pulse Generators. It is ordinarily not possible to distinguish

between a narrow gate and a pulse, particularly since the two may be
generated in the same manner. A sharp distinction is hardly necessary,
but one might perhaps distinguish between them on the basis of the ulti-
mate purpose of the signal. Another distinguishing feature might be
based on the time duration of the signal. There are, however, certain
SEC. 19-13] HEAVILY BIASED RELAXATION CIRCUITS 435
methods available for generating very narrow pulses with reasonably
sharp rise and fall which are not suitable for generating wider gates. It is
not always necessary that the pulse be rectangular, although certain
applications require such a pulse.
Several methods for generating such
narrow rectangular pulses will be con- t-
sidered. These methods fall into two
general classes. In one, tube circuits 100
are employed, and these may be gas-
filled or vacuum tubes, with provision
for damping to keep the pulse narrow FIG. 19-36. A thyratron pulse gener-
and rectangular. In another type, ator with an oscillatory extinguishing
circuit.
artificial transmission or pulse-form-
ing lines are used to control the duration of the pulse and also to effect
a rectangular shape.
A simple circuit for generating pulses incorporates an oscillatory circuit
between the plate and cathode of a thyratron. Such a circuit is given in
Fig. 19-36. It is the function of the oscillatory circuit when shock-excited
to cause the plate potential to fall below that required to maintain the
discharge of the thyratron, once it has been fired. The pulses so gen-

FIG. 19-37. A thyratron pulse generator with thyratron clipper.

era ted are not rectangular in shape but are acceptable for some types of
serVlCe.
The circuit of Fig. 19-37 shows a circuit for generating narrow rectangu-
lar pulses ranging from less than 1 p,sec duration to perhaps 10 or 20 p,sec
duration. In this circuit an 884 thyratron with its ionization time of
approximately 10-8 sec is used to generate the pulse, the trailing edge of
which is controlled by a second thyratron. The output pulse from such
a circuit has relatively sharp sides.
Suppose that the 884 has been triggered by the application of a positive
triggering pulse to the grid. This causes the 884 to conduct, and the
potential that appears across the O.OOl-p,f capacitor from plate to ground,
less the drop in the 884 appears across the cathode resistor, capacitor,
436 ELECTRON-TUBE CIRCUITS [CHAP. 19

and the 2050 clipper thyratron. Note that the 200k plate resistor and
the 1M cathode resistor are so large that a continuous discharge through
the 884 cannot be sustained, and that the energy of the pulse must be
supplied from the O.OOl-}1f capacitor.
Clearly, for different pulse lengths,
different sizes of plate-ground capac-
itors will be required.

,-1 Once the 884 has been fired, the po-

tential on the plate of the 2050 in-
creases to some positive value, and if
the energy in the pulse is small com-
pared with that on the plate capacitor,
FIG. 19-38. A triggered blocking 08- the 2050 plate potential remains rel-
cillator for pulse generation.
atively constant. Owing to the RC
circuits between the plate and the control grid and the grid and the cath-
ode of the 2050, there will be a time delay before the grid-cathode potential
increases from the normally nonconducting value of -15 volts, determined
by the voltage divider between the -150-volt source and ground, to the
approximately zero potential required for the tube to fire. The time dura-
tion is controlled by the adjustable 1M resistor in the circuit. Once the
2050 fires, the cathode of the 884 is effectively short-circuited to ground,
thus terminating the output pulse. Of course, the 2050 continues to con-
duct until enough of the charge has leaked off the capacitor in the 884
circuit for the potential across the 2050 to fall below that required for
maintaining the discharge.
L L L L L

~c
i 1Frrr'J_T2
1 I
FIG. 19-39.
JR O

The basic circuit incorporating a pulse-forming line.

The use of a blocking oscillator which is biased to be off normally and

which is triggered as desired will yield pulses which are sometimes suitable
for some services. Such a circuit which provides for an adjustable pulse
width is given in Fig. 19-38.
19-14. Line-controlled Pulse Generators. The use of artificial trans-
mission lines in the formation of short rectangular pulses has become
widespread in recent years. The basic circuit is shown in Fig. 19-39.
The line is considered to consist of a sufficient number of sections so that
it closely approximates a continuous transmission line. One end of the
SEC. 19-14] HEAVILY BIASED RELAXATION CIRCUITS 437

line is open-circuited; a resistance Ro equal to the characteristic imped-

ance of the line Zo = VL1C is connected across the other end. The line
is initially charged by means of the battery, and hence all capacitors are
charged to full battery potential. When the switch is closed, one half of
the battery voltage appears across the
terminating resistance Ro and the remain-
ing half of the voltage traverses the line
as a voltage wave. The wave front is
delayed t = viLC sec for each section
/ ~ /

through which it passes. As it pro-

m"77"7"7C/""""/""~"77"""""'?"r"""'77""""'"
gresses down the line, the voltage wave E/zO~~~~~~~~~~
removes one-half of the voltage present
across the line. When it reaches the end
of the line, the wave encounters the open
circuit and is reflected without change of
polarity. The remaining voltage on the
line is then progressively removed as the FIG. 19-40. The traveling voltage
wave returns toward the starting point. wave on an open-circuited trans-
When the wave front reaches the starting mission line tll.at is originally
charged and is being dis~harged.
point, it cancels the voltage across the
resistance Ro and the entire system comes to rest. In traversing the line
forward and backward, the occupied time is
t = 2N VLC sec (19-39)

where N is the number of sections.

The sequence of events is illustrated in the block diagram of Fig. 19-40,
which shows the voltage distribution on the line at various times during
the existence of the running wave. As a result of the sequence of actions,
a voltage pulse appears across the resistor Ro. When the switch is
closed, the voltage instantaneously assumes one-half of the applied volt-
age; this voltage persists until the voltage wave has traversed the line
and returned to R o, and it thereupon drops to zero suddenly.
The circuit provides a very simple method of forming a rectangular
pulse of controllable length, provided that the pulse length is short
enough to be obtained in an artificial line of practical dimensions. The
practical difficulty that might be encountered in such a pulse-forming cir-
cuit is the number of sections required to approximate a continuous trans-
mission line. Each section introduces a small deviation in the flat top
of the pulse. If the top is to be essentially flat, a large number of sec-
tions should be used.
The Guillemin line is a network which simulates the section of a trans-
mission line which is open-circuited at the far end. A circuit of this line
438 ELECTRON -TUBE CIRCUITS [CHAP. 19

is given in Fig. 19-41. This network is a closer approximation to the

continuous transmission line than a simple line of an equal number of
sections. Several typical line-controlled pulse circuits are given below.

FIG. 19-41. The general characteristics of a pulse-forming circuit employing a

Guillemin line.

-! ___ OJ\I\ft..+.

FIG. 19-42. A line-controlled thyratron pulse generator.

Figure 19-42 is essentially the circuit of Fig. 19-36, except that the
pulse duration is controlled by the "length" of the line in the plate cir-
cuit, the thyratron being extinguished when the plate potential falls to
zero.
In the circuit of Fig. 19-43 a line is added to the grid circuit of a block-
ing oscillator to control the duration of the pulse. In this circuit the
triggering pulse is applied to the art i-
ficialline and is followed immediately
thereafter by the beginning of the
rectangular wave. The wave of
voltage traverses to the open end of
the line and then is reflected back
without change of sign. When the
wave front reaches the input end, the
FIG. 19-43. A line-controlled blocking positive grid voltage is removed from
oscillator.
the line, reducing the input potential
to zero. This causes the grid-cathode potential to be driven below cut-
off, thus yielding a sharp pulse in the output of the circuit.
A delay line may also be used as the cathode impedance in an amplifier
tube, as shown in Fig. 19-44. The operation of this circuit is essentially
the following: The sharp rise at the beginning of the applied gate to the
grid is accompanied by a corresponding sharp increase in the plate cur-
rent. This causes a cathode-ground voltage wave, which then proceeds
SEC. 19-14] HEAVILY BIASED RELAXATION CIRCUITS 439
down the transmission line, which is essentially open-circuited, as the
terminating resistance Rk is made much higher than the characteristic
impedance of the line. Rk serves
only to complete the d-c path from
cathode to ground. The voltage
wave is reflected from the open JL --ft---.--+
end and retraces its path back
along the line. This reflected
voltage at the input end adds to
the applied voltage to cause the
cathode-ground potential to reach FIG. 19-44. A pulse generator with the
a value of nearly twice the initial line in the cathode circuit.
voltage. This increase in the cathode-ground voltage is enough to cause
the tube to cut off, thus terminating the pulse.
REFERENCES
1. Williams, F. C., and N. F. Moody, J. lEE, 93, 1188 (1946).
2. Eccles, W. H., and F. W. Jordan, Radio Rev., 1, 143 (1919).
3. Reich, H. J., Rev. Sci. Instruments, 9, 222 (1938).
4. Regener, V. H., Rev. Sci. Instruments, 17, 180 (1946).
5. Seal, R. K. F., Electronics, 20, 150 (September, 1947).
6. Sharpless, T. K., Electronics, 21, 122 (March, 1948).
7. General Electric Co., Decade Scaling Unit, Type YYZ-l.
8. Langberg, E. L., Rev. Sci. Instruments, 18, 796 (1947).
9. Potter, J. T., Electronics, 17, 110 (June, 1944).
10. Close, R. N., and M. T. Lebenbaum, Electronics, 21, 100 (April, 1948).
As a general reference, see Chance, B., et al., Waveforms, Vol. 21, M.LT. Radia-
tion Laboratory Series, McGraw-Hill Book Company, Inc., New York, 1949.
PROBLEMS
19-1. A 6SN7 tube is to be used to produce a positive rectangular gate of
108 p,sec duration. The pulse repetition frequency is 400 per second. The gate
is driven by a negative trigger pulse, and the positive gate is to begin at the time
that the trigger is applied.
a. Draw the circuit for such a gate. Indicate on the circuit the terminals to
which the negative trigger is applied, the terminals from which the gate is
taken, and the circuit elements which determine the length of the gate.
b. One grid should be returned to B+, and the other to ground. Give reasons.
c. Calculate the coupling capacitor connected to the grid of tube Tl. Use
the following data:
Section 1 Section 2
Grid to B+ Grid to ground
Ebb .............................. . 300 300
R •............................. 0.5 M 1M
Rz .... · ................ . 20k 30k
It· . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 20
Static Te • • • • • • . • • • • . • • • . • • . . . . . . . . • 500 ohms 500 ohms
Static rb when eo = O.............. . 10k 10k
440 ELECTRON-TUBE CIRCUITS [CHAP. 19

19-2. a. Calculate the recharging time constant of the capacitor C in the

cathode-coupled delay gate.
b. To what values does e,2 jump when the circuit recovers, i.e., what is the
value of e,2 at point A of Fig. 19-9?
c. What is the value of ebi at point B?
19-3. Carry out the analysis that leads to Eq. (19-14) for the duration of the
gate of Fig. 19-8.
19-4. The circuit shown in the diagram is to be designed to give a pulse output

PoSti'/~\--;.~-.--+
trigger',

of 100 J.Lsec width. Assume the following:

6SN7 tube; Ebb = 250 volts

Tc = 1k when e, = 0 or positive
rb = 10k when e, = 0 or positive
Eo = -10 volts
Trigger prf = 400 cps

a. For operation under these conditions, what limits are imposed on the values
that Eo might assume?
b. Calculate the value of the coupling capacitor C.

19-5. The circuit of Fig. 19-23 uses a 6SN7 tube and is supplied from a 300-volt
regulated supply. Suppose that the bias voltage Eo can be varied over the
limits from 20 to 65 volts.

a. Are the approximations in Eqs. (19-26) satisfied over this voltage range?
b. Calculate the duration of the gate at each limiting value. The plate current
is related to the voltages by the approximate expression

amp ±1O%
19-6. The variable resistor of the pulse circuit of Fig. 19-37 is set at 0.5 M •
Estimate the duration of the pulse generated by the circuit. Ebb = 300 volts.
19-7. The fundamental problem in the design of a tuned pulse-forming network
is to determine a network composed of but few circuit elements which will respond
with a closely rectangular wave of current when excited with a unit voltage.
Show that the rectangular wave may be repres~nted by a Fourier series, the first
five terms of which have the form:

J(t) = al sin wt + a3 sin 3wt + as sin 5wt + a7 sin twt + ag sin 9wt
 HEAVIL Y BIASED RELAXATION CIRCUITS 441

Calculate the values of the coefficients an. Plot the resulting curve repre8tmted
by these results, and compare with the results for which
al = 1.2575 a3 = 0.3725 a5 = 0.1735
a7 = 0.08315 a9 = 0.02777

19-8. The diagram for this problem shows a circuit of the pulse-forming stag('
of a pulse modulator. Calculate the width and amplitude of the generated pulse.
+400

19-9. It is required to design a circuit which will provide the several outputs
as shown on the accompanying time diagram.
Base
"igger ~---- ______________A-____

Trigger
A
601 1110
I

Trigger t----H-Il-...,,...--------------
B

A B

a. Draw a block diagram of a circuit that will satisfy the requirements.

b. Show the complete circuit diagram. Do not evaluate the circuit elements
to be used.
All times are given in microseconds.
 CHAPTER 20

SWEEP GENERATORS

T HE function of the sweep circuit as applied to a cathode-ray tube is to

cause the luminous spot that results from the impact of the electron beam
on the fluorescent screen phosphor to move across the screen with a known
velocity, whether it be constant or whether it be varying in some definite
way with respect to time. The proper selection of the form of the sweep
is most important, since otherwise a relatively simple investigation may
prove to be almost impossible. If, for example, it is desired to examine
the shape of a periodically recurring wave, a sweep Circuit that is propor-
tional to the time would normally be desired. Such a linear sweep might
not be appropriate in some application where a sweep that is proportional
to some other quantity might be indicated.
The problems that arise in producing sweep circuits for electrostatic
cathode-ray tubes are quite different from those in producing the sweep
circuits for electromagnetic-type tubes. This chapter will consider the
problems that arise in the generation of sweeps for electrostatic-type
tubes; the next chapter will consider the problems arising in the genera-
tion of magnetic and other sweeps.
20-1. Simple Saw-tooth Sweep. One of the most common sweep
requirements is for a repetitive trace along a straight line. Owing to the
dimensional limitations of the cathode-ray tube screen, the wave form
required to produce it is of the
..
~"VV\
., saw-tooth form, as shown in Fig .
t: ~
"u
>V)
20-1. The wave form illustrated
~"~
~5 I is the ideal, since it varies linearly
~ ~ I with time, but following standard
5~ I practice, no attempt is made to
I-Time oftrace+r~ linearize the fly-back portion of
FlybOlck time
the wave. With the wave form
FIG. 20-1. A simple saw-tooth sweep.
shown, the electron spot moves
across the tube face with a uniform velocity and with a speed depend-
ing upon the slope of the curve. The spot is caused to return to the
starting point very rapidly, and ordinarily it is not seen or , at worst , its
brilliancy is much reduced over that of the writing portion of the cycle.
The ideal saw tooth illustrated is not easily obtained by direct means.
442
SEC. 20-2] SWEEP GENERATORS 443

and various methods have been employed to linearize or to compensate

nonlinear waves. The details of such circuits will be examined below.
20-2. Capacitive Sweep Circuits. Many of the more common circuits
employ the changing potential across a
capacitor upon the application of a
constant source of potential through a
resistor. These circuits usually employ
separate charging and discharging cir-
cuits in order that the retrace, or fiy- FIG. 20-2. The elements of a
switching circuit for producing a
back, time may be made much shorter saw-tooth wave.
than the writing time. Ordinarily the
charging circuit is permitted to function continuously, the rapid dis-
charge being accomplished by a rapidly acting switching circuit. The
equivalent form of such a circuit is shown in Fig. 20-2.
The potential across the capacitor in the circuit shown follows the form
t
Cc = Ebb(l - c -RC) (20-1)
where Ebb is the voltage applied to the RC combination, t is the time in
seconds, C is the capacitance of the capacitor in farads, and R is the
resistance in ohms. For a short time after the application of the voltage

FIG. 20-3. Achieving a linear saw tooth

E cf p S
FIG. 20-4. A saw-tooth generator using
from an RC circuit. a glow discharge tube.

Ebb, the voltage increases across the capacitor will be reasonably linear
with time. For example, for a time equal to one-fifth the time constant
of the circuit, the voltage increase will be linear' within about 5 per cent.
If the switching is properly timed, a reasonably satisfactory wave, of the
form illustrated in Fig. 20-3, is possible.
One of the earliest and probably one of the simplest capacitive saw-
tooth generators utilizes a glow discharge tube as the switch in the circuit
d Fig. 20-2. The circuit of this saw-tooth generator is given in Fig. 20-4.
Suppose that the capacitor is initially uncharged when the switch S is
closed. The voltage across the capacitor will increase according to
Eq. (20-1). When Cc equals the breakdown potential of the glow tube
Ed, charge will fiow through the tube. This discharge is accompanied
by a rapidly falling capacitor potential, because charge is being drained
444 ELECTRON -TUBE CIRC UITS [CHAP. 20

from the capacitor. When the potential across the capacitor falls below
the extinction potential Ee of the glow tube, the discharge ceases.
The discharge current is high since it is limited only by the resistance and
inductance of the leads that connect the capacitor to the glow tube, and
the discharge time is very small.
Once the tube becomes extinguished, the capacitor immediately begins
to recharge. When the potential across the capacitor again reaches the
discharge Ed, the tube will again break down and conduction will continue
until the capacitor voltage again falls to E e , when the process repeats
itself. This process is periodic and results in oscillations that have a
period T equal to the time required for the capacitor to charge from the
potential Ee to Ed and then discharge to Ee. The expression governing
the process is found from Eq. (20-1) to be
T = RC loge Ebb - Ee (20-2)
Ebb - Ed
The amplitude of the oscillation equals the voltage difference Ed - Ee.
Such simple glow-tube circuits
may be used for frequencies of 1
cycle every few minutes to fre-
quencies well up in the a-f range.
The wave shape of the resulting
saw tooth is dependent upon the
(a) (6) magnitude of the supply source
FIG. 20-5. The wave shape of the output Ebb. The use of a VR-90, for
in a simple glow-tube sweep generator. which Ed equals approximately
Curve (a) is for a supply voltage of 600
120 volts and for which Ee equals
volts; curve (b) is for a supply voltage of
135 volts. The amplitude of oscillation is 85 volts approximately, and a 600-
the same in both cases. volt supply yields a wave that
is approximately triangular, as
shown in Fig. 20-5a. The same circuit when used on a 135-volt supply
yields the curved wave form of Fig. 20-5b.
20-3. Thyratron Sweep Generators. Two obvious shortcomings of
the simple glow-tube sweep gen-
erator are that high voltages must R'
be used in order to obtain rela-
tively linear saw-tooth waves and c To CRT
I
R
---1
that the amplitude of the output
wave is relatively small. The use '-----,H----4------4'-fII~
of a thyratron as a switch al- J!:.cc Ebb
FIG 20-6 A saw-tooth generator em
lows reasonable flexibility in con- . . -
ploying a thyratron.
trol of the period of oscillation
and also in the amplitude of the saw-tooth wave so generated. The cir-
SEC. 20-3] SWEEP GENERATORS 445
cuit of a saw-tooth voltage generator employing an 884 thyratron in such
a relaxation circuit is given in Fig. 20-6. In this circuit Ru is a protective
resistor in the grid circuit, and R' is a protective resistor in the plate cir-
cuit. il' is a current-limiting resistor that is made as small as possible
consistent with the tube-current rating in 240
order that the capacitor may discharge very a
quickly through the tube. "- r-.,.., 160~

To understand the operation of the circuit, ~

o
reference must be made to the critical grid
I'.
80
~
" (L

curve of the thyratron, which is given in Fig. o

..Q
(L

20-7. Suppose that the battery potentials -32 -24 -16 -8 o

are set at Ecc = -20 and Etb = 250 volts. Gdd Potential
Also suppose that the capacitor is initially un- FIG. 20-7. Typical critical
grid characteristic of an 884
charged. After the switch S is closed, the argon-filled thyratron.
voltage across the capacitor will increase
exponentially according to Eq. (20-1). This charging process will con-
tinue until the voltage across the capacitor reaches approximately 160
volts. At this time the tube will break down. Charge will leak off the
capacitor very rapidly, and the tube will stop conducting when the
cathode-anode potential falls below about 16 volts, the arc-maintaining
voltage of the tube. The capacitor will again begin to charge through

-- --- -:,--;;:;:::=---- -E6b-Sl.lPP/Y vO/fage

./
/'
/
/
/

--fi"rinq pofent;a/(smtlf/l bias)

If--"---"--''--'LL.JL....JL....JL-_--'-_ _ Dei(m,zing pofent,;;,,/

Time
FIG. 20-8. The change of amplitude and frequency of a thyratron saw-tooth generator
by change of grid bias.

the local RC circuit, and the entire process will repeat itself. The period
of oscillation can be adjusted by varying R, C, or the voltages Ecc or
Ebb. Certain of these results are made evident from an inspection of
Fig. 20-8.
The oscillations of a gas-tube generator are not very stable in fre-
quency. The thyratron oscillator can be synchronized with a given fre-
quency by injecting a small voltage of the desired frequency on the grid,
in precisely the same way as synchronization is achieved· for a multi-
vibrator or bloc!dng oscillator (see Chap. 18). The entire synchroni-
446 ELECTRON-TUBE CIRCUITS [CHAP. 20

zation process is illustrated in Fig. 20-9. The operation is essentially the

following: The natural period of the oscillator is adjusted to be slightly
less than that of the synchronizing voltage. The"natural action is shown
dotted in the figure. Without the synchronizing voltage, the tube would
fire at point A. The presence of the synchronizing voltage on the grid
ec
. - Firing pofenfia/ wlded
by synchronizloq potenf/~
I
1/--· ec wifh synch.
/ ! .ec wifhoufsynch
I/'K
1/
L - - Oeli:m;zlr,q polentia/

FIG. 20-9. Synchronizing a thyratron saw-tooth generator with a sine wave.

causes the firing point to vary in accordance with the grid signal. At
some time during the synchronizing cycle the firing potential will be low,
so that the tube fires at point B. On the next cycle the voltage across
the capacitor reaches the firing point at D. The period, or time per cycle,
is thus reduced from AC to BD, and the oscillator is now synchronized,
or locked, to the frequency of the injected voltage. In a similar manner
it may be synchronized to a submultiple or multiple of the synchronizing
voltage.

c c

FIG. 20-10. Free-running vacuum-tube sweep generators.

20-4. Free-running Vacuum-tube Circuits. The thyratrons of the
previous circuits may be replaced by vacuum tubes with substantially
the same operation. However, since a vacuum tube as such is neither
self-triggering nor self-extinguishing, circuits must be .employed which
SEC. 20-5] SWEEP GENERATORS 447
accomplish this action. The use of a hard-tube relaxation-oscillator cir-
cuit serves the purpose very well, and such circuits are used extensively.
Four such circuits are illustrated in
Fig. 20-10. One circuit employs a
conventional plate-coupled multivi-
brator; one employs a cathode-coupled
multivibrator; the third employs a U c
blocking-oscillator circuit; and the
fourth employs a pentode relaxation
FIG. 20-11. A hard-tube sweep cir-
oscillator. These circuits may be cuit.
synchronized by injecting a synchro-
nizing voltage into the grid or into the cathode circuits, as already dis-
cussed in connection with the several oscillator circuits. Hence all the
properties of the gas-tube relaxation circuits are also possessed by these
relaxation circuits.
20-5. Triggered Sweep Circuits. The use of a vacuum tube as a
switch allows a very accurate timing of the start of the sweep voltage,
since there is substantially no delay between the application of a signal
I on the grid and the consequent

"'10 ------j--
D effect in the plate circuit. In par-
ticular, refer to the circuit of Fig.
20-11. In this circuit, the grid is
so biased that the tube is conduct-
I
I ing when no square wave is applied
I
to the grid. The potential across
ell
the capacitor C is then Eb, the drop
o~~--~--------~--~---
across the tube. With the appli-
FIG. 20-12. The voltage across the capac- .
itor C in the circuit of Fig. 20-11. catIOn of the negative gate, the
tube ceases to conduct, and the
voltage across C begins to increase toward Ebb through the plate-load
resistor R l • Then, depending on the width of the gate and the constants
of the circuit, the saw-tooth will follow one of the paths shown in Fig.
20-12.

lL_--+

FIG. 20-13. A negatively FIG. 20-14. The output-voltage wave form from
biaBed Baw-tooth generator. the &weep generator of Fig. 20-13.
448 ELECTRON -TUBE CIRCUITS [CHAP. 20

If the tube is normally biased negatively so that no current flows

through it, then upon the application of a positive gate the potential
across the capacitor will fall from the value Ebb to the value determined
by the drop across the tube Eb and the drop across the resistor R 12 , if one
is used to limit the current through the tube. The generator circuit is
given in Fig. 20-13, and the general character of the discharge curves is
illustrated in Fig. 20-14.
20-6. Linearizing Saw-tooth Waves. When a linear-s,,,,eep voltage is
necessary, the most direct recourse is to employ a high potential toward
which the capacitor is charging and then
to use only the lower portion of the charg-
ing curve. An alternative and somewhat
more satisfactory method is to replace the
charging resistor by a constant-current de-
vice, such as a pent ode or other constant-
1J current generator, and charge the capacitor
through this. For the majority of cases,
the degree of linearity obtainable with such
FIG. 20-15. Linearization of a a constant-current generator is usually suf-
saw-tooth wave by using a high ficient. Other methods of linearizing saw-
charging voltage, obtained by
interrupting the current in a tooth generators do exist, and these gen-
large inductor. erally employ one or another of the
following methods of obtaining the desired
compensation: (a) the inverse curvature of a vacuum-tube character-
istic; (b) an auxiliary time-constant circuit; (c) feed-back methods. Each
of these methods wiII be considered below.
Charging to a High Voltage. As already pointed out, the use of a frac-
tion of the charging curve will yield a sub-
stantially linear voltage increase, particu-
larly if the capacitor is charging to a high
potential. It is not necessary that a high
d-c potential source be available, and the
circuit of Fig. 20-15 illustrates the method of
generating a high voltage by interrupting
the current in a large choke (of the order
of 500 henrys). Such circuits have long
flyback times and are limited to circuits FIG. 20-16. A thyratron saw-
tooth generator with a pentode
having a small duty cycle (the ratio of the
charging circuit.
duration of the sweep to the sweep recur-
rence period).
Use of Constant-current Generators. The use of a constant-current
generator in place of the ordinary charging resistor allows a very satis-
SEC. 20-6] S WEEP GENERATORS 449
factory linear saw-tooth voltage to be generated. With such a device,
the capacitor charging current passes through the constant-current
generator, and the capacitor potential increases linearly with time.
Figure 20-16 illustrates the circuit of a thyratron generator with a pentode

1S

FIG. 20-17. Positive- and negative-going linear saw-tooth generators employing

pentode constant-current generators for linearization.

as the constant-current generator. With such a circuit, the frequency

may· be varied by changing the bias of the pentode, since this controls
the tube current and hence the
charging rate of the capacitor.
The circuits of Fig. 20-17 show
the connections and the wave
1J
forms expected from a vacuum-
tube circuit employing pentode
charging. For a circuit that in-
corporates a triode as a constant-
current generator see the figure of FIG. 20-18. A saw-tooth generator incor-
porating pentode charging with adjustable
Prob. 20-7. feedback.
A slightly modified circuit pro-
vides pentode charging with adjustable feedback. In this way, any
curvature that might still remain in the saw tooth can be compensated by
controlling the feedback. This circuit is illustrated in Fig. 20-18. A
variation of this circuit was suggested by Bedford. His arrangement is
given in Fig. 20-19.
450 ELECTRON-TUBE CIRCUITS [CHAP. 20

Linearization by Means of Inverse Curvature. These methods compen-

sate to a certain degree for the curvature of the charging characteristic by
inserting a device having a similar but inverse characteristic. An
arrangement which is often adopted is to amplify the potential across the
capacitor by means of a tube in such a way that the curvature of the
tube characteristic is employed as a linearizing means. It is fortunate
in this respect that the curvature of a part of the transfer curve of the
tube approximates the inverse of an
exponential characteristic. Such a
A circuit is illustrated in Fig. 20-20.
In this circuit the output potential
1S is amplified by a second tube which
is operated with such a bias that the
nonlinear operation compensates
FIG. 20-19. An alternative circuit that the nonlinear input. The added
provides a combination of pentode capacitor C 2 is to maintain the an-
charging and feedback to linearize the ode potential reasonably constant
saw tooth.
during the discharge period of C 1 •
Linearization by Means of an Auxiliary Time-constant Circuit. A
simple method that yields fairly satisfactory results modifies the exponen-
tial wave form across the capacitor by the addition of an integrating
circuit. 1 The essentials of the circuit are illustrated in Fig. 20-21. In
this circuit, the charging capacitor consists of two parts C1 and C2 in
series. These are charged together, through the resistance R 1, and dis-
charge through the tube.

n
1J

FIG. 20-20. A circuit for linearizing a saw FIG. 20-21. A circuit that employs an
tooth by employing the curved character- integrating network in the output to
istic of a vacuum-tube amplifier. linearize the saw-tooth wave.

Assume that the capacitors C 1 and C2 have just been charged. The
capacitor Cs will be at a potential less than that across C1, as C s charges
through R 2. The potential of C s depends on the time constant C SR2
and the length of time that Cs is charging through R 2 • When the gate
is applied to the grid, C1 and C2 begin to discharge, while the potential
across Cs falls more slowly owing to the large time constant CSR 2 • When
C1 and C2 are discharged and the tube is again nonconducting, C s will
SEC. 20-6] SWEEP GENERATORS 451
retain a considerable charge. C 1 and C2 now start to charge once more,
but C1 receives charge from two sources, the B+ supply through R1 and
from C3 through R 2 • C3 supplies charge to C1 until the potentials are
equalized, and C3 charges thereafter from C 1• As a result, the potential
across C1 and C2 is exponential, while that across C3 is approximately
parabolic. If C2 and C3 are properly chosen, the output is approximately
linear.
Linearization by Means of Feed-back Arrangements. Feedback may be
applied at two points in a saw-tooth generator for improving the linearity
of the output. The effect that results from the stray capacitances of the
leads and the plates of the cath-
ode-ray tube can be minimized by
feeding the output from the cath-
ode of a cathode-follower stage.
The other effect arises because the
potential across the charging ca- 1.J - __-+
pacitor is exponential as it ap-
proaches the fixed charging poten-
tial. This can be compensated by
. d· t· FIG. 20-22. A linear saw-tooth generator
mtro ucmg a compensa mg po- employing a cathode follower for feedback.
tential in the charging circuit to
counterbalance the potential of the capacitor. A circuit that accom-
plishes this result is illustrated in Fig. 20-22.
In this circuit, TI is normally conducting, and the potential across the
capacitor is E b1 • With T2 absent and with the application of a negative
gate to the grid of TI, the capacitor will begin to charge from Eb1 toward
Ebb along an exponential curve, with a time constant CR1 + R 2 )C 1.
With T2 connected as shown and with the application of the negative gate
to cut TI off, as the potential of C 1 increases the potential at Rk increases,
and if the gain of the cathode follower is unity, the two potentials are
exactly equal. Consequently, by the application of the output of the
cathode follower into the circuit in such a way that the voltage across the
capacitor is just balanced by the cathode-follower output, the net effect
is a linear output. The capacitance of C2 is made sufficiently large so
that the potential across it remains steady, and it is kept charged from the
B+ supply through R 2 •
It should be noted that the charge leaking off C2 through R1 and R2
must bo replaced during the recycling time, and since the time constant
of the circuit must be large compared with T, this requires an appreciable
time. Unless the recycling time is long compared with the duty time, a
d-c shift will result. Moreover, the presence of R2 reduces the feed-back
gain. The use of a diode in place of R2 avoids these difficulties, as it
452 ELECTRON-TUBE CIRCUITS [CHAP. 20

cuts off during the duty interval but closes to give a low impedance during
recycling. Such a circuit is illustrated in Fig. 20-23.
The extent to which the resulting saw tooth is linear depends upon how
closely the gain of the cathode fol-
lower approaches unity. In effect,
therefore, the circuit is such that the
capacitor, instead of charging to a
constant potential Ebb, charges to-
ward a continually increasing poten-
1I tial, the extent of the increase being
such as to compensate very nearly
for the potential across the capacitor.
FIG. 20-23. The use of a diode in the
plll,te circuit to reduce any d-c shift and
T h us t h e current t h roug h t h e circuit
to maintain the feed-back gain. remains constant, with a consequent
linear potential across CI .
Of course, if amplification is provided before applying the output across
the capacitor C\ to the grid of the cathode follower, it is then possible to
get almost any degree of compensation. Such a circuit would have the
form illustrated in Fig. 20-24.

lJ

FIG. 20-24. A linear saw-tooth generator with amplification before the application of
feedback through a cathode follower.

It is a disadvantage of the above circuits that the grid of the cathode

follower is driven negatively during the flyback period. Because of the
various stray capacitances across the output circuit, these will tend to
delay the fall of the cathode potential, and the cathode follower may
momentarily cut off, with a consequent poor flyback. For those applica-
tions in which the duty cycle is small, the results with the circuit of Fig.
20-23 are highly satisfactory. This circuit is used in the linear delay
gate and was discussed in Sec. 19-14.
20-7. Push-pull Deflection. By tying one of each pair of deflecting
plates in a cathode-ray tube to the second anode, the total number of
external leads is reduced by two. On the other hand, when this is done,
both defocusing and trapezium distortion will result. 2 Consequently, in
SEC. 20-7] SWEEP GENERATORS 453

those cases where these effects cannot be tolerated, it is necessary to con-

nect the anode potential to the mid-point of a high resistor across each
pair of plates and apply the deflecting potential across the plates. Using
the connection illustrated in Fig. 20-25 provides the opportunity for using
push-pull deflection. This is particularly desirable in that a number of
advantages are gained. These are improved linearity of the trace,
improved deflection sensitivity, and avoid-
ing the need for clamping circuits. Ap;//ed
A defleC!infj
Actually, the use of push-pull deflection potential
I
with one of the pair of deflecting plates
being connected directly to the saw-tooth FIG. 20-25. The connection to
generator, the other being fed from an in- the deflecting plates in a cathode-
verting amplifier of unity gain, results in ray tube to reduce defocusing'and
trapezium distortion.
a sacrifice in linearity and in flyback speed.
However, it is possible to arrange the circuit so that the distortion intro-
duced by the tube approximately cancels out the effect of an exponential
curvature in the wave form of the saw tooth. It is essential that the
gain of the amplifier remain unity-otherwise some distortion will result
but this is not difficult to achieve if the amplifier is provided with nega-
tive feedback.
The improved deflection sensitivity results because the instantaneous
deflecting potential difference with push-pull deflection is twice what it is
with a single-sided, or unbalanced, deflection.
Clamping circuits are unnecessary because the average ordinate of one
wave is equal and opposite to that of the other wave, and the effects are
in such a direction as to cancel each other.
Five important methods of obtaining push-pull deflection potentials
from a single input wave are available. These are
1. A single-tube phase-reversing stage which mayor may not provide
more than unity gain. This may be subdivided into

a. Those circuits in which the input saw-tooth wave form is sensibly

linear and the tube distortion is arranged to be a minimum.
b. Those circuits in which the input wave form is exponential and the
amplifier characteristic is eIllployed also for compensation of the
input curvature.

2. A single-tube push-pull, or paraphase, amplifier.

3. A two-tube push-pull amplifier which has been modified to handle
the peculiarities of a saw-tooth wave form. This may be subdivided into

a. A standard form of push-pull amplifier.

b. A cathode-coupled push-pull, or paraphase, amplifier.
454 ELECTRON-TUBE CIRCUITS [CHAP. ::0

4. Split saw tooths.

5. The use of two equivalent saw tooths operating in phase opposition.
It is not possible to state which of these methods is the most satis-
factory, since each possesses certain merits. For most purposes, type
2, the single-tube paraphase circuit, where no amplification is required,
and type 3b, the cathode-coupled paraphase amplifier, where amplifica-
tion is required in addition to the provision of a push-pull output, will
ordinarily provide satisfactory results. The use of 4 and 5 usually permit
deflection voltages with a minimum of distortion.
Single-tube Phase-reversing Amplifier. Any vacuum tube used as a
conventional plate-coupled amplifier provides an output voltage that is
opposite in polarity to the input voltage.
However, since the gain of such a stage is
ordinarily greater than unity, some means
must be found to reduce the amplifica-
tion. One common method of reducing
the gain is to introduce sufficient negative
feedback; the anode follower, discussed in
Sec. 6-7, would be suitable for this service.
FIG. 20-26. A single-tube para- H owever, I't IS. qUI'te l'k 1 th a t th e p h ase
I e y
phase amplifier.
difference between input and output po-
tentials might differ slightly, particularly at the higher frequencies, where
stray capacitances become important.
A second way to reduce the gain of the simple amplifier circuit to unity
is to use a voltage divider in the input circuit to reduce the amplitude of
the grid signal by an amount equal to the gain of the amplifier. If the
wave form which is to be inverted contains many harmonics, special care
must be taken to compensate the voltage divider for the shunting effect
of the stray capacitances associated with it.
Single-tube Paraphase Amplifier. A combination of amplifier and phase
inverter to provide a push-pull output from a single input wave is known
as a phase splitter or paraphase amplifier. A single-tube paraphase
amplifier in which the plate resistor is divided equally between the plate
and cathode circuits is the simplest form of paraphase amplifier and was
examined in Sec. 9-11 (page 187). The circuit of this amplifier is redrawn
for convenience in Fig. 20-26.
An important feature of this circuit is to be noted. This arises from
the presence of stray capacitances between the cathode and ground. If
a positive-going trace is applied to the grid, this capacitance will charge
slowly during the trace, and when the grid is brought down sharply at
the end of the trace, the tube current decreases and hence during the
flyback time the capacitance must discharge through the cathode resistor,
SEC. 20-7] SWEEP GENERATORS 455
a relatively long time-constant circuit. If, on the other hand, a negative-
going saw tooth is applied to the grid, the cathode-ground capacitance
will charge slowly and during the retrace time the tube current will be
large; as a result the cathode potential is enabled to change rapidly, and
the flyback is not impaired.
Another point to be observed is that the tube behaves as a cathode
follower sufficiently to prevent the output at the cathode from being
appreciably influenced by the value of the cathode load impedance.
The presence of Rk increases the effective impedance of the tube as seen
from the anode, so that the anode current is almost independent of the
anode circuit. The output potential at the anode is directly dependent
upon the value of Rz in parallel with the deflecting-plate leak resistor Rl
and the total stray capacitance associated with the anode and the deflect-
ing plates. It is for this reason that the stray capacitances across Rl
and R2 must be equalized. This balancing arrangement prevents the
flyback from overshooting the start of the
trace.
Two-tube Paraphase Amplifiers. These par-
aphase amplifier circuits are discussed in some
detail in Sec. 9-11 (page 187), and reference
should be made to this section for details.
Cathode-coupled Push-pull Amplifier. The
cathode-coupled paraphase amplifier is dis-
cussed in some detail in Sec. 6-9 (page 117) as
the difference amplifier. This paraphase am-
FIG. 20-27. A cathode-
plifier has many desirable features which rec- coupled para phase ampli-
ommend it as a satisfactory solution of the fier.
paraphase problem. The circuit is redrawn in
convenient form in Fig. 20-27.
The attractive features of this circuit are the following:
1. Low distortion and relatively small value of grid current when
overloaded.
2. Freedom from any tendency to self-oscillation.
3. Permits d-c connections at the input and output.
4. Permits providing shift, astigmatism correction, and balance
controls.
5. Permits sweep expansion.
Figure 20-28 shows a circuit in which a cathode-coupled paraphase
amplifier is arranged to provide sweep expansion and other controls.
Split Saw Tooths. It is sometimes desirable to avoid the necessity for
amplification and phase inversion by generating a saw-tooth wave form
having sufficient amplitude to allow half the output voltage to be applied
456 ELECTRON-TUBE CIRCUITS [CHAP. 20

to each deflecting plate. The principle involved is indicated in Fig.

20-29a. Here the two charging resistors have a capacitor between them,
although one might equally well use two equal charging capacitors with
the charging resistor between them. The output provides equal poten-
-/50 +/60 +/60

-rI
I
: Clam
I
I
Inpuf
I
I
I
X-sN!'f
I
I
confrol
~
-.72S
X-asfiqmafism
confrol -440 -J40 1440
FIG. 20-28. A cathode-coupled paraphase amplifier with sweep expansion.

FIG. 20-29. A split saw-tooth generator. (a) employs a single capacitor and two
resistors. (b) employs two capacitors and one charging device.

tials of opposite phase. The discharging device is indicated by the

switch S. Figure 20-29b shows a practical circuit for obtaining a split
saw tooth from two capacitors and one charging device.
Saw Tooths in Phase Opposition. In Fig. 20-30 is shown a circuit which
comprises two saw-tooth generators, one of which is connected in the
reverse sense to the other. In this
r------It---- circuit a single discharging device may
be employed, but the plate supply
would be larger than would be nec-
essary if separate discharging devices
L----...----lt---- are employed. The resistors Rl and
R2 may be replaced by constant-
current devices. The switch S which
FIG. 20-30.
p\l~h-pull.
Saw tooths operating in
represents t h e d'ISC h argmg
. d' .
eVlCe IS
used to bring the capacitors to the
same potential. Since Rl and R2 are equal, this potential is half that
SEC. 20-7] SWEEP GENERATORS 457
of the plate source. When the switch S is opened, one plate is driven
in the positive direction and the other plate is driven in the negative
direction.
The circuit of Fig. 20·31 is one in which the elements are so arranged
as to provide a push-pull output in which the whole of the available supply
potential may be made to appear across each of the capacitors C l and C2 •

L1 Synch.

Top/aks
FIG. 20-31. Two thyratron saw-tooth generators operating in push-pull.

In this circuit the two thyratrons T1 and T2 are connected across the
capacitors C l and C2 , respectively, and the bias potentials applied to
their grids are determined by the adjustment of the resistors Rkl and R k2 .
The potentials across these resistors remain constant.
When the capacitors Cl and C2 are being charged, the potential of the
anode of T2 remains fixed, while those of the cathode and grid rapidly
become more negative with respect to the anode. Since the grids of the
two thyratrons are joined together through C3 and the discharge current
of C3 passes through R gl , the grid-cathode bias of T1 is greater than that
due to the potential drop across R l . For this reason Rl must be adjusted
to provide a smaller bias potential than is the case with R 2 , since other-
wise there will be a time lag between striking of T1 and T2. The object
is to obtain simultaneous striking, and this is best achieved by making
the potentials at the two grids equal at the instant immediately prior to
striking by accurate adjustment of R l .
REFERENCES
As a general reference, see
Puckle, O. S., "Time Bases," John Wiley & Sons, Inc.., New York, 1943.
1. Hawkins, G. F., Wireless World, 5,425 (1939).
2. Millman, J., and S. Seely, "Electronics," Sec. 3-10, McGraw-Hill Book Com-
pany Inc., New York, 1941.
PROBLEMS
20-1. An 884 thyratron is used in the saw-tooth generator of Fig. 20-6. The
parameters are adjusted to be
458 ELECTRON -TUBE CIRCUITS [CHAP. 20

E~~ = 300 volts Bee = -20 volts C = 0.003 pJ R = 10 6 ohms

a. Calculate the frequency of oscillation.
b. Calculate the amplitude of the generated waves.
c. Plot the wave shape of the generated waves. Assume that the flyback
time is zero.
20-2. Design a thyratron relaxation saw-tooth generator to yield a substan-
tially linear saw tooth having a peak amplitude of 25 volts, with a recurrence
frequency of 1,000 cps. Specify reasonable values of Ebb, E", Rand C, and give
reasons for your choice.
20-3. Derive an equation for the potential across capacitor C in the circuit of
Fig. 20-11:
a. During the time that the gate is applied.
b. After the gating period.
c. Sketch these results.

20-4. The parameters in the simple sweep circuit of Fig. 20-11 are the following,
for tube 6SN7:
C = 0.005 p.f Ebb = 300 volts prf = 1,000 cps
The input negative gate has an amplitude of 50 volts, and a duration of 150 p,sec.
a. Calculate and plot curves of the form illustrated in Fig. 20-12.
b. Repeat a for Rl = 50 k •
20-5. Set up and solve the differential equation that controls
a. The charge to capacitor C of Fig. 20-17a when the negative cutoff gate is
applied.
b. The discharge of capacitor C at the end of the gate period.
20-6. The circuit of Fig. 20-17b is to be used in a saw-tooth generator circuit.
The triode is a 6J5, and the pentode is a 6SJ7. A linear saw tooth with a 100-volt
excursion is required, the duration of which is 250 p,sec. The recurrence fre-
quency is 1,000 cps. The available plate-supply source Ebb = 300 volts. Specify
all elements of the circuit.
20-7. a. Find expressions for the potential across C during the charge and
discharge portions of the cycle of the circuit in the figure for this problem.

c
 SWEEP GENERATORS 459
Assume that the beam resistance of T2 is rb when the tube conducts and is infi-
nite when the gate is applied.
b. Plot the results of part a.
c. On this same curve sheet, plot the results of Prob. 20-3, assuming the
same value of C and that R/ = R 1•
20-8. Derive an expression for the output potential from the circuit of Fig.
20-21 when a positive pulse is applied to the grid. The tube is normally biased
beyond cutoff. What should be the relation among the circuit elements for an
approximately linear output?
20-9. The circuit shown in the diagram is essentially the feed-back circuit of

1J
Fig. 20-22. Discuss the advantages of this arrangement over that of Fig. 20-22.
20-10. Derive an expression for the voltage across capacitor C1 in Fig. 20-22.
20-11. If the circuit of Fig. 20-22 is modified as shown in the accompanying
diagram, and assuming that the cathode of T2 rises linearly at a constant rate K \,

show that the voltage that appears across C3 during the saw tooth is given by
__t__ 1
e3 = Kit + K1RCa(e RCa )

Neglect the effect of charging C2 from the main source.

20-12. Discuss the operation of the following circuit, suggested by Puckle.

,
e
 CHAPTER 21

SPECIAL SWEEP GENERATORS

IT MAY be shown that the deflection of the electron ~eam in an electro-

magnetically deflected cathode-ray tube is proportional to the field
strength of the deflecting field. The field strength is proportional to the
current passing through the deflecting coil, if saturation is avoided. Con-
sequently, if it is desired to deflect the electron beam linearly with time,
the current through the coil must be increased linearly with time. When
the end of the sweep is reached, the electron beam must be returned to its
starting point quickly. It will be seen below that achieving a linear
current through a deflecting coil (a series RL circuit) is a difficult task; in
general, one seeks to achieve satisfactory results, even though these may
leave something to be desired.
21-1. Sweep Generators for Electromagnetic Deflection. The're are
four general methods in use for obtaining a linear current through a
deflecting yoke. They make use of (a) the initial portion of the exponen-
tial charge of current through the yoke when a step voltage is impressed
on the series RL circuit, (b) the initial change of current in an inductance
during an oscillation when the current in a parallel RLC circuit is varied
suddenly, (c) the increase of current in an
inductance due to the application of a
trapezoidal voltage of properly chosen
dimensions, and (d) the use of feedback in
Oeflechng
a way to provide for linearity. Each of
yoke these methods will be examined below.
Exponential Rise of Current. The basic
FIG. 21-1. A simple circuit for circuit of the sweep generator for produc-
producing a saw-tooth current. ing an exponential current through an
inductance is given in Fig. 21-1. In this circuit it is supposed that no cur-
rent is initially flowing in the deflecting-yoke circuit. When the switch S
is closed, the current will begin to rise exponentially according to the
equation
E R,t
i = ~ (1 - e -L+L,) (21-1)
Rc
The rate of rise of current is given by
460
SEC. 21-1] SPECIAL SWEEP GENERATORS 461
di Ebb
(21-2)
dt = L + Lc
and hence the sweep speed may be varied by adjusting the magnitude of
the inductance in the circuit.
To return the sweep to the initial condition, the switch S is opened.
When this is done, however, the current tends to oscillate owing to the
oscillatory circuit consisting of the yoke inductance and the distributed.
capacitance of the yoke. In order to damp out these oscillations, it is
sometimes necessary to connect a resistor or a damping diode across the
deflecting yoke.
An electronic circuit that is based
on these principles is illustrated in
Fig. 21-2. In this circuit, the tube ' l - - - - * " - - ../\....
is normally biased beyond cutoff, It _--.--+
and hence no current is flowing in
the deflecting-coil circuit. At the
time that the sweep is to be pro-
FIG. 21-2. A circuit for generating a
duced, a positive pulse is applied to sweep current by impressing a rectangu-
the grid of the tube. The purpose lar pulse on an RL circuit.
of the resistor and diode that are
connected across the deflection coil is to damp out the oscillations that
are excited at the end of the sweep. The diode serves to disconnect the
damping resistor during the rise of current. When the driver tube is cut
off at the end of the positive pulse, a reversal of potential across the
inductance results, which will make the diode conduct and permit cur-
rent to flow through the damping resistor.
Initial Part of Oscillation. When an attempt is made to interrupt the
current in a circuit consisting of a coil shunted by a capacitor, an oscilla-
tory current is produced in the tuned circuit. The first part of the first
i

FIG. 21-3. A circuit for ob- FIG. 21-4. The current in

taining an oscillating cur- the deflecting coil in the
rent in a deflecting coil. circuit of Fig. 21-3.
cycle of the oscillation of the current is reasonably linear and may be
used as a sweep current. The elements of the circuit are illustrated in
Fig. 21-3.
462 ELECTRON-TUBE CIRCUITS [CHAP. 21

The shape of the current in the coil is somewhat as illustrated in Fig.

21-4. The rate of change of current in the interval from a to b depends
upon the resonant frequency of
the deflecting-coil inductance, and
the shunt capacitance, and the
amplitude of the initial current.
Of course, if different values of
lS shunt capacitance are used, dif-
ferent resonant frequencies result,
whence the sweep speeds (con-
sidering the region from a to b to
FIG. 21-5. An oscillatory sweep-current be the effective portion of the
circuit with arrangements for adjusting the
sweep) may be varied. Likewise,
zero position of the sweep.
the amplitude of the oscillation
will depend upon the magnitude of the d-c current in the coil before it is
interrupted.
A sweep circuit that employs such an oscillatory circuit is given in
Fig. 21-5. In this circuit the deflecting coil is coupled to the plate circuit
of a current-amplifier stage by means of a large capacitor C. The induc-
tors L1 and L2 are very large and hence serve to isolate the a-c oscillatory
current from the zero-position circuit and the plate supply Ebb and thus
confine the current to the deflecting-yoke circuit. However, the bias
battery Ed sends a steady current through resistor R, inductor L 2 , and
the deflecting coil for the purpose of centering the spot.
In this circuit the application of the pulse to the amplifier tube causes
the current in the deflecting-coil
circuit to decrease and oscillate ld
about one level, and the removal
of the pulse causes the current to
oscillate about the d-c value de-
termined by the local circuit con-
taining Ed. The shape of the
curve is somewhat as illustrated FIG. 21-6. The current in the defiecting-
coil circuit when a pulse is applied to the
in Fig. 21-6. The second oscilla- grid of the current tube in Fig. 21-5.
tion is damped more rapidly than
the first because the beam resistance of the tube is in parallel with the cir-
cuit during this portion of the cycle.
Linear Current by Trapezoidal Voltage. Both the foregoing methods
for obtaining a linear-sweep current utilize a small essentially linear por-
tion of a nonlinear function. It is possible to find a voltage wave form
which, when applied to the inductive circuit of the deflecting coil, will
yield a current that is linear with time. To examine the required voltage
wave shape, refer to Fig. 21-7.
SEC. 21-1] SPECIAL SWEEP GENERATORS 463

For a linear current having the form

i = kt (21-3)
the voltage across the deflecting coil must be
di
e = L dt + R·'/, (21-4)

which requires, for the linear current, that the voltage have the form
e = Lk Rkt + (21-5)
This has the form illustrated in Fig. 21-8.
e

D
:~ R ,
~-
:I L ~ __________-+__ ~t

I
-------T ------...\
FIG. 21-7. A linear deflecting current. FIG. 21-8. The voltage wave to produce
a saw-tooth current in an RL circuit.
A circuit which is capable of developing a trapezoidal voltage wave form
is shown in Fig. 21-9. The switch S that is connected across Rc and C
permits the capacitor to be charged and
discharged. When the switch is opened,
the current in this circuit increases accord-
ing to the relation
. Ebb (RL~R')C
'/, = RL + Rc e (21-6)
FIG. 21-9. A circuit for gen-
The output voltage of the circuit is erating a trapezoidal wave.
e = Ebb - iRL
or
EbbRL
e = Ebb - R L + Rc e (21-7)
The slope of the voltage is
de
(21-8)

l
dt =
At the initial time, t = 0,

( at +
de) 0 = Ebb (R L RLRc)2C
and (21-9)

(e)o = Ebb RL ~ Rc
464 ELECTRON -TUBE CIRCUITS [CHAP. 21

If the time constant (RL +

Rc)C is large compared with the sweep time
T, so that only a reasonably linear portion of the exponential curve is
used, the output voltage will be a sufficiently good trapezoidal voltage
wave having the properties shown in Fig. 21-10.
e
1 EhbRZ,
(RL ;Rc)2C
T

Eb,/lc+ ------ --.i

BL+Hc~~__________~~____~
h-- ----:r ------~
FIG. 21-10. Trapezoidal voltage and the relation to the circuit constants.

An electronic circuit that incorporates the foregoing features for pro-

ducing a linear saw-tooth current through the deflecting coil is given in
Fig. 21-11. It will be observed that tube T1 serves as the switch of

lS

FIG. 21-11. A circuit for generating a saw-tooth current wave for electromagnetic
deflection.

Fig. 21-9, the output trapezoidal wave being applied to the grid of a
current amplifier tube T2. A biased clamp T3 is used to set the reference
level.
It would be possible to deduce the features of the voltage that must be
apphed to the grid of the driver or current amplifier tube in Fig. 21-11 in
order to give the desired saw-tooth current in the deflecting yoke. How-
ever, this will require a knowledge of the tube parameters. Since these
will not be constant, owing to the fact that the tube is nominally biased
beyond cutoff, and will conduct only on the application of the grid driving
potential, such an expression is only approximate, at best. More
accurate results are possible by direct recourse to the plate character-
istics of the tube, and deducing therefrom the requisite grid signal.
This will require a knowledge of the yoke current and the features of the
voltage across the yoke, as well as the grid clamping potential. This grid
signal may, in fact, be nontrapezoidal.
SEC. 21-1] SPECIAL SWEEP GENERATORS 465

A variation of the circuit of Fig. 21-11 that does not require a clamp is
shown in Fig. 21-12.

FIG. 21-12. A saw-tooth current generator that provides for d-c restoration by means
of a separate circuit.

There are two factors which tend to influence adversely the considera-
tions given above, viz., the distributed capacitance of the coil, and the
capacitance of the circuit wiring. Since the voltage across a capacitor in
a series RC circuit cannot change suddenly, the distributed capacitance
at the input to the current or driver stage will reduce the steepness of the

~-------------+t
FIG. 21-13. The effect of distributed FIG. 21-14. The use of a narrow spike to
capacitance in retarding the start of a aid in starting a linear-sweep current.
sweep current.
initial jump of the trapezoidal voltage, making the trapezoid tend to
approach a saw tooth. The application of such a voltage to the RL cir-
cuit through the current tube is a current which increases slowly at first,
as shown in Fig. 21-13, and gradually becomes linear with time. Fre-
quently one applies a sharp spike at the beginning of the sweep to help
overcome this effect. Such a
wave has the form shown in Fig.
21-14.
The effect of the distributed ca-
pacitance of the deflecting-coil
circuit, which may be considered
FIG. 21-15. The wave form of the current
to appear as a capacitance across in the deflecting coil.
the coil, is to produce an oscil-
latory circuit. The rapid change of current passing through the coil at
the time of the return trace will ordinarily shock-excite the coil into
oscillation. But since it is necessary to provide some means of dissi-
466 ELECTRON-TUBE CIRCUITS [CHAP. 21

pating the energy in the electromagnetic field of the coil so that the cur-
rent can fall to zero in a short time, a resistor is often connected in
parallel with the deflecting coil. This resistor serves to damp the oscil-
lations produced by the shock excitation. The complete curve is gen-
erally of the form illustrated in Fig. 21-15.
Feed-back Circuits. Although the circuit methods described under
Exponential R.ise of Current have been used extensively for magnetic
sweep-current generation, the present-day methods generally incorporate
feedback in the circuit as a means for linearizing the current wave. That
is, the circuit is so arranged that the application of a linear voltage saw
tooth to the grid of the input amplifier tube will result in a linear saw-
tooth current through the deflecting yoke.
Consider the circuit of such a feed-back sweep amplifier shown in Fig.
21-16. The input saw tooth to this amplifier may be generated by a
6SL7

807

FIG. 21-16. A feed-back sweep-current amplifier.

simple RC sweep generator, since only a small input signal is required,
and for most purposes the resulting linearity in using a small portion of
the wave is adequate. This input is passed through two voltage-ampli-
fying stages and is then applied to the grid of the current-amplifier tube.
If the total current through the current tube differs from a linear varia-
tion, there will be a difference in potential between the cathode of the
input stage and an appropriately chosen sample of the output current
stage. Owing to the feedback between the output and the input stages,
the effective signal will vary in a manner to yield a linear current in the
output. Note also in the figure that the grid of the output stage is d-c
restored to a value beyond cutoff.
An improved circuit is shown in Fig. 21-17. In this improved circuit,
the sampled output current is amplified before being fed back to the inp.ut.
There are two features of the above circuits that should be discussed.
In the first place, the sampling of the output current is done by means of
the potential drop across the cathode resistor. Evidently, therefore, it
is the total current that is being sampled, and not the current through
SEC. 21-1] SPECIAL SWEEP GENERATORS 467
the deflecting yoke. Of course, if the fraction,of the total current that
flows through the deflecting yoke remains constant as the current
Increases, then there will be no error resulting from this manner of
sampling.

.
FIG. 21-17. An improved version of the amplifier of Fig. 21-16.

A second feature is one which is important in certain types of applica-

tion. It will be noted from Figs. 21-16 and 21-17 that the deflecting yoke
is in the plate lead of the current-amplifier tube. Likewise, with the
clamping tube that is provided, the position of the cathode-ray beam at
the start of the sweep will be independent of sweep amplitude or sweep
speed. This is a very desirable characteristic in a radar indicator, since it
avoids the requirement for a centering voltage, which will depend on the
sweep speed.

FIG. 21-18. A feed-back amplifier with an a-c coupled output.

If provision must be made for controlling the position of the start of

the sweep, the yoke may be a-c coupled to the current amplifier by means
of a transformer and by providing a separate source for effecting the
desired displacements. The elements of a circuit that permits this type
of operation are given in Fig. 21-18. In this circuit the current in the
deflecting yoke is sampled, amplified, and combined with the input saw
tooth in a difference amplifier. The output from such an amplifier is a
very satisfactory linear current wave.
468 ELECTRON-TUBE CIRCUITS [CHAP. 21

21-2. Circular-sweep 9"enerator. In certain applications it is found

desirable to employ other than a linear sweep of the cathode-ray beam
across the face of the tube. In some instances the beam may be deflected
across the tube face according to some prescribed function of time, not
necessarily linear. In other cases the cathode-ray beam may be caused
to describe a circular path, a spiral path, a rotating radial path, or, in
fact, any of a wide variety of paths. Several of such sweep generators
will be studied.
An elliptical or circular sweep of the electron beam can be readily
accomplished by making use of the fact that the application of sinusoidal
voltages which differ in time phase
,, by 90 deg to the two sets of de-
Sine
wave
flecting plates of the cathode-ray
inpuf tube which are 90 deg apart in
space will generate an ellipse, if
the vertical deflection is different
FIG. 21-19. Phase-shifting circuit for pro- from the horizontal deflection, or
ducing an elliptical or circular trace.
.will generate a circle, if the two
deflections are the same. The amplitude of the voltages must be slightly
different in general in generating a circular sweep in order to take into
account the different deflection sensitivities of the deflecting plates, owing
to their different distances from the screen.
The simplest circuit for obtaining the necessary two voltages which are
90 deg apart in time phase is illustrated in Fig. 21-19. The circle diagram
of this circuit is given in Fig. 21-20, and it will be seen that, for a fixed

FIG. 21-20. The circle diagram of the FIG. 21-21. An improved simple phase-
phase-shifting network of Fig. 21-19. shifting circuit.
value of C, the degree of ellipticity, which depends upon the relative
voltage across Rand C, is controlled by controlling the value of R. The
potentiometer P controls the amplitUde of the sine-wave voltage that is
applied to the phase-shifting circuit. If this voltage is large, the voltages
across both Rand C will also be large. Consequently, the circle traced
on the screen will be of large diameter.
If the source of voltage for the sweep generator contains appreciable
harmonics, the trace cannot be made circular. A modification of the
circuit of Fig. 21-19 is possible which will avoid most of the errors arising
SEC. 21-2] SPECIAL SWEEP GENERATORS 469

from this cause. It consists in replacing the resistance of the circuit by a

parallel circuit that is tuned to the frequency of the supply source. The
circuit then has the form illustrated in Fig. 21-21. At the supply fre-
quency, the impedance of the parallel circuit is equivalent to a pure
resistance of value Lt/C1R 1, where Rl is the effective resistance of the
induct. With this circuit, the parallel branch impedance is capaci-
tive for the harmonic frequencies, and so the variation of this impedance
with frequency is substantially the same as that for the capacitor C, with

tI
1
J
I '--'----+I-
~he ~--+-~~~~~--~
wave
vO/~l7fge ~
1
1
1
1
:
I1

FIG. 21-22. A circular-sweep generator circuit.

the result that the amplitudes of the harmonic potentials applied to the
two deflector plates are sUbstantially reduced and the circularity of the
trace is not seriously impaired.
The main advantages of such a circular trace are the ease of generation,
the increase in the length of trace possible on a given cathode-ray tube
face, and the avoidance of flyback. A disadvantage is that such sweeps
are ordinarily used with tubes which require the presence of a central
electrode to which the signal is applied in order to obtain a deflection of
the trace. However, it will be shown that the use of such special tubes
is not required since it is possible to obtain a radial deflection without the
use of the central electrode.
The disadvantage of these generators is that the omission of push-pull
sweep deflection results in trapezium distortion unless the tube is specially
designed to reduce this. The effect of the distortion is readily avoided
470 ELECTRON-TUBE CIRCUITS [CHAP. 21

through the use of the somewhat more elaborate circuit illustrated in

Fig. 21-22. This circuit, which is based on the same principles as above,
provides a greater flexibility.
In this circuit the values of C and R are chosen sO.that a circular sweep
will be generated at the frequency of the applied voltage. Push-pUll
deflection is provided in order to reduce trapezium distortion. The gain
of the four amplifier tubes is controlled by the voltage on the screen grids,
which is controlled by the voltage at the plate of T5, a quantity that is
set by the setting of the bias po-
tentiometer in the grid circuit of
this tube.
If no signal is applied to the grid
of T5, the voltage at its plate will
be constant and a circle will be
generated. However, if a signal ~) ~)
is applied to the grid of T5, the FIG. 21-23. A circular and a spiral trace
voltage at the plate will be possible with the circuit of Fig. 21-22.
changed in accordance with the
shape of the input signal, assuming that there is no distortion. This will
cause the diameter of the circle to change in accordance with the applied
signal, whence the signal is made to appear on the beam. Such a pattern
may have the form illustrated in Fig. 21-23a.
To obtain the spiral trace illustrated in Fig. 21-23b, it is necessary only
to apply a saw tooth to the input of T5, the frequency of which is a sub-
multiple of the frequency of the applied sine wave. In particular, suppose
that the applied sine-wave voltage is 1,000 cps and that a saw-tooth
voltage of 200 cps is applied to the grid of T5. The gain of the amplifier
tubes will be varied continuously during the 5 cycles of the sine wave.
As the gain increases from zero to the maximum, the spot rotates on a
circle of ever-increasing diameter, returning to the center during the
flyback period of the saw tooth. If the saw tooth is not a submultiple
of the sine wave, the spiral wiII revolve.
By substituting current-amplifier tubes for the voltage-amplifier tubes
in the push-pull circuit and replacing the electrostatic deflecting system
by an electromagnetic deflecting yoke, the foregoing discussion is valid
for the electromagnetic type of tube.
21-3. Rotating Radial Sweep. A radial linear sweep which can be
rotated with uniform angular velocity is extensively used in radar appli-
cations. Such an indicating system is known as a plan position indicator
(PPI) since, in its operation, the rotation is synchronized with the scan-
ning of the radar system, the resulting display being in effect a plan view
of the scanned area. A number of methods have been used to generate
such a display.
SEC. 21-3] SPECIAL SWEEP GENERATORS 471

The simplest method of producing the plan-position-indicator sweep

incorporates the magnetic deflecting yoke which produces the nominal
linear deflection in a mechanical assembly which can be rotated physi-
cally. The physical rotation of the mechanical assembly can be synchro-
nized with the primary driving source either by means of rigid mechanical
couplings or by means of a servomechanism, a device which permits
accurate follow-up of driven and driver systems by purely electrical
interconnection.
A second method that has been extensively used utilizes a so-called
two-phase selsyn transformer. Such a device consists of two stator

Hj:=#=ITI DefYeclintJ
i-l yoke

FIG. 21-24. A selsyn transformer circuit for producing a rotating radial sweep.

windings which are placed at right angles to each other and which are
excited from a single-phase movable rotor. The application of, say,
a trapezoidal wave form (which would theoretically provide a linear-
sweep current) to the primary will yield two output voltages which are
in time phase but the amplitudes of which vary respectively as the sine
and cosine of the angle of the rotor or primary with respect to the second-
ary windings. These outputs when amplified and applied to two deflect-
ing coils at right angles to each other will produce a display of the type
here considered. That this is so follows readily from the fact that the
horizontal and vertical magnetic fields are proportional to
- Bx = cp(t) cos cp
By = cp(t) sin cp
from which it follows that
B = VB~ + B; = cp(t)
That is, the resulting field, and hence the deflection, at any angle cp is
constant in amplitude and of the form to produce a linear saw-tooth cur-
rent wave. Of course, the requirements imposed on the selsyn trans-
472 ELECTRON-TUBE CIRCUITS [CHAP. 21

former is that it faithfully reproduce in the output circuit the wave shape
of the input wave.
The main elements of such a circuit are given in Fig. 21-24.
REFERENCES
As general references, see
Puckle, O. S., "Time Bases," John Wiley & Sons, Inc., New York, 1943.
Soller, J. T., M. A. Starr, and G. E. Valley, Jr. "Cathode Ray Tube Displays,"
M.LT. Radiation Laboratory Series, Vol. 22, McGraw-Hill Book Company, Inc.,
New York, 1948.
PROBLEMS
21-1. The deflecting yoke of a magnetic cathode-ray tube has an inductance of
8.9 mh and a resistance of 15 ohms. It is connected in the circuit of Fig. 21-3.
The initial current in the coil is 80 rna when S is opened. Calculate and plot the
current through the coil when C = 0.01, 0.1 and 1 J.f,f, respectively. Estimate
the range over which a linear current within 5 per cent exists.
21-2. The deflecting yoke of Prob. 21-1 is connected in the circuit of Fig. 21-11.
Tube Tl is a 6J5, T2 is a 6V6, Ebb = 300. The clamp sets the grid bias of T2 at
-50 volts. If the current wave through the yoke is to vary linearly from 0 to
80 rna in 60 J.f,sec,
a. Specify the trapezoidal wave that must be applied to the grid of T2.
b. Specify a set of constants of the circuit of Tl to yield such a trapezoidal wave.
 CHAPTER 22

ELECTRONIC INSTRUMENTS

A NUMBER of instruments have been developed which incorporate

vacuum tubes as an integral part of these devices. These ip.struments
possess certain advantages over the standard electromechanical varieties.
In particular, most of the vacuum-tube types of instruments possess
relatively high input impedance and also may be used over a very wide
range of frequencies. These instruments are divided into the following
general classes: voltmeters, ammeters, power meters, frequency meters,
und phasemeters.
Perhaps the most versatile of all electronic instruments is the cathode-
ray oscilloscope, which may be used as a voltmeter, ammeter, frequency
meter, and phasemeter, depending upon the manner of its connection in a
test circuit. The applications to these various services will be discussed
below in the appropriate section.
22-1. Electronic Voltmeters. The Oscilloscope. The applicability of
the oscilloscope as a voltmeter arises from the fact that the deflection
sensitivity of the cathode-ray tube, for a given accelerating potential, is
a constant and the deflection of the cathode-ray beam on the screen is
directly· proportional to the applied deflecting potential. Thus, by set-
ting the gain of the amplifiers associated with the oscilloscope at some
convenient value, the peak of any applied potential may be compared
with a calibrating potential.
Such devices are very flexible but are usually limited to the lower fre-
quencies, owing to the frequency distortion that occurs in the amplifiers
at the higher frequencies. If connection is made directly to the plates of
the tube, this frequency limitation is avoided and the tube may be used
to the very high frequencies. However, as the deflection sensitivity is
usually low, of the order of 40 volts/in. on a 5-in. tube with an applied
accelerating potential of 2,000 volts, such a voltmeter would be limited
to the higher potentials, in the range from 5 to 150 volts. For higher
voltages, a potential divider might be used. Lower voltages cannot be
measured, except with the aid of amplifiers, and with their consequent
fre1uency limitations.
Vacuum-tube Voltmeters. The vacuum-tube voltmeter provides a con-
venient method for the accurate measurement across high-impedance
473
474 ELECTRON -TUBE CIRCUITS [CHAP. 22

sources of both d-c and also a-c voltages up to very high frequencies.
The use of the conventional type of tube usually limits the operation to
frequencies below about 1.5 megacycles. For frequencies higher than
this, the grid-circuit loading, resulting from the input capacitances of the
tube and the electron transit time, usually becomes sufficiently serious to
influence the operation. Where the loading effects are permissible, the
use of suitable standard receiving-type tube is permissible to frequencies
of approximately 20 megacycles without serious calibration errors.
For the measurement of voltages having frequencies above 1.5 mega-
cycles, to perhaps 30 megacycles, Acorn type tubes in a special probe
construction may be used without introducing appreciable loading. With
such a probe construction, the lea.d impedance is reduced to zero by allow-
ing direct connection to be made from the voltage source to the grid of the
tube. Also, the shunting effect of the grid resistor is made negligible
either by neglecting it entirely or by using resistors of low self-capacitance
in series to provide from 5 to 10 megohms across the grid to cathode.
A wider range of operation, perhaps to 100 megacycles or higher, is
possible by the use of the Acorn type diode rectifier. Such a circuit
provides a low input capacitance and low transit time, and although a
liigh efficiency of rectification is possible and a high average impedance
may be obtained when the load is of the order of 50 megohms, the input
impedance during the peak of the positive half cycle of the applied volt-
age decreases. Also, the reading is dependent on the impedance of the
source, which results in a reduction of the applied voltage at the terminals
of the voltmeter. If the tube diode is replaced by one of the present-day
crystals, the range of a vacuum-tube voltmeter is extended somel"hat, to
perhaps as high as 500 megacycles.
All a-c voltmeters are essentially rectifiers, employing diode, grid-
circuit, or plate-circuit rectification. They may be grouped according
to the value of the wave form of the applied voltage to which their read-
ings are proportional, as rms, average, peak, or logarithmic. A number
of circuits that are currently in use will be discussed.
RM S voltmeters. The response of this type of voltmeter is proportional
to the rms value of the applied
voltage. To achieve such a re-
sponse, the rectifier that is used
must have a square-law relation
between the applied input voltage
and the mean rectified current.
FIG.22-1. The basic circuit of a vacuum- Such voltmeters may therefore be
tube voltmeter. used to measure the rms values of
voltages, regardless of what the wave form may be.
SEC. 22-1] ELECTRONIC INSTRUMENTS 475
The basic circuit of such an instrument is given in Fig. 22-1, which is
essentially a conventional thermionic vacuum-tube circuit, except that
an auxiliary d-c circuit is applied across the indicating milliammeter in
the plate circuit in order to balance out the steady component of the
tube current when. the input voltage is zero. In this way, the milliam-
meter reads the average rectified current that results when the a-c voltage
is applied to the input terminals.

FIG. 22-2. Vacuum-tube voltmeters with balanced output.

It is possible to arrange the circuit in such a way that the tube forms
one arm of a balanced Wheatstone bridge. In this way, the static
operating current may be balanced out without the need for a separate
d-c source. Two such circuits are illustrated in Figs. 22-2. In these
circuits, to achieve good sensitivity, the resistor Ra must be made large
in comparison with the meter resistance.
A characteristic which is closely square law for a limited range of
applied voltage, and this is usually less than about 1 volt peak, may be
obtained by operating a triode or
a pentode on the curved portion
of the characteristic curve. To
ensure that the negative peak of
the applied voltage will not ap-
proach too closely to the cutoff
bias of the tube, the static plate
current must be slightly greater
than twice the increment of plate
current required to produce full-
FIG. 22-3. Illustrating the operation of an
scale deflection of the indicating rms voltmeter.
meter. The situation is best un-
derstood by referring to Fig. 22-3, which illustrates the operation of the
above circuits.
Values of applied voltage higher than 1 volt peak may be measured
without affecting the square-law property of the voltmeter by using a
voltage divider across the source of voltage and applying a known small
fraction of the total voltage to the input terminals. When this is done,
however, the desirable property of very high input impedance is lost.
476 ELECTRON-TUBE CIRCUITS [CHAP. 22

In general, also, the calibration of this type of instrument is dependent

upon the tube maintaining its square-law characteristic. As a result,
frequent calibration is usually necessary.
Average-reading voltmeters. The response of an average-reading volt-
meter is proportional to the average value of the jnput voltage. To
achieve such a response, there must be a linear relation between the
applied input voltage and the mean rectified current of the rectifier that
is used. But as the mean rectified current is proportional to the mean
of the positive excursions of the applied voltage, the instrument is depend-
ent on the wave form of the input voltage.
The rectifier that is employed may be either a simple diode rectifier
without a shunt capacitor, a diode bridge circuit, or a biased triode or

FIG. 22-4. The circuits of several average-reading vacuum-tube voltmeters.

pentode that is operated from approximately cutoff over the linear portion
of the dynamic curve. Owing to the fact that the dynamic curve is
nonlinear at low voltage levels, a substantially linear response is obtained
only for relatively large values of applied voltage. The linearity may
be considerably improved by using a large lead resistor in the diode
circuit or by the use of a high plate resistor and negative feedback in the
case of triodes or pentodes. Several circuits of such average-reading
voltmeters are given in Fig. 22-4.
Since the reading of these voltmeters is dependent upon the wave form
of the input voltage wave, then when the applied voltage is not a sine
wave or other symmetrical wave form, a reversal of the polarity of the
input wave will, except for the bridge circuit, change the reading of the
instrument. This effect is known as turnover.
Peak-reading voltmeters. The response of a peak-reading voltmeter is
proportional to the peak value of the applied voltage and is, therefore,
independent of the wave form when it is calibrated in peak volts. Peak-
reading voltmeters may be calibrated to read rms values for sinusoidal
voltages, which then correspond to 0.707 of the peak value. A number
of peak-reading voltmeters are possible. They include diode, grid-leak,
feed-back, and slide-back types.
The diode peak voltmeter provides one of the most convenient and accu-
rate methods of measuring peak voltages, especially at radio frequeneies.
SEC. 22-1] ELECTRONIC INSTRUMENTS 477

It consists of a conventional diode rectifier provided with a capacitor input

filter. The capacitance that shunts the load resistance is so chosen that
the time constant of the circuit is large compared with the period of the
applied voltage. A nonlinearity exists for low input voltages which
causes an error, but the indication is linear for voltages above about 10
volts. The circuits of two such voltmeters are shown in Fig. 22-5.

(a) (bJ
Positive peak, NegOitive peak
FIG. 22-5. Peak-reading voltmeters.

The shortcoming of these voltmeters is that the input impedance during

the conducting portion of the cycle is different from that during the non-
conducting portion of the cycle. This difference arises from the fact that
the diode consumes power during the conducting portion of the cycle.
This effect is precisely that discussed in Sec. 16-3, where it was shown that
the effective input resistance to the diode Re = R/2'YJ and is approximately
R/2, where R is the load resistance. Ordinarily R is so large that voltage

T
e
I

ra) (h)
Difference amplifier Gener",i ROiclio Co.
FIG. 22-6. Diode peak voltmeters with d-c amplifier.

fluctuations which might result from such a changing load impedance are
negligible.
When the rectified current through the load resistance is too small to
be measured conveniently by a d-c microammeter, the voltage developed
across the load resistance may be applied to the input of a d-c amplifier.
Several such circuits are illustrated in Fig. 22-6. In circuit a, a difference
amplifier with e2 = 0 is used as the d-c amplifier. In circuit b, a single
tube balanced amplifier is used.
The grid-leak peak voltmeter consists of a grid-circuit rectifier employing
either a triode or a pentode. Rectification occurs in the grid-cathode
478 ELECTRON-TUBE CIRCUITS [CHAP. 22

circuit in the same manner as in a diode, and grid current flows over the
positive half cycle of the applied voltage. The input impedance is sub-
stantially the same as that of the diode peak-reading voltmeter.
A negative feed-back peak voltmeter consists essentially of a self-biased
tube which operates from approximately cutoff over the linear portion of
the dynamic curve. To maintain the tube almost at cutoff for all except
the peak portions of the cycle, the bias resistor must be shunted by a
capacitor which is adequate for essentially class B operation. As a
result, the plate current flows only at the positive peaks of the applied
voltage. The indications on this type of voltmeter are dependent on the
wave form. Since the grid is not driven positive over any portion of the
cycle of the applied voltage, the input impedance of the voltmeter is high.
The circuit connections of such a voltmeter for use with either a 6J7 or
a 954 Acorn tube, depending upon the frequency range required, are
shown in Fig. 22-7. The 6J7 is
satisfactory for use to several meg-
acycles, but where the voltmeter
is intended for use across high-
impedance tuned circuits at fre-
quencies as high as 10 megacycles,
and in cases where the input ca-
FIG. 22-7. A negative feed-back peak pacitance must be kept small, the
voltmeter. 954 is used. Note that the tube is
connected as a triode, with the
suppressor connected to the cathode. This instrument may be used to
measure d-c voltages by omitting the blocking capacitor in the input,
although a separate calibration must be used.
The slide-back voltmeter consists essentially of a threshold indicator, this
threshold being indicated by a d-c volt-
meter when the d-c voltage is made equal
to the peak value of the applied potential.
The circuit of the instrument is illustrated
in Fig. 22-8. The triode or pent ode is
operated at a very low value of plate cur-
rent, and the bias is read on the d-c volt-
meter. The voltage to be measured is then
applied to the input terminals, and the
grid bias is increased until the plate cur- FIG. 22-8. A slide-hack peak
rent is reduced to its initial value. The voltmeter.
peak of the applied voltage is then equal to the increase of grid bias, as
obtained from the d-c readings.
This type of voltmeter is true peak-reading, and owing to the method
of operation it is self-calibrating. It is completely independent of varia-
SEC. 22-1] ELECTRONIC INSTRUMENTS 479
tions in operating voltages and tube characteristics. Moreover, it has a
very high input impedance and operates over a very wide range of volt-
ages simply by changing the slide-back voltage. Because of the method
of operation, such voltmeters are restricted to steady sources of potential.
Logarithmic voltmeter. A voltmeter with a logarithmic scale is possible
by using a variable-mu tube in which the amplification factor is an expo-
nential function of the grid voltage. Thus for such a tube circuit

(22-1)

where a and A are constants and Ec is the d-c bias on the tube.

FIG. 22-9. Block diagram of a logarithmic voltmeter.

In this voltmeter, the output is maintained constant by rectifying

and filtering the output and using the resulting d-c voltage to bias the
amplifier tube. That is, with e2 constant,
1
Ec = K - - loge e
a
But since
db = 20 loglO ~ = K' - 8.69 loge e
e
then
db = K' - 8.69a(K - Ec)
or
db = k + mEc (22-2)
where K, K', k, and m are constants. A block diagram of this circuit
appears in Fig. 22-9.
In another type of instrument, illustrated in Fig. 22-10, use is made of
the fact that the average diode
cur r e n t varies logarithmically
with the input voltage to the
variable-mu pentode, over a large
range of input voltage. ----.I--+--,...,\--+-;
D-c voltmeters. Most of the
circuits considered as a-c volt-
FIG. 22-10. A logarithmic voltmeter.
meters will operate to give a·
meter deflection for an input d-c potential applied directly to the grid
of the tube. However, since rectification is not required for such an
480 ELECTRON-TUBE CIRCUITS [CHAP. 22

instrument, the tubes might be used for amplification and also to provide
a very high input impedance.
The cirCUIt of Fig. 22-11 is one which possesses a very high input imped-
ance. The indicating instrument is a low-resistance milliammeter,
having a resistance Rm. An analysis of the equivalent circuit shows that
the meter current is
1m = fJ.RE (22 3)
[R + rp + (fJ. + 1)Rk](2R + Rm) - 2R -

where E is the d-c potential being measured. However, for a pentode

and the current through the milli-

ammeter is given by

I
E
= ---- (22-4)
m 2Rk

This shows that a linear relation

exists between the applied poten-
FIG. 22-11. A high-impedance d-c volt- tial and the reading of the indicat-
meter.
ing instrument.
A voltmeter employing triodes, which utilizes a high-resistance volt-
meter as the indicating instrument, may be built using the simple differ-
ence amplifier. The circuit of such an instrument is given in Fig. 22-12.
As shown in Sec. 6-9, the potential difference across the high-resistance
indicating instrument is given by
fJ.R
E = 2(rp + R) e (22-5)

22-2. Electronic Ammeters. The oscilloscope may be used to measure

currents, although an indirect
method must be adopted. In this
measurement, the current to be
measured is passed through a cal-
ibrated resistor, the resulting volt-
age across the resistor being given
on the oscilloscope screen. The
current may then be calculated
from the known voltage and FIG. 22-12. A d-c voltmeter employing
Iiesistance. However, owing to triodes.
the fact that the input impedance of the oscilloscope is only mod·erately
high, the oscilloscope is limited to the measurement of relatively high
SEC. 22-3] ELECTRONIC INSTRUMENTS 481
currents, of the order of that possible with good-quality d-c instruments,
say 0.1 ,uamp.
Electrometer Tubes. The amplification properties of a vacuum tube
may be used to amplify extremely small currents, which are then indicated
on a sensitive galvanometer. In order to measure these small currents,
it is essential that the grid current of the tube be very small compared
with the currents to be measured. Normal tubes, if selected for quality,
and if operated at reduced electrode voltages, are satisfactory over
rather wide limits. Special tubes with unusually small values of grid
current are available. These include, among others,
General Electric Co. FP-54 I g roov 10-15 amp
Victoreen Inst. Co. VX-41 I g roov 10-15 amp
Raytheon Mfg. Co. CX-570A I g roov 10- 14 amp
A circuit utilizing a General Electric FP-54 electrometer tube and
operating from a 12-volt storage battery is given in Fig. 22-13. Even
though the FP-54 tube is a low-grid-current type, great care must be

!onlz""tlon
chamber

FIG. 22-13. An electrometer circuit using an FP-54 tube.

exercised in keeping leakage and surface currents to a minimum if high

sensitivity and dependable operation are desired. This is frequently
accomplished by incorporating the entire circuit in a suitable probe type
of construction, which is shielded electrostatically and which is kept
dry by the use of a drying agent. Also, the grid current is kept small by
driving the circuit at low potential from batteries, the circuit being kept
in continuous operation for long periods of time in order to reduce tube
drift. With the simple circuit shown, the indicating instrument being a
sensitive wall galvanometer, currents of the order of 10-14 amp may be
reliably measured.
22-3. Phase meters. Lissajous Patterns. A Lissajous pattern is the
figure created on an oscilloscope screen when sine-wave voltages are
applied to both the horizontal and vertical deflecting plates. If the
frequencies of these two component voltages are the same but differ in
phase, the resulting pattern is a measure of the phase difference between
the two waves.
482 ELECTRON-TUBE CIRCUITS [CHAP. 22

To see that this is so, suppose that the voltage across the horizontal
deflecting plates of the oscilloscope is denoted as
Ex = E1 sin (wt +8 1)

and that across the vertical deflecting plates is given by

Ey = E2 sin (wt +8 2)
Then
E
E x. 8
1 = sm wt cos 1
+ cos wt sm
. 8 1 (22-6)
and
Ey . .
E2 = sm wt cos 82 + cos wt sm 8 2 (22-7)

Now eliminate the time factor wt. To do this, multiply the first equation
by cos 82, and subtract the second, multiplied by cos 81 • This gives

~: cos 82 - ~: cos 81 = cos wt(sin 81 cos 82 - cos 81 sin 82)

Similarly, multiplying the first by sin 82 and the second by sin 81 and
subtracting, there results
Ex . 82
E1 sm -
Ey.
E2 sIn 81 = sm
. wi. (cos 8 ' 82
1 sm -
. 81 cos 82)
sm

Squaring and adding these equations gives

E;,
2 + 2E; - 2ExEy
- - cos (81 -
82) = sm
. 2 (8 1 -
82 ) (22-8)
E1 E2 E1E2
This is the equation of an ellipse whose principal axes coincide with the
coordinate axes when 81 - 82 = 11"/2. Hence, in most cases, the result
is an ellipse, the orientation of which depends upon the phase difference
between the two waves. The results have the form illustrated in Fig.
22-14.

.:J(}=o

FIG. 22-14. Lissajous patterns showing the effect of the phase difference of two
l:lources.
SEC. 22-3] ELECTRONIC INSTRUMENTS 483
If the amplitudes of the voltage applied to the vertical and horizontal
deflecting plates are equal, then the pattern at the phases t10 = 7r /2 and
37r/2 will be circular. For the meas-
urement of phase, it is customary to
set the two amplitudes equal to each
other, although this is not necessary.
The experimental procedure neces-
sary for measuring the phase differ-
ence between the two voltages is quite
direct and consists in measuring the FIG. 22-15. To measure the phase
two distances Eo and Em illustrated in difference between two voltages by
means of the Lissajous pattern.
Fig. 22-15. The distance Eo is evi-
dently the value of Ey when E:r; is zero. This requires that
Eo = E2 sin (01 - O2)
Likewise, the value of Em is the maximum value of Ey and fromEq. (22-7)
is given by
Em = E2
Solving these expressions simultaneously yields
. (
sm 01 -
0)
2 = 'Eo
Em (22-9)

Electronic Phasemeter. The elements of a simple electronic phasemeter

are illustrated in Fig. 22-16. It will be observed that the circuit com-
prises in essence two diode peak-reading voltmeters which have been

FIG. 22-16. An electronic phasemeter.

differentially connected. With no input signal, the voltages appearing

across the input of each diode-voltmeter circuit arise only from the
impressed reference voltage. When the circuit is balanced, a zero reading
appears on the zero-center high-resistance voltmeter. This high-resist-
ance voltmeter might be a d-c vacuum-tube voltmeter with a zero-center
indicating instrument.
If now a voltage of preset amplitude is impressed across the input, the
484 ELECTRON-TUBE CIRCUITS [CHAP. 22

potential at one anode decreases and becomes, say, el - e2, while the
potential at the other anode increases and becomes el + e2. As a result,
the output potentials will no longer be the same, and a reading will appear
on the output voltmeter. The extent of the change will depend directly
upon the phase of the input voltage with respect to the reference potential,
being a maximum in one direction for zero phase difference, and being a
maximum in the other direction for 180 deg phase difference, there being
a calculable phase difference for every intermediate reading. The mathe-
matical analysis parallels that of the f-m discriminator and follows below.
Suppose that the voltages el and e2 are
el = E. sin wt (22-10)
and

e2 = E. cos (wt + cp) = E. sin (Wi + ~ + cp) (22-11)

The a-c potential that appears in the circuit diode Tl is

and that in the circuit of T2 is

But ea may be written explicitly as

ea = E. sin wt + Es cos (wt + '1')
which reduces to

ea = 2E. cos ~ ('I' + ;) sin [ wt + ~ (cp + ;) J

Similarly
eb = - 2E. sin ~ (cp + ~) cos [ wi + ~ ('I' + ;) J

l
The corresponding d-c output potentials are then given by

Ea = 2rJE. cos ~(cp +;) (22-12)

Eb = 2rJE. sin ~ ('I' + ;)
where rJ is the detector efficiency (see Sec. 16-3). The corresponding
indicating-voltmeter reading is then

E = Ea - Eb = 2rJE s [I cos ~ ('I' + ;) I-I sin ~ ('I' + ;) IJ (22-13)

The result has the form illustrated in Fig. 22-17.

SEC. 22-3] ELECTRONIC INSTRUMENTS 485

A somewhat similar type of phasemeter is possible using triodes. 1 A

circuit is shown in Fig. 22-18. It is supposed that the two voltages el
and e2 are of equal frequency and
of constant amplitude. The equal 1.0 E/2'lEs
voltages appearing across the second-
aries of the input transformers are
connected in series opposition so that, qJ
when the input voltages are in phase, 11"
the voltage across ab is zero. When
the input potentials are slightly out
of phase, a small voltage appears -1.0
across ab the magnitude of which is F IG. 22-17. Tlle output of the
proportional to the phase angle. For phasemeter of Fig. 22-16.
small angles this voltage may be
considered to be gO deg out of phase with either input.
The phase-shifting network causes a voltage e' which is gO deg out of
phase with e, the voltage across ab, to appear across cd. This voltage is
applied across the grid resistor R and excites the grids of tubes T1 and T2,
which are connected in push-pull. Owing to the initial gO-deg phase shift
and the added gO-deg shift, then for an a-c potential applied to the plates

~ ~-------~--------~~-----+
FIG. 22-18. A double-triode phasemeter.

the grid .voltage of one tube is in phase with its plate, whereas the grid
potential of the other is in phase opposition with the plate. As a result,
one tube conducts, while the other is cut off, either because of a highly
negative grid or because of a negative plate. The conduction causes the
zero center instrument to deflect according to which tube conducts.
The instrument must be zero-set initially with no signal on the grids,
whence the unidirectional pulses of plate current through the tubes divide
equally through the plate resistor, so that zero resultant voltage appears
across the d-c instrument. In this instrument, since e' is proportional to
(j for small angles, the magnitude of the voltage across the d-c instrument
is proportional to (j.
Should either el or e2 drop below or exceed normal rated voltage, the
difference voltage ab increases in length and rotates. This sinor may
486 ELECTRON-TUBE CIRCUITS [CHAP. 22

be resolved into two components, one of which is in phase with c, and

will be effective in producing the voltage across the indicating instru-
ment. The other component, which is 90 deg out of phase with e, will
not appreciably effect the accuracy of the instrument.
22-4. Wattmeters. The circuit of an electronic wattmeter is given .in
Fig. 22-19. In this circuit, the resistors R2 are equal and sufficiently small
so that the voltage drop across
Inpuf
them does not appreciably re-
duce the load voltage. The
fio
resistor Rl is high in order to
prevent appreciable power
loss.
To examine the operation,
Load it is supposed that the first two
terms of the series expansion
FIG. 22-19. An electronic wattmeter.
adequately represent the dy-
namic characteristics of the tubes. Thus
i pl = al(el + e2) + a2(el + e2)2 (22-14)
and
i p2 = al(el - e2) + a2(el - e2)2 (22-15)
The output meter reads the difference between i pl and i p2. Hence
io r-v i pl - i p2 = 2ale2 + 4a2ele2 (22-16)
But with
kel = kEm cos wt }
and (22-17)
e2 = R21 = Rdm cos (wt + (J)
and since the indicating instrument reads the average, then

10 = -
1 fT. ~o dt
T 0
2al (T () {T
= T Jo Rdm cos wt + (J dt + 4a2R2EmImk
T Jo cos wt cos (wt + (J) dt

so that finally
(22-] 8)
If the load contains harmonics, it can be shown that the reading of the d-c
meter is proportional to the total load.
22-5. Frequency Meters. Lissajous Patterns. If the component volt-
ages that are applied to the horizontal and vertical deflecting plates of an
oscilloscope differ in frequency, as well as in phase, the pattern that results
is far more complicated than the simple ellipse that has already been con-
sidered. If the frequencies are in a constant ratio with each other, the
SEC. 22-5] ELECTRONIC INSTRUMENTS 487
resulting patterns permit a comparison of the ratio of these frequencies.
If one of these frequencies is known, the method permits a comparison of

A rnA B
IT~_-- given standard.
!v1'J,. L-\,}--V -
the unknown frequency with the

Suppo,", that a voltage E. ~ E

~ EZt~= : i j l i e d to the hod,"ntal

i ti Voltaqe applied 10 the vertical

defledmq plates
I Signal on horizontal
deflecting plates

kP
J
I

quency ratio.
Voltage applied to the
1 horizontal def/ecfinq plates

FIG. 22-20. Lissajous figure for 1: 2 fre- FIG. 22-21. Lissajous pattern for 2: 3
frequency ratio.

plates and a voltage Ey = E sin (W2t + 0) is applied to the vertical

plates. If the frequencies are in the ratio wz/ WI = 2, the form of the
resulting patterns is readily calculated. The construction is illustrated
in Fig. 22-20. In a similar way, if the frequencies are in the ratio 2: 3,
the Lissajous pattern assumes the form shown in Fig. 22-21.

1:1 2: 1 1 :5 5:3
FIG. 22-22. A number of Lissajous patterns.

A feature that is common to all Lissajous figures is that the horizontal

line AA and the vertical line BB are tangent to the pattern at a number
of points, the number depending, respectively, on the frequency applied
to the vertical and to the horizontal deflecting plates. Thus the fre-
quency ratio
Wh _ no. of points at which the figure is tangent to the vertical line
w" - no. of points at which the figure is tangent to a horizontal line
A number of Lissajous patterns are contained in Fig. 22-22.
Electronic Frequency Meters. The circuit of such an electronic device
is given in Fig. 22-23. The input causes periodic switching from Tl to
488 ELECTRON-TUBE CIRCUITS [CHAP. 22

T2. When the current begins to flow in TI, the potential across Rkl
causes C1 to charge through the par-
allel paths RI and one plate of the
rectifier T3. The average diode cur-
rent is indicated by the meter in
series with the diode cathode. The
diode current is proportional to the
frequency, over the range for which
eland c2 charge "fully" in less
than the time of one half cycle of
the impressed voltage. For the
circuit shown, the calibration of the
FIG. 22-23. An electronic frequency instrument is linear to approximately
meter. 7,000 cps.
REFERENCES
As general references, see
Reich, H. J., "Theory and Application of Electron Tubes," McGraw-Hill Book
Company, Inc., New York, 2d ed., 1944.
Langford-Smith, F., "Radiotron Designer's Handbook," Amalgamated Wireless
Valve Co., 1941.
1. Mikelson, W., Gen. Elec. Rev., 41, 557 (1938).
PROBLEMS
22-1. The plate current in a 6SK7 tube with Ecc2 = 100, Ebb = 250, is approx:
10 = 18eO. 20E, ma
This tube is used in a Ballantine voltmeter circuit. An a-c potential of 18 volts
peak is applied. What must be the shift in bias potential if the voltmeter is to
read correctly?
22-2. The voltage applied to the input of an average-reading vacuum-tube
voltmeter has the form shown in the figure.

0,.What is the zero input reading of the d-c milliammeter?

b. Calculate the reading when the signal is applied.
c. If the input voltage were sinusoidal with the same peak amplitude, what
would be the corresponding reading?

22-3. Show the connections for employing the multiplying circuit of Fig. 8-11
:1S an electronic wattmeter.
 APPENDIX A

MILLMAN THEOREM

T HE usual methods for the solution of the currents in or the potential

differences across the elements of a network involve a direct application
of Kirchhoff's laws. This involves setting up the mesh currents and then
equating the sum of the potentials around any closed loop equal to zero.
The number of independent loops that must be chosen must equal the
number of unknown currents to be found. The resulting algebraic
equations are then solved, ordinarily by using determinants, for the
desired unknown quantities. If a large number of branches are involved,
the resultant solution may become quite laborious. A network theorem
suggested by Millman * will frequently materially reduce the amount of
work involved in effecting the solution.
Consider a general network of the form illustrated in Fig. A-I. The
impedances Zl, Z2, and Z3 terminate in the common junction 0'. The
.I :hI

O--t--t

3 Z3

FIG. A-I. A general network for deriving the Millman theorem.

opposite ends of these impedances are numbered 1, 2, and 3. The point

o is any other point in the network. It is not necessary to know the net-
work interconnections between the points 0, 1, 2, and 3. The network
theorem states that the potential difference E oo ' between points 0 and 0'
is given by
(A-I)

where Eo. (8 = 1, 2, or 3) is the potential difference between point 0 and

the 8th impedance and where Y. = liZ. is the admittance corresponding
to the 8th impedance. Note specifically that no requirements are
imposed on the character of the element connected between points 0 and
* Millman, J., Proc. IRE, 28, 413 (1940).
489
490 ELECTRON-TUBE CIRCUITS lAPp. A

8 and between which there exists the potential difference Eo.. The proof
of the theorem follows.
The potential difference across impedance Zt is
E 10, = E 10 + EOO' = EOO' - EOI
The current through this impedance is, therefore,
E 10 , ,
110' = Zl = (EOO' - E01)Y1

Similarly, the currents in the other two branches are given by

E ,
120 , = - 20 = (EOO' - E 02 )Y 2
Z2
and
E ao'
Iso' = - = (Eoo' - Eoa) Y a
Za
But from Kirchhoff's point law, the sum of the three currents 1 10 " 1 20"
I ao', meeting at the point 0', must be zero.
Then
(EoO' - E 01 )Y 1 + (EOO' - E 02 )Y2 + (EOO' - E 03 )Y3 = 0
from which Eq. (A-I) is readily obtained.
This result is not restricted to three impedances, as shown in Fig. A-I,
but may be extended to any numbe~ n terminating in the common junc-
tion 0'. The generalized result is
n

I Eo.Y.
(A-2)
E oo' = 8,-=-,,1=,.---
~Y.
n

where the summation I

8=1
denotes the summation of the n terms.
 APPENDIX B

PLATE CHARACTERISTICS
OF RECEIVING-TYPE TUBES
AVERAGE PLATE CHARACTERISTIC
800
TYPE 5U4-G
E f =5.0 VOLTS A.C.

/
600
"
'"
w
a:
w ""
Q.
::; V"
'"
.J
.J
400
.; "
::;

O",'t, I.... "

~~ ....
~
200
"'~
/'
V
o
- :...--- .............
20 .00 60
PLATE

FIG. B-l.
VOLTS D.C.

5U4-G diode.
80 100 120
92C- 610'1

491
 ~
\0
t--:I

..,
fTl

1/
!:J
200
U'
<
~
(TJ
0
r ::0
-i
VI »
C) l':J
~ P1 to<
() l':J
"~ 15
r -U
r
Ci
~
f'1 »
~ ~
~ r'l 0
r Z
r () I

}; I ~
~
~IOO
»
::0
Lj
to
:l)
fT1
»
()
l':J
VI
-I Ci
fTI
.....
::0 ~
Ci
VI Lj
-4
("")
.....
~
VI C/:l

o 10 200 3 0 400 500 600

PLATE VOLTS
FIG. B-2. 2A3 triode.
Class A amplifier: Eb = 250 volts, Ec = -45 volts, h = 60 rna. f.L = 4.2, rp = 800 ohms, gm = 5,250 f.Lmhos.
Interelectrode capacitances: C, = 7 f.Lf.Lf, C 2 = 5 f.Lf.Lf, C, = 16 f.Lf.Lf.
 .
40
f 6 3 OL S
c2 • SO VO, TS
" /b Ec s
. ..,
(/) . / l-
I.JJ
0::
I.JJ
a..
I
:::E
« II
...J
30 ~,; 0
...J 1
:::E
:z / l-A: r-
t-
:z
III
I.JJ
0::
\ _/.0
0::
::J
<..>
20
V '\

0 '" I<. \

II -;;
:z
I.JJ 1\
I.JJ
0::
<..> tV \
I",
(/)

0:: IV .
2.1
0

.0
10
1/'x ~

1'.
- - -- -- -- ~: -- -- - _ .5
I.JJ
t-
Y. - - 1-- - -3 0
oCt
...J
a.. iY
IV::
IV
""<;;
-- r-'
--
1-' r-~
1--
'b
:~
C
-13·5
- .y-

0 100 zoo 300 400 500 600 70C

PLATE VOLTAGE IN VOLTS
FIG. B-3. 6AC7 pentode.
Class A amplifier: Eb = 300 volts, E,2 = 150 volts, E, = -2 volts, h = 10 rna, L2 = 2.5 rna, Tp = 1
megohm approx, (1m = 9,000 pmhos. Interelectrode capacitances: C, = 11 pl-'f, C 2 = 5 ppf, C s = .015 I"Pf.
 ~

....
\0

n
30 ~ ..,fTI
--i
;II
0
r ++- "
Cl' ~
"l>
r 206
;II
w rrJ
:;0
--i 0 (§ l>
fTI120 ........ r C)
.... n --i
(J)
:
-~
rrJ
t:rJ
lJ 10 ~
~ -I"1J r"
o;II t:rJ
r :I:r C"l
r
(II
» I'll> >-3
()
0 n
(J)
-i
~ ~
rrJ~
JJ
f"I ;II

~ 80 "fTI
;II
fTI
fTI l;n G)
0
~
f"I Z
f)
V> :1:"'" I

N
< ~l> >-3
0
r :u:;O ~
~ -1 »l> I;I;j
(J)
r 01 0 t:rJ
r
» " r-i
I'lrrJ C"l
~ 40
(]I
0 :;0 ....
~
f"I VI
;II C"l
f"I -i
~
(J)
0 ....
VI >-3
[JJ

o 100 200 300 400 500

PLATE VOLTS
FIG. B-4. 6AG7 pentocle.
Class A amplifier: Eo = 300 volts, E'2 = 150 volts, E, = -3 volts, 10 = 30 rna, [,2 = 7 rna, rp = 0.13 megohm approx,
gw = 11,000 Ilmhos. ~
Interelectrocle capa~~tanc()s: C , = 13 1l4, C2 = 7.5 Illlf, C, = 0.06 Ililf. :0
I;I;j
 14

~
$
12

(/)
UJ 10
a::
UJ
0.. "0
::::0
...: ~
-'
>
-'
8 >-3
::::0 trl
:z Ci
I-
:z
UJ
0::.
6 =
>
!:l:l
0::
:::J
U
>
Ci
.... >-3
...
I-
-' 4 trl
0.. !:l:l
1-1
if.!
>-3
1-1

2 Ci
if.!

o 100 200 300 400 500

PLATE VOLTAGE IN VOLTS
FIG. B-5. 6C5 triode.
Class A amplifier: Eb = 250 volts, E, = -8 volts, h = 8ma, I-' = 20, rp = 10,000 ohms, g", = 2,OOOI-'Illhos.
Interp]cctrode capacitances: C , = 3 I-'I-'f, C 2 = 11 I-'I-'f, C 3 = 2.0 1-'fJ-f.
 TRIODE CONNECTION
250
IEf 6 3 OL S

if)
w
a::
w
CL
I ~:-

:>
<I 200 I
,V
I
:z
~! x~ I
/ / ()/ I
I-
z
Ii v / i
I
~ 150
n::
I.L L
,()
I
/ / V /
/ V V / /
o
a: 100 / / / / / /
o
/ / / / :;
a::
o
1/ / / / / \Y i--

I~J
If. / / L L / /

50
,\[ / / L L 7:"
I \
/ 1)( / / V ~/
'X", V K v ,/
/' ,)V I
x '>./'
-
?2::
-- 1-;> ~ I-E..;. ~ (~ /'
,;:::. ~ ~ ~ ~1-", i2 ~ ::z:: ~ ;;:;.- ~ +]..{. -::: -E ~ ~
./
./
V
....- /' }';"'-
I

I
1-1
o
~

100 200 300 400

--- 500 600 700

PLATE VOLTAGE IN VOLTS

FIG. B-6. 6F6 triode (screen connected to plate).
Clas" A amplifier: Eb = 2,50 volts, E, = -20 volts, h = 31 rna, Jl. = 6.8, rp = 2,600 ohms, g,n = :;:J
2,600 ",mhos.
 PENTODE CONNECTION
250r--r-.--r-.--r~r-.--r-.--~-r-'--~-r-.--r--r-.--r-.--'--r-.--r--r-.--r-,40~
Ef· .3 VOL TS aJ
Ec = 25 V( LTS OCcn
r--r--1--+--+--r-~-+--+--r~--+--+--r-~-+--+-~~~4--+--+-~~--+--+--r-4---?0 3~
_w
LW(L

~~
200r-~-4--+-~.n~~'--+--+-,r-4--+--+-~-4--+--+--r-4--+--+-~~r-4--+--+-~~---20~~
I
E~~.J- 0:::
~:::E
\ - _ Ec = 30 'f:z:
~ ~,~,~~-iI.~_~t=_~_~-_-+--t~~~0~-4--+--+--~4--+--+-~~--~--+--~~-4--+--+--~4--- 10,g-

~ 150r--r~t__
r-~~~'-~~
__rP--4~
__~~__-+-__r-4--+__+-~~~4--+__+-~-4__~_-+__r-4--+__+-~~~ 08
I v
/ II

50
J v
I V v~
•I

• I
.
~ ........ I-- • 0
25

- -35

o 100 200 300 400 500 600 700

PLATE VOLTAGE IN VOLTS
FIG. B-7. 6F6 pentode.
Class A amplifier: Eb = 250 volts, Ec2 = 250 volts, E, = -16.5 volts, h = 34 rna, rp = 80,000 ohms approx,
gm. = 2,500 Mmhos. Interelectrode capacitances: C , = 6.5 MMf, C, = 13 MMf, Co = 0.2 MMf.
498 ELECTRON -TUBE CIRCUITS [APP. B

40

Ef • .3

0 ./
/
/
/
,. 1/
z 20 /
....z V
UJ
a:
a:
:::J
u n~ /
"
....
UJ
d/
::3 ~(;~.
a.l 0 V
L"
./
V
V
.... /'

5 10 15 20 25
O-C PLATE VOLTAGE IN VOLTS
FIG. B-8. 6H6 diode.
 14
~
~
$
12

If)

~ 10
UJ
0-
:::E
<!
-'
-'
'"
t"'
>
~
:::E 8 trJ
z
C1
I-
z ~
UJ
a:: >
a:: 6 ::0
:=>
u >
UJ C1
I-
<[
~
--' trJ
a... 4 .....
---
......
(f)
~
......
C1
2 (f)

a 100 200 300 400 500

PLATE VOLTAGE IN VOLTS
FIG. B-9. 6J5 triode.
Class A amplifier: Eb = 250 volts, E, = -8 volts, h = 9 rna, p. = 20, rp = 7,700 ohms, gm = 2,600 p.mhos.
Interelectrode capacitances: C , = 3.4 p.p.f, C 2 = 3.6 p.p.f, (;3 = 3.4 p.p.f.
 UI
o
o
400
Ef • 6 .J OL S

-1-+-1 Ec . 50 VO TS

f- f-I- +-!-j"-+-i
E'
;.,.c::.i;7 ,--:-i -
:;; 300
')
.-!.--
+-,~ l - f- trj
l'
.....
z
UJ
cr
cr
H I trI
(')
=>
U ~I - l - I-- ~
r.;.-}- ::0
0
Z,
i5
w
cr
J...- +- e- -, .~
u
(J) ~ 0" e
I j....:: t<: s to
.
cr
o t"l
10
w
S If 1'"
a..
100
-
\,V
I- .... :::--.
14
0
(')
.....
....,
"'"
(')
I, J...- I - ........ t--..... _t
, 1':::: - C
.....
I ~ V I-"
~
~"
- /.- - - - .- '.fJ

a 100 200 300 400 500 600 700

PLATE VOLTAGE IN VOLTS

FIG. B-I0. 6L6 beam tube.
Class A amplifier: E" = 350 volts, E" = 2.50 volts, E, -18 volts, h = 54 rna,
Ie' = 2.5 rna, rp' = 3:.1,000 ohn:s approx, g,n = 5,200 ,umhos.
Intcrclcctrocl. .., ('~p~citanc8s: C 1 = 10 ,u,uf, C, = 12 ,u,uf, C, = 0.·1 ,u,uf.
 2.0
Ef = .3 va 1S

<::>
'L
"
.,;/
1.5 A ~I I
VJ
LU 1 \, I 'I
a::
LU
CL
:::;;
/ \d, ,
<I:
--'
'" , ....:
--'
:; 1, , I'\v- 7' , I I
z
- 1.0
I / '\F/ , I "il I
I-
Z
~ 1/' ~c;, I"'~';:. • 1/ JI
,
LU
a::
a::
:::>
/ "- I,~ I ,1
u /.. ....... t I l"-J. " I ~I /"
1/ f. . . '\ i/,., Ik '~" V 'I ",1/
LU
I-
<I: 'j
--'
CL
0.5
... ..,
~~ I \ Pol " , II ';; \

t '"-....,-.r:;:.../ / "1-. O::S...... ~ 'f. '/ / , Y ..,

I[ ~f '\. V'\ ... V....... 'v· ,
If-- ~= L !'i -- -2
.......
-... V ...j '~ , lL V , V.0 V
V- t-,I( r--.... ~ tl- ~ -- VC Lx- V, L-
1/ V V ./
-
v. . . po< ~ ~ . . . .V V
I'- ...
j>< t:> ~ V
o 100 200 300 400 500 600
PLATE VOLTAGE IN VOLTS
FIG. B-11. 6SF5 triode.
Class A amplifier: Eb = 250 volts, Ec = -2 volts, h = 0.9 rna, IJ. = 100, Tp = 66,000 ohms, gm = VI
1,500 ILmhos. Interelectrode capacitances: C 1 = 4.0 IJ.lJ.f, C 2 = 3.6 IJ.lJ.f, C 3 = 2.4 ILILf. :=:
 AVERAGE PLATE CHARACTERISTICS
PENTODE CONNECTION
14

'"
UJ
ex;
u.J
a..
::> 12
<C
....
....
::>
~
10
OJ
0

~
:z 8
0 W
\ ex;
ex;
::::>
(J

;Z
UJ
w 6
ex;
(J

'"
ex;
a
:;; 4

t-
Z
UJ
ex;

'"
::::>
(J

UJ
t-
:s
Q.

a 100 200 300 400 500

PLATE VOLTAGE IN. VOLTS
FIG. B-12. 6SJ7 pentode.
Class A amplifier: Eb = 250, E" = 100 volts, Ee = -3 volts, h = 3 rna, Ie2 = 0.8
rna, rp = 1.0 megohm approx; gm = 1,650 .umhos. ~
Illterelectrode capacitances: C 1 = 6.0 .u.uf, C 2 = 7.0 .u.uf, C 3 = .005.u.uf. :t'
0:;
 ~
(/)
UJ
a::
UJ
a...
::e $
« 15
--'
--'
::e
z

0 '" "0
t"'
t-
:z:
lJ.J 10
>-
a:: >-3
a:: t;I:j
:::>
u {i
z
lJ.J :=
UJ
a:: >-
U
(/)
~
a:: >-
{i
0
>-3
t- t;I:j
:z: 5
UJ
c:::
c:::
:::>
....fJJ~
u
>-3
UJ
t-
....
{i
:5
a... fJJ

o 100 200 300 400 500

PLATE VOLTAGE IN VOLTS

FIG. B-13. 6SK7 variable-mu pentocle.
Class A amplifier: Eb = 250 volts, Ec2 = 100 volts, Ec = -3 volts, h = 9.2 rna, Ic2 = 2.6 rna, rp = 0.8
megohm approx, gm = 2,000 }Lmhos. Interelectrode capacitances: C 1 = 6.0 }L,u,f, C 2 = 7.0 }L}Lf,C s = .003 }L}Lf.
 EACH UNIT
14

12

~ 10
c::
.....
a..
tlj
,;;;: t"'
« tlj
-' (j
-' 8 >-3
:;;
~
:z 0
I- Z
I
:z
6 >-3
""
c::
c:: C
::::>
u 0:1
UJ
tlj
I-
;« (j
-'
a..
4 ....,
~
(j
C
....,
2 >-3
[J)

o 100 zoo ~OO 400 500

PLATE VOLTAGE IN VOLTS
~.""
FIG. B-14. 6SN7-GT dual-triode.
Class A amplifier: Eb = 250 volts, Ec = -8 volts, h = 9 rna, fJ. = 20, Tp = 7,700 ohms, g", = 2,600
ILmhos. Interelectrode capacitances: Unit 1: C , = 3.21LILf, C 2 = 3.41LILf, C3 = 41LfJ.f. Unit 2: C , = 3.8 f.J-p.f,
C 2 = 2.6 ILlLf, C 3 = 4 fJ.1Lf.
 EACH UNIT
7

en 5
w
0::
w 'i:i
a..
::::;; r
<[
>
....J >-3
....J 4 t':j
::::;;
z C":l
::r::
!z 3 >
w ~
0::
0::
:::>
>
C":l
u
>-3
w t':j
~ 2 ~
....J
a.. ....
UJ
>-3
....
C":l
1 UJ

o 50 100 150 200 250 300 450 500

350 400
PLATE VOLTAGE IN VOLTS
FIG. B-15. 6SL7-GT dual triode.
Class A amplifier: Eb = 250 volts, Ec = -2 volts, Ib = 2.3 ma, p. = 70, rp = 44,000 ohms, gm = 1,600 p.mhos. C.11
Interelectrode capacitances: Unit 1: C 1 = 3.0 p,p,f, C 2 = 3.8 p,p,f, C 3 = 2.8 p.p.f. Unit 2: C 1 = 3.4 p.p,f, a
c.n
C 2 = 3.2 p.p.f, C, = 2.8 p.p.f.
 (/.)
UJ
PENTODE CONNECTION a::
UJ
280 ,30 CL.
- :=IE
4:
0::-
20 UJ - '

t/l ~=
UJ 240 10 §E::=E
0:: . :z:
UJ 0-
0..
:::>
<[
O~'Z
LLJ
-' a::
-' a::
::z 200 :::>
<.J
:z

N
(.)
160
t-
:z
UJ
a:
a:
=>
u 120
:z
UJ
UJ
a:
U
t/l
a: 80
0
h-
:z
UJ
a:
a:
=> 40
u
UJ
t-
«
-'
0..

o 100 200 ,300 400 500

PLATE VOLTAGE IN VOLTS
FIG. B-16. 6V6 beam tube.
Class A amplifier: Eb = 250 volts, E e, = 250 volts, Ee = -12.5 volts, h = 45 rna, Ie' = 4.5 rna, Tp = 52,000
ohms approx, gm = 4,100 pmhos. Interelectrode capacitances: C 1 = 10ppf, C 2 = 11 ppf, C 3 = 0.3 ppf.
 An BJ PLATE CHARACTERISTICS 507
200
f 6 3 OL S
/ "
/
/
I
150

'"
W
0:
W
/

a.
~
........
::>
::: 100
I-
z
W
0:
0:
=>
U
W
!;(
.... "/
.... V L
y.'>
a. 50

V
/'
V
..,... V
o 10 20 30 40 50
O-C PLAT€ VOLTAGE IN VOLTS
FIG. B-17. 6X5 triode.
 APPENDIX C

CHARACTERISTICS OF TRANSMITTING TUBES

0
0
0

'"
0
If)

.........

0
,0
....
If)

(/)
I-
0 -' cD
U If) 0'"0
> .9
0 ....
'" ......,
...
II) -z: 0
w 0
<5 <..? 00
«
>
II')
0
0
l-
-' .
....
0 0 ....

"€
II
>Q
w
l- d
H
S
a. ~

0
.....
If)

o
o
If)

o
If)

'"

o
o o o o o o o
...o If) If) o
....
If)

'" N'"I lN3~~nJ

""S3~f3dVj\1111IVj
o
.... If)

31\1ld
508
APP. C] CHARAC'>tERISTICS OF TRANSMITTING TUBES 509

~~~~~~
o'~~~~~~~e~oo~~'~2o~o~~lro~o~.~oooo~~~~~O~o~-2~eoo~~~~~-L-~~eoo~~~~
PLATE VOLTAGE IN VOLlS.

FIG. 0-2. 851 triode.

 APPENDIX D

TABLE OF BESSEL FUNCTIONS OF THE FIRST KIND

511
 J.(1) J.(2) In(3) I n(4) In(5) J.(6) In(7) J,,(8) In(9) In(lO) J.(ll) In(12) In(l3) In(14) In(15) In(16) In(17) In(18) I n(19) In(20)
n

- --- --- - - - - --- - - - - --- --- ~-- --- --- --- --- --- --- --- --- --- - -
0 +0.7652 +0.2239 -0.2601 -0.3971 -0.1776 +0.1506 +0.3001 +0.I717 -00903 -0.2459 -0.1712 +0.0477 +0.2069 +0.171I -0.0142 -0.1749 -0.1699 -0.0134 +0.1466 +0.1670
1 +0.4400 +0.5767 +0.3391 -0.0660 -0.3276 -0.2767 -0.0047 +0.2346 +0.2453 +00435 -0.1768 -0.2234 -0.0703 +01334 +02051 +00904 -0.0977 -0.1880 -0.1057 +0.0668
2 +0.1I49 +0.3528 +0.4861 +0.3641 +0.0466 -0.2429 -0.3014 -0 1I30 +0.1448 +0.2546 +0.1390 -0.0849 -0.2177 -0.1520 +00416 +0.1802 +0 1584 -0.0075 -0.1578 -0.1603
3 +0.0025 +0 1289 +0.3091 +0.4302 +03648 +0. I!48 -0.1676 -0.291I -0.1809 +00584 +0.2273 +0.1951 +0.0033 -0.1768 -0.1940 -0.0439 +01349 +0.1863 +0.0725 -0.0989
4 +0.0002 +0.0340 +0.1320 +0.28I! +0.3912 +0.3576 +0.1578 -O.ID54 -0.2655 -0.2196 -0.0150 +0.1825 +02193 +0.0762 -0.1I92 -0.2026 -0.1I07 +00696 +0.1806 +0.1307
5 +0.0070 +0.0430 +0.1321 +0.261I +0.3621 +0.3479 +01858 -0.0550 -0.2341 -0.2383 -0.0735 +0.1316 +0.2204 +0.1305 -0.0575 -0.1870 -0.1554 +00036 +0.1512
6 +0.0012 +0.0I!4 +0.0491 +0.13I! +0.2458 +0.3392 +0.3376 +0.2043 -0.0145 -0.2016 -0.2437 -0.1I80 +00812 +0.2061 +0.1667 +00007 -0.1560 -0.1788 -0.0550
7 0.0002 +0.0025 +0.0152 +0.0534 +0.1296 +0.2336 +0.3206 +0.3275 +0.2167 +0.0184 -0.1703 -0.2406 -0.1508 +0.0345 +0.1825 +0.1875 +0.0514 -0. II 65 -0.1842
8 +0.0005 +0.0040 +0.0184 +0.0565 +0.1280 +02235 +0.3051 +0.3179 +0.2250 +0.0451 -O.l4ID -0.2320 -0.1740 -0.0070 +0.1537 +0.1959 +0.0929 -0.0739
9 +0.0004 +0.0055 +0.0212 +0.0589 +0.1263 +0.2149 +0.2919 +0.3089 +0.2304 +0.0670 -0.II43 -0.2200 -0.1895 -0.0429 +0.1228 +0.1947 +0.1251
10 +0.0002 +0.0015 +0.0070 +0.0235 +0.0608 +0.1247 +02075 +0.2804 +0.3005 +0.2338 +0.0850 -0.0901 -0.2062 -0.1991 -0.0732 +0.0916 +0.1865
II +0.0004 +0.0020 +0.0083 +0.0256 +0.0622 +01231 +0.2010 +0.2704 +0.2927 +0.2357 +0.0999 -0.0682 -0.1914 -0.2041 -0.0984 +0.0614
12 +0.0005 +0.0027 +0.0096 +0.0274 +0.0634 +0.1216 +0.1953 +0.2615 +0.2855 +0.2367 +0.1I24 -0.0486 -0.1762 -0.20M -0.1I90
13 +0.0001 +0.0008 +0.0033 +00108 +0.0290 +0.0643 +0.1201 +0.1901 +02536 +0.2787 +0.2368 +0.1228 -0.0309 -0.1612 -0.2041
14 +0.0002 +OOOID +0.0039 +0.0120 +0.0304 +0.0650 +0.1I88 +0.1855 +0.2464 +0.2724 +0.2364 +0.1316 -0.0151 -0.1464
15 +0.0003 +00013 +0.0045 +0.0130 +0.0316 +0.0656 +0. II 74 +0.1813 +0.2399 +0.2666 +0.2356 +0.1389 -0.0008
16 +0.0004 +0.0016 +0.0051 +0.0140 +00327 +0.0661 +0.I!62 +0.1775 +0.2340 +0.261I +0.2345 +0.1452
17 +0.0001 +0.0005 +0.0019 +0.0057 +0.0149 +0.0337 +0.0665 +0.1I50 +0.1739 +0.2286 +0.2559 +0.2331
18 0.0002 +0.0006 +0.0021 +0.0063 +0.0158 +0.0346 +00669 +0.1I38 +0.1706 +0.2235 +0.25II
19 +0.0002 +0.0008 +0.0025 +0.0068 +0.0166 +0.0354 +0.0671 +0. I!27 +0.1676 +0.2189
20 +0.0003 +0.0009 +0.0028 +0.0074 +0.0173 +00362 +0.0673 +0.11I6 +0.1647
21 +0.0003 +0.0010 +0.0030 +0.0079 +0.0180 +00369 +0.0674 +0.1106
22 I +0.0001 +0.0004 +0.0012 +0.0034 +00084 +0.0187 +0.0375 +0.0676
I
23 I +0.0001 +0.0004 +0.0013 +0.0036 +00089 +0.0193 +0.0381
24
I I +0.0005 +00015 +0.0039 +0.0094 +0.0199
I i
I I I I I
-
 INDEX
A Amplifier, class A, equivalent circuit for,
43ff·
A supply, 64 (See also Equivalent circuit, linear
Abnormal glow discharge, 28 class A)
Ammeters, electronic, 480 grid-bias voltage in, 39
electrometer tubes, 481 (See also Self-bias in amplifiers)
Amplification, voltage (see Amplifier, grid signal voltage in, 39
voltage-gain) inductance-capacitance coupled (see
Amplification factor, measurement of, 45 Inductance-capacitance coupled
numerical values for several tubes, amplifier)
492ff· input admittance of, 60ff.
tetrode, 23 inverse-feedback (see Feedback)
triode, 17 load lino for, 41
Amplifier, audio-frequency, anode-fol- load resistance for maximum powcr
lower, 108ff. output from, 168
comparison with cathode follower, maximum power output from, 168,
111 173
gain, 109 output transformer in, 170
input admittance, 110 parallel operation in, 178
output impedance, 109 pentode in, 176
broad-band, anode-follower (see Ampli- phase relations in, 42
fier, anode-follower) plate dissipation, 174
cathode-follower (see Amplifier, cath- plate efficiency of, 174
ode-follower) power output from, 168
compensated, high-frequency, 95ff. push-pull (see Push-pull amplifier)
low-frequency, 101ff. quiescent operating point for, 41
direct-coupled (see Amplifier, direct- radio-frequency (see Tuned ampli-
coupled) fiers)
cascode, 314 resistance-capacitance coupled (see
cathode-coupled, i13ff. Resistance-capacitance coupled
cathode-follower, 102ff. amplifier)
comparison with anode follower, 111 self-bias in (see Self-bias in ampli-
double, 120 fiers)
gain, 104 single-stage, 39ff.
input admittance, 107 tetrode in, 21ff.
output impedance, 107 transformer-coupled (see Trans-
class A, beam power tubes in, 176 former-coupled amplifiers)
cascade, 68ff. tuned (see Tuned amplifiers)
voltage gain of, 70 voltage gain of (see Amplifier, volt-
definition of, 53 age-gain)
degenerative (see Feedback) class AB, definition, 53
distortion in, 169 push-pull (see Push-pull amplifier)
dynamic characteristic of, 42 class B, definition, 53
513
514 ELECTRON -TUBE CIRCUITS

Amplifier, class B, push-pull (see Push- Amplifier, inverse-feedback (see Feed-

pull amplifier) back)
tuned (see Tuned amplifiers, class inverted, 122
B power) limiters (see Limiter, amplitude)
class C, definition, 53 linear, class B (see Tuned amplifiers,
tuned (see Tuned amplifiers, class class B power)
C power) conditions for, 43, 216
compensated (see Amplifier, broad- load line for, 41
band) maximal flatness in (see Tuned ampli-
computing (see Electronic computer fiers, class A, double-tuned)
circuits) modulated (see Modulator, amplitude,
degeneration in (see Feedback) amplifier as)
difference (see Amplifier, cathode- for modulated wave (see Tuned ampli-
coupled; Amplifier, direct-coupled) fier, class B power)
direct-coupled, 111ff. narrow-band (see Tuned amplifiers)
battery-coupled, 111 oscillations in, 89
cathode-coupled, 113ff. output impedance, anode follower, 109
resistance-coupled, 112 cathode follower, 107
series balanced, 121 effect of feedback on, 88
double-tuned (see Tuned amplifiers) output transformer in, 170
driver stage for (see Push-pull ampli- output wave form, 42
fier) parallel or shunt-feed, 76
dynamic characteristic of, construc- path of operation in, 41, 217, 223
tion from static, 42 phase characteristic of, 74
equations of, 48 phase-inverter, 187
push-pull (see Push-pull amplifier, phase relations in, 42
class A, composite dynamic char- plate-circuit efficiency (see Plate-cir-
acteristic of) cuit efficiency)
in feed-back oscillator, 258 plate-current components, 42, 49
feedback in (see Feedback) plate dissipation (see Plate dissipation)
frBquency characteristics of, 74 plate-modulated (see Modulator, am-
frequency classification, 54 plitude, amplifier as)
frequency distortion (see Frequency power, 167ff.
distortion) (See also Push-pull amplifier; Tuned
frequency response of (see Frequency amplifier, class B power; Tuned
distortion) amplifier, class C powcr)
extension of (see Amplifier, broad- power supply in, common, 64
band) push-pull (see Push-pull amplifier)
gain (see Amplifier, voltage-gain) quiescent operating point for, 41
graphical analysis of output, 41 radio frequency (Tuned amplifiers)
grid-base voltage of, 40 regenerative (see Feedback)
(See o1so Self-bias) resistance-capacitance coupled (see Re-
grid signal voltage in, 40 sistance-Capacitance coupled am-
grounded-grid, 122 plifier)
harmonic distortion in (see Nonlinear resistance coupled (see Amplifier, di-
distortion) o reet-coupled).
interelectrode capacitances in, 58 shunt feed in, 170
intermediate-frequency (see Tuned am- stagger-tuned (see Tuned amplifiers;
plifiers, class A) class A, stagger-tuned)
internal impedance of, effect of feed- transformer-coupled (see Transformer-
back on, 86ff. coupled amplifiers)
 INDEX 515
Amplifier, tuned (see Tuned amplifiers) Beam power tubes, 25
video (see Amplifier, broad-band) Beam resistance of tube, 133
voltage-gain, db gain, 56 Beat-frequency oscillator, 357
effect of feedback on, 83 Beating (see Converter)
inductance-coupled, 76 Bel, 55
interelectrode capacitances, effect of, Bessel functions, 367, 511
59 Bias, grid, 53
multistage amplifier, 70 impedance, effect of, in amplifiers, 85
resistance-capacitance coupling, 71 Bias control of thyratrons, 286
resistance coupling, 113 Bias phase control of thyratrons, 287
resistance load, 56 Bleeder resistance, 303
transformer coupling, 76 Blocking, oscillator (see Oscillators, relax-
tuned, double-tuned, 200 ation, blocking)
single-tuned, direct coupling, 194 in oscillators, 250
transformer coupling, 197 Blocking capacitor, 69
Amplitude distortion (see Nonlinear dis- Bombardment by positive ions, 7, 28
tortion) Breakdown, gas tubes, 27
Amplitude modulation (see Modulation, mercury-arc tubes, 28
amplitude) Breakdown potential, 27
Angstrom unit, 5 Bridge measurements of triode coeffi-
Anode fall, 28 cients, 45
Anode follower (see Amplifier, anode-fol- Bridge rectifier, 279
lower) By-pass capacitor, 64
Are, anode fall, 28
cathode fall, 27 C
cathode spot, 27
externally heated, 30 C supply, 64
high-pressure, 31 Capacitance, interelectrode, numerical
initiation, grid control (see Thyratrons) values for several tubes, 492ff.
ignitor rod (see Ignitron rectifier)
pentodes, 63
mercury pool (see Mercury-arc recti-
tetrodes, 62
fier)
triodes, 58
Arc back in mercury rectifiers, 36
Capacitor filters (see Filter, rectifier,
Argon, in high-pressure diodes, 31
capacitor)
in photo tubes, 7
in thyratrons, 33 Carrier suppression, 325
Automatic amplitude control, oscillators Carrier wave, disappearance of, in fre-
quency modulation, 369
with, 258
Automatic volume control, 355 suppression of, 325
Cathode, directly heated, 4
B indirectly heated, 4
inward-radiating, 4
B supply, 64 mercury as, 28
Backfire, 36 oxide-coated, disintegration, 31
Balanced modulators (see Modulator, emission, 3
amplitude, balanced) work function, 3
Band width, tuned amplifiers, 194, 195, photoelectric, 4
197 power for heating, 64
reduction in cascade stages, 205, 206 thermionic (see Thermionic cathodes;
untuned amplifiers, 74, 95 Thermionic emission)
Barkhausen criterion for oscillators, 247 thoriated-tungsten, 3
516 ELECTRON-TUBE CIRCUITS

Catbode, tungsten, 3 ClaBB A amplifiers (see Amplifier, class A)

Cathode efficiency, 3 Class B, C, amplifiers (see Push-pull
Cathode fall, arc, 28 amplifier; Tuned amplifiers, class
glow discharge, 28 B power; Tuned amplifiers, class C
Cathode follower (see Amplifier, cathode- power)
follower) Class C modulated amplifier (see Modu-
Cathode modulation (see Modulation, lator, amplitude, amplifier as)
cathode) Clipping, circuit (see Limiter)
Cathode-ray tube, electrostatic deflection side-band, in modulators, 323
in, 442 Cloud, electron (see Space charge)
magnetic deflection in, 460 Coefficient, coupling, 201
sweep circuits for (see Sweep generators, Coil, Q of, 194
for cathode-ray tubes) Cold-cathode triodes, 30
tracing speed of beam in, 442 Collisions, in arc tubes, 28
Characteristic tube curves, FG-27 (thyra- ionizing, 7
tron), 33 by positive ions, 7
FG-33 (thyratron), 33 Colpitts oscillator, 253
FG 98 (shield-grid thyratron), 34 Compensated amplifiers (see Amplifiers,
PJ 22 (vacuum photocell), 5 broad-band)
PJ 23 (gas photocell), 6 Composite characteristics, push-pull, dy-
RCA OA4G (cold-cathode triode), 30 namic, 184
2A3 (power triode), 492 static, 181
5U4 (double diode), 491 Computer circuits (see Elcctronie com-
6AC7 (television pentode), 493 puter circuits)
6AG7 (power pentode), 494 Condensed mercury temperature and
6C5 (voltage triode), 495 vapor pressure, 30
6F6 (power tube), 496, 497 Conductance, mutual, 17
6H6 (duo diode), 498 Conduction, gaseous, 27fJ.
6J5 (voltage triode), 499 (See also Arc)
6L6 (beam power tetrode), 500 Constant-current characteristic, 20, 225
6SF5 (voltage triode), 501 Constants, multielectrode tube, 23
6SJ7 (voltage pentode), 502 triodes, 18
6SK7 (voltage supercontrol pentode), Controlled rectifiers, applications of,
503 281
6SL7 (voltage duo-triode), 505 filter for, 304
6SN7 (voltage duo-triode), 504 ignitron, 288
6V6 (beam power tetrode), 506 thyratrons, average current, 283
6X5 (duo-diode), 507 bias control, 286
800 (transmitting triode), 508 bias phase control, 287
851 (transmitting triode), 509 on-off control, 287
866 (mercury diode), 36 phase-shift control, 285
884, 885 (thyratron), 33 circuits, 286
889A (transmitting triode), 19, 20 circle diagram of, 286
Child's law, 11 voltage across tube, 284
(See also Space-charge limited cur- Conversion, theory, general, 259
rent) transconductance, 358
Choke input filters (see Filter, rectifier) square-law, 358
Clamping circuit, continuously acting, C'.-onverter, pentagrid, 360
133 Counter circuits (see Trigger circuits)
switched, double-sided, 139 Coupling, amplifier, impedance, 75
single-sided, 139 resistance, 68
 INDEX 517
Coupling, amplifier, resistance-capaci- Demodulator, definition of, 343
tance, 112 (See also Demodulation; Detection;
transformer, 76 Detector; Modulation)
coefficient, critical, 201 Detection, for amplitude-modulated
transitional, 203 waves, 343ff.
networks, design of, 214 conversion, 357
Critical grid control curves, 33 definition of, 343
Critical inductance, filters, 302 diode (see Detection, linear)
Crystal oscillators (see Oscillators, crystal) for frequency-modulated waves, 387
Current density, 11 linear, 345
Current feedback (see Feedback, current) a-c to d-c impedance in, 353
Current-source equivalent circuit, 47 distortion in, 351
Current-voltage locus, push-pull ampli- effect of load on, 353
fier, 181 input impedance of, 350
resistance load, 41 rectification characteristic, 353
Cutin, capacitor input filter, 297 single-side-band, 357
inductance filter, 294 square-law, 343
thyratron, 281 distortion in, 344
Cutout, capacitor input filters, 297 with triode, 343
controlled rectifiers, 282 suppressed-carrier, 357
inductance filter, 294 Detector, diode, 345
L-section filters, 302 first (see Converter)
Cylinders, space charge limited current linear, 345
(see Space-charge limited current) square-law, 343
(See also Demodulation; Detection)
D Deviation, frequency, in frequency modu-
lation, 366
Deviation ratio in frequency modulation,
D-c amplifier (see Amplifier, direct-
366
coupled)
Difference amplifier, 113ff, 146ff
D-c restorer, 134
cascode type, 121, 147
(See also Clamping circuit)
cathode-coupled, 113, 146
Decibel gain, 55 Differentiation, circuits, 128ff.
Decoupling filters in amplifiers, 65 amplifiers, 161
Deflection in cathode-ray tubes (see Diode, characteristics, operating, 12, 35
Sweep generators) clipper, 123ff.
Degenerative feedback (see Feedback, cylindrical, 12
negative) detector, 345
Deionization, 36 discriminator, 387
Delay, avc systems, 356 high-pressure gas, 31
generation of time, 424ff. high-vacuum, as clamper, 133ff.
linear delay multivibrator, 425 hot-cathode gas, 30
linear-sweep delay, 432 phasemeter, 483
Miller integrating methods, 427 plane parallel, 9
phantastron, 431 ratings, 12
sanatron, 430 as rectifiers (see Rectifiers)
(See also Gate circuits) square-law modulation with, 323
Delay distortion, 54 static characteristics, 11
Demodulation, 343ff. voltmeter, 476
definition of, 343 Discharge, electrical, in gases (see Elec-
(See also Detection) trical discharge in gases)
518 ELECTRON -TUBE CIRCUITS
Discriminator for frequency modulation, Electrical discharge in gases, glow (see
38'7 Glow discharge)
Disintegration of cathodes, 31 Electrometer (see Ammeters, electronic)
Disintegration voltage, 31 tubes for, 481
Dissipation (see Plate dissipation) Electron, secondary (see Secondary emis-
Distortion, amplitude (see Nonlinear sion)
distortion) Electron-coupled oscillator (see Oscilla-
in class A amplifiers, 54, 169 tors, electron-coupled)
causes of, 54 Electron-current density, drift, 11
permissible maximum, 171 Electron emission, bombardment, 7
for push-pull operation, 185 photoelectric (see Photoelectric emis-
delay, 54 sion)
feed-back effect on, 84 secondary (see Secondary emission)
frequency (see Frequency distortion) thermionic (see Thermionic emission)
harmonic (see Nonlinear distortion) Electron tubes (see Diode; Pentode;
in linear detectors, 354 Tetrocle; Thyratrons; Triode)
in modulation, 337 classification of, 9ff.
nonlinear (see Nonlinear distortion) Electronic computer circuits, 146ff.
phase, 54 adding, 148
in square-law detection, 345 difference, 146
Distributed capacitance, 79 difference of squares, 153
Doubler, frequency, 243 differentiating, 129, 161
voltage, 280 dividing, 154
Driver stage (see Push-pull amplifier, integrating, 129, 158
driver stages for) multiplying, 152
Dushman's equation, 2 square root, 154.
Dynamic characteristics, class A, 48 squaring, 150
construction of, 43 Electronic instruments, ammeter (see
general power series representation, 48 Ammeters, electronic)
nonlinear distortion, 48, 54 frequency (see Frequency meter)
parabolic, 48 phase (see Phasemeter)
push-pull, 184 power (see Wattmeter)
rectifier, 272 voltmeter (see Vacuum-tube voltmeter)
resistance load, 43 Electrostatic deflection in cathode-ray
Dynatron, 262 tubes (see Cathode-ray tube)
oscillator (see Oscillators, negative- Emission, electron (see Electron emission)
resistance) Equivalent circuit, linear class A, 43ff.
analytical derivation of, 43
E applicability, 44
current source, 47
Eccles-Jordan trigger circuit (see Trigger interelectrode capacitance, effect of,
circuits, Eccles-.Tordan) 58
Efficiency, cathode, 3 pentocle,63
plate-circuit (see Plate-circuit effi- push-pull, 180
ciency) tetrode,62
rectification, 274 voltage source, 43
tuned circuit transfer, 216 Excitation voltage, in amplifiers, a-c d-c
Electric arc (see Arc) components, 40
Electrical discharge in gases, 27ff. multistage, 70
arc (see Arc) push-pull, 178
breakdown, 27 thyratrons, 282
 INDEX 519
F Frequencies, half-power, 72, 74, 195
mid-frequency range, 71
Feedback, in amplifiers, 81:0'. side band, in amplitude-modulated
effect of, on distortion, 84 wave, 322
on output impedance, 88 in frequency-modulated wave, 366
gain, 83 video, 95
negative, 81 Frequency, angular, instantaneous, 364
positive, 81 carrier, 321
circuits, 85ff., 155 division, blocking oscillator, 412
compound, 87 multivibrator, 409
current, 85 scaling circuits, 420
and stability in amplifiers, 89 mid-band, 71
criterion for oscillation, 90 modula tion (see Modulation, fre-
(See also Oscillators, feed-back) quency)
voltage, 86 thresliold, 5
Feed-back factor, 82 Frequency characteristic, 74, 80
Feed-back oscillators (see Oscillators, with inverse feedback, 81
feed-back) Frequency conversion, 358
Feed-back ratio, complex, 246 Frequency converter tubes, 360
Field emission, 7 Frequency distortion, description, 54
Filament (see Thermionic cathodes) inductance-capacitance coupled ampli-
Filament voltage, 64 fier, 76
Filter, capacitor input, 296
resistance-capacitance coupled, 74
cutin in, 297 transformer-coupled, 80
cutout in, 297
tuned power, 213ff.
effect on inverse peak voltage, 300
tuned voltage, 192ff.
peak tube current with, 297
video, 98
ripple, 300
Frequency doubler, 243
choke input, 293
Frequency measurement, with oscillo-
cutout in, 294
scope, 486
ripple, 295
frequency meter (see Frequency meter,
controlled rectifier, 304
electronic)
decoupling, in amplifiers, 65
L-section, 300 Frequency meter, electronic, 487
bleeder resistance, 303 Frequency-modulated oscillator (see Os-
critical inductance, 302 cillators, frequency-modulated)
cutout, 302 Frequency modulation (see Modulation,
ripple, 301 frequency)
swinging choke, 303 Frequency modulation detection (see
multiple L-section, 304 Detection, for frequency-modulated
pi-section, 305 waves)
Flash-back voltage, 36 Frequency modulation receivers (see
Focusing electrodes in beam tubes, 26 Receiver, frequency modulation)
Forward voltage rating, rectifiers, 36 Frequency response, 74
Fourier series, amplifier output, push- extension of, in amplifiers, 95ff.
pull, 179 Frequency stability of oscillators (see
single tube, 49 Oscillators, stabilization)
rectifier output, single-phase, full- Frequency translation (see Detection;
wave, 293 Modulation)
half-wave, 293 Full-wave rectification (see Rectifiers)
520 ELECTRON-TUBE CIRCUITS
G Grid voltage, cutoff, 53, 2Hi
Grounded-grid amplifier (see Amplifier,
Gain, voltage (see Amplifier, voltage- grounded-grid)
gain)
Gain-band width product, 206 H
Gas amplification, mechanism, 7
phototube, 6
Harmonic distortion (see Nonlinear dis-
Gas diodes, 30
tortion)
Gas phototubes, 6
Harmonic generation, in linear detector,
Gaseous conduction (see Electrical dis-
355
charge in gases)
Gate circuits, a-c coupled, 414 in modulation, 327
cathode-coupled, 416 multivibrator, 395
d-c coupled, 415 in square-law detection, 345
pentode, 418 Harmonics (see Fourier series; Nonlinear
(See also Delay, generation of time) distortion)
Glow discharge, 27 Hartley oscillator (see Oscillators, feed-
Glow tubes, diodes, 29 back)
triodes, 30 Heat-shielded cathodes, 4
Graphical determination of Fourier com- Heptode,25
ponents,49 Hexode, 25
Grid, control, definition of, 14 High field emission, 7
cold-cathode triode, 30 High-level modulation (see Modulation,
grid control curves, 16 high-level)
in high-vacuum tubes, control of cur- High-pressure arcs, 31
rent, 14
heating of, 15, 218 I
nonuniform, 25
screen, 18, Ignitor rod, 34
suppressor, 24 Ignitron rectifier, 35
in thyratron, 32 Image force, 2
Grid-bias modulation (see Modulation,
Impedance matching, 168
grid-bias)
by transformer, 170
Grid-bias voltage, 64
Impedance-stabilized oscillator (see Os-
self-bias, 64
cillators, stabilization of, impedance)
Grid control curves, thyratron, 33
Inductance, for filters, 293
Grid current, causes, of, 15, 218
critical in L-section filters, 302
class B amplifier, 218
leakage, 79
class C amplifier, 226
electrometer, effects in, 480 primary of transformer, 77
oscillators, 249 swinging choke, 303
thyratrons, 34 Inductance-capacitance coupled ampli-
Grid driving power, class C amplifiers, fier, 75
230 Input admittance, pentode, 63
Grid glow tube, 30 tetrode, 62
Grid leak and capacitor bias voltage, 249 triodes, 60
build up of, 249 Input capacitance of amplifiers, 70
effect on oscillator stability of, 250 effect on operation, 73
Grid-plate transconductance, 17 Input resistance of triodes, 61
Grid voltage, bias, 64 negative, conditions for, 61
(See also Self-bias) Input volts (see Excitation voltage)
 INDEX 521
Instantaneous amplitude, 320 Leakage reactance (see Reactance, trans-
Instantaneous angular frequency, 320, former, leakage)
364 Limiter, amplitude, 123ff.
Instantaneous phase, 320, 364 diode, 123
Insulation stress in transformers, 274, in frequency modulation receivers, 286
277,280 grid, 126
Integrating amplifier, 158, 427 plate, 127
Integrating circuits, 128 Line-controlled pulse generator (see Pulse
Interelectrode capacitance, 58ff. generation, line-controlled)
in the cathode follower, 103 Linear amplifier, class A, 48
effect of, on input admittance, 60 class B, 216
equivalent circuit including, 59 Linear delay circuits (see Delay, gener-
numerical values for several tubes, ation of time)
492ff· Linear detector (see Detector, linear)
pentode,63 Lissajous pattern, for measurement of
reduction of, by screen grid, 21 frequency, 486
tetrode, 62 for measurement of phase, 482
Intermediate frequency, 358 Load line, nonlinear in push-pull ampli-
Inverse feedback (see Feedback, in fier, 185
amplifiers, negative) pentode amplifier circuit, 176
Inverse peak voltage rating, bridge cir- rectifier, single-phase, half-wave, 272
cuit, 280 triode amplifier circuit, 41
single-phase, full-wave circuit, 277 Load resistance, choice in pentodes, 176
single-phase, half-wave circuit, 274 class A amplifier, 41
with capacitor input filter, 300 maximum power output, 168, 173
voltage doubler, 280 matching, 170
Inverted amplifier, 122 push-pull amplifier, 181
Inward radiating cathodes, 4 Local oscillators (see Converter; Mixer)
Ionization, description, 7 Low-level modulation (see Modulation,
by electron bombardment, 7 low-level)
in gas photo tubes, 6
photoelectric, 7 M
in plasmas, 28
by positive ions, 7, 28 Magnetic deflection in cathode-ray tubes,
Ionization time, 36 460
Ions, 7 Master oscillator (see Oscillators, master)
function in gaseous conduction, 27 Maximal flatness, 98
in glow discharge cathode fall space, 28 Maximum power output, class A ampli-
in plasma, 28 fier, 168, 172
push-pull connection, 186
K from linear modulator, 330
undistorted, 172
Kinetic energy, electrons and electron Mercury, vapor pressure of, 30
velocity, 11
Mercury-arc rectifier, arc back, 36
Konal metal, 3
arc drop, 27
L cathode disintegration, 31
evaporation of mercury, 30
L-section filter, 300 flash-back voltage, as function of tem-
Langmuir-Child's law, 11 perature, 36
(See also Space-charge limited cur- ratings of, average current, 35
rent) forward voltage, 35
522 ELE CTRON -TUBE CIRCUITS
Mercury-arc rectifier, ratings of, inverse Modulation, percentage, 321
voltage, 35 phase, 371ff.
surge current, 36 comparison with frequency modula-
vapor pressure in, 30 tion, 372
(See also Rectifiers; Thyratrons; definition of, 371
Controlled rectifiers) expression for, 372
Metals, image force, 2 plate, 327
secondary emission, 6 advantages of, 332
thermionic emission, 2 process of, 320
work function, 3 side bands in, 322
Miller integrating circuit, 427 single side band, 357
Millman network theorem, 489 sinor representation of, 322
Misch metal, 29 square-law, 323
Mixer, 357 suppressor grid, 337
Mixing (see Conversion) transformer for, 331
Modulated wave, components of, ampli- types of, 321
tude-modulated, 322 van der Bijl, 323
frequency-modulated, 366 Modulation characteristics, 328
envelope of, 323 Modulation factor, 321
Modulation frequency, 321
Modulation, amplitude, 320ff.
definition of, 320 Modulator, amplitude, 320ff.
amplifier as, 327
expression for, 320
balanced, 325
frequency spectrum for, 322
van der Bijl, 323
linear, 327
frequency, 363ff.
modulation characteristic for, 327
Armstrong method, 381
side bands in, 322
phasitron, 384
sinor representation of, 322
reactance tube, 373
square-law, 323
phase, 371
angular, 364 single side band, 357
instantaneous frequency, 364 suppressed carrier, 326, 357
carrier suppression of, 325 (See also Modulation)
cathode, 337 Multielectrode tubes, 18ff.
defined, 320 (See also Beam power tubes; Pen-
frequency, 363ff. tode; Tetrode; Thyratrons, shield-
characteristics of, 364 grid)
comparison with phase modulation, Multiple L-section filters, 304
372 Multiplication in high-vacuum tubes, 152
definition of, 363 frequency, 243
deviation ratio for, 366 voltage, 280
expression for, 366 Multiplier, circuits for frequency, 243
frequency deviation in, 366 voltage, 280
narrow band, 381 Multistage amplifiers, 68ff., 204
spectrum of, 366 Multivibrators (see Oscillators, relaxa-
wide band, 373 tion)
frequency translation by, 320 one-shot (see Gate circuits)
grid-bias, 332 Mutual characteristics, 17
advantages of, 336 Mutual conductance, 17
high-level, 327 measurement of, 46
index, 321 numerical values of several tubes,
low-level, 325 492ff:
 INDEX 523
N Oscillators, dynatron, 262
electron-coupled, 254
Negative feedback (see Feedback, in feed-back, 244J!.
amplifiers, negative) self-excitation in, 244
Negative input resistance of amplifiers, Franklin, 267
61, 262 frequency-modulated, 373J!.
Negative plate resistance in tetrodes, 22, frequency stability of, 254
263 grid bias in, 249
Negativc-resistance oscillator (see Oscilla- Hartley, 253
tors, negative-resistance) intermittent operation of, 250
Negative-transconductance oscillator (see local (see Converter, Mixer)
Oscillators, transitron) master, 254
Network theorem, Millman, 489 multivibrator, 396
Neutralization, electron space charge, 29 negative-resistance, 262
in amplifiers, 239 negative-transconductance, 263
Nonlinear circuit element as modulator, nonlinearity and stabilization, 247
343 phase-shift, 262
as rectifier, 271jJ. power, 251
Nonlinear distortion, description, 48J!., relaxation, 395J!.
54 blocking, 404
distortion factor, 50 synchronization of, 411
effect of feedback on, 84 gas diode, 443
five-point schedule, 50 gas triode, 444
general dynamic curve, 48 IDultivibrator, 296
grid current, cause of, 126 biased plate coupled, 400
parabolic dynamic curve, 50 cathode-coupled, 402
push-pull amplifiers, 185 pentodes in, 403
Nonlinearity and oscillator stability, 246 period of, 399
Normal current density, 28 plate-coupled, 396
Normal glow discharge, 28 pentode, van der Pol, 406
synchronization of, sine wave, 409
o pulse, 408
resistance-capacitance tuned, 259
On-off control of thyratrons, 287 resistance-stabilized, 255
Operating point, 41 self-starting of, 248
Oscillation, in amplifiers, 82 square-wave, 396
Barkhausen criterion for, 246 stabilization of, amplitude, 254
Nyquist criterion for, 90 impedance, 255
Oscillators, 244J!. resistance, 255
amplitude stability of, 258 transitron, 264
automatic amplitude control of, 258, tuned-grid, 253
261 tuned-grid-tuned-plate, 253
basic circuits, 253 tuned-plate, 250
class A, 258 frequency of oscillation in, 252
classification of, 244 self-excitation in, 251
Colpitts, 253 sinor diagram for, 252
crystal-controlled, 256 Wien bridge, 261
circuits for, 257, 269 Oscilloscope as frequency meter, 486
equivalent circuit of, 257 as phase meter, 481
stability of, 258 as voltmeter, 473
524 ELECTRON-TUBE CIRCUITS

Output impedance, 88 Piezoelectric crystals in oscillators, 256

anode follower, 109 Plasma, description of, 28
cathode follower, 107 Plate characteristics, 492ff.
effect of feedback on, 89 (See also Characteristic tube curves)
Output voltage of amplifier (see Ampli- Plate-circuit efficiency, amplifier, class A,
fier, voltage-gain) 174
Oxide-coated cathodes, 4 class B, tuned, 221
disintegration, 31 untuned
work function, 3 class C, tuned, 229
Plate current, average value of, 226
P total differential of, 17
Plate dissipation, 174
Parallel feed amplifiers, 170 in class C amplifier, 229
Paraphase amplifiers, 187 during modulation, 331, 335
Peak inverse voltage rating (see Inverse permissible, 222
peak voltage rating) in tuned class B amplifier, 221
Peaking circuits, 128ff late efficiency, amplifier, class A, 174
Pentode, 24ff. series-fed, 175
in class A amplifier, 176 shunt-fed, 176
in class B a-f amplifier, 184 tuned class B, 221
distortion in, 25 tuned class C, and operating angle,
equivalent circuit for, 63 233
input admittance of, 63 Plate modulation (see Modulation, plate)
interelectrode capacitances in, 63 Plate resistance, diode, 273
load resistance for, 177 measurement of, 46
modulation with, 337 negative, in tetrodes, 22, 263
plate characteristics of, 24 numerical valucs of several tubes, 492ff.
power, 176 push-pull amplifiers, 186
remote cutoff, 25 tetrode, 23
in tuned amplifier, 192 triodes, 17, 47
voltage gain with, 63 variable nature of, 18
Phase characteristic, 74 Plate-plate resistance, 186
Phase distortion, 54 Polar diagram of feed-back amplifier, 89
Phase inverter, 187 Positive feedback, 81
Phase modulation (see Modulation, Positive ion bombardment, 7, 28
phase) Positive ion current, 7, 27
Phase relations, amplifier, 42, 74 Positive ions (see Ions)
thyratron, 285 Potential, breakdown (sparking), 28
Phase-shift characteristics of tuned am- Potential distribution, 10, 27
plifier, 196 glow discharge, 27
Phase-shift control, ignitron, 288 parallel-plane diode, sIlacc-charge flow,
Phase-shift distortion (see Phase distor- 10
tion) in plasma, 27
Phasemeter, electronic, 483 Power, rectifier, output, 273
Photocells (see Phototubes) input, 274
Photoelectric emission, 4ff. maximum (see Maximum power
gas amplification, 6 output)
Phototubes, vacuum, 5 Power amplifiers (see Amplifier, power)
gas-filled, 6 Power oscillator (see Oscillators, power)
volt-ampere curve, 5 Power sensitivity, 55
Pi-section filters, 305 Power series, 48
 INDEX 525
Power supply, voltage-regulated, 306ff. Reactance tube, 373
glow-tube, 306 Receiver, automatic volume control in,
series vacuum-tube, 308 355
electronically regula ted, 309 band width in
output resistance, 312 for amplitude-modulated waves, 212
stabilization ratio, 311 for frequency-modulated waves, 386
PPI presentation, 470 frequency modulation, 386
Predistorter in frequency modulation, 383 discriminator, 387
Preemphasis of high frequencies, 377 limiter, 386
Pressure, mercury, as function of tem- selectivity of, 192, 206
perature, 30 single side band, 357
Pulse generation, 434ff. superheterodyne, 362
by differentiation (see Peaking circuits) tracking of, 359
with gas tubes, 435 volume control in, 355
line-controlled, 436 Recombination, ions, 28
Push-pull amplifier, class A, 178ff. Rectification, definition of, 271
circuit, 178 Rectification characteristics, 353
composite dynamic characteristic of, Rectifier filters (see Filter, rectifier)
184 Rectifier meter, 280 .
composite static-characteristic Rectifiers, circuits, bridge, 279
curves, 181 full-wave single-phase, 276
distortion in, 185 gas tubes, 277
equivalent circuit, 180 half-wave, single-phase, 271
harmonics in, 185 voltage-doubling, 280
plate-plate rcsistance, 180 voltage-quadrupling, 281
power output, 185 controlled types (see Controlled recti-
class B, 186 fiers)
driver stages for, 187 Rectigon tube, 31
single-tube para phase, 187 Reflected resistance, 170
two-tube paraphase, 117, 187 Regeneration, in amplifiers (see Feed-
back, positive)
Regulation, voltage (see Power supply)
Q Regulators, electronic (see Power supply,
electronically regulated)
Q point, 41
glow-tube, 306
Quadrupler, voltage, 281
Relaxation oscillator (see Oscillators, re-
Quiescent point, 41
laxation)
Remote-cutoff tubes, 25
R
Resistance, bleeder, 303
grid, for high-vacuum tubes, 39
Radio communication, amplitude modu- input, of tubes, 61
lation, 320 load (see Load resistance)
frequency modulation, 386 negative (see Dynatron)
modulation essential for, 320 plate (see Plate resistance)
single side band, 357 multielectrode tubes, 21
Radio frequency amplifiers (see Tuned numerical values for several tubes,
amplifiers) 492ff·
Reactance, primary, effect on amplifier triode, 18
operation, 77 plate-plate (see Push-pull amplifier;
transformer, leakage, effect on ampli- Plate-plate resistance)
fier operation, 79 reflected, 170
526 ELECTRON-TUBE CIRCUITS

Resistance-capacitance coupled amplifier. Second harmonic, in rectifiers (see Ripple

68fJ. factor)
applicability of, 68 in tubes (see Nonlinear distortion)
feedback in, 88 Secondary emission, 6
frequency characteristic, 74 by ion bombardment, 6
gain of, high-frequency, 73 suppression, in beam tubes, 27
low-frequency, 71 in pentodes, 24
mid-frequency, 71 in tetrodes, 22
half-power frequency, 72, ~73 Self-bias in amplifiers, 64
extending (see Amplifier, broad- feedback through, 64
band, compensated) in push-pull amplifiers, 179
universal amplification curve, 74 Semigraphical analysis, class B and class
Resistance-capacitance tuned oscillator C amplifiers, 223
(see Oscillators, resistance-capaci- Series, Fourier (see Fourier series)
tance tuned) power (see Power series)
Resistance-coupled amplifiers, 112 Taylor (see Taylor series)
(See also Amplifier, direct-coupled, re- Shield-grid thyratrons, 34
sistance-cou pled) Shunt-feed amplifiers, 76
Resistance-stabilized oscillator (see Os- Shunt peaking in amplifiers (see Ampli-
cillators, stabilization of, resistance) fier, broad-band, compensated)
Resonance, parallel, 194 Side bands, amplitude modulation, 322
transformer-coupled circuit, 196 separation of, from carrier, 357
frequency modulation, 366
universal resonance curve, 196
Side frequencies, detection of (see Detec-
Ripple factor, capacitor filter, 300
tion)
definition, 275
for frequency modulation, 366
with electronic regulators, 315
receivers for (see Receiver)
full-wave circuit, 277
separation of carrier from, 325
half-wave circuit, gas tube, 279
Signal voltage (see Excitation voltage)
vacuum tube, 276
Sources of electrons, 1
inductance filter, full-wave, 295 (See also Electron emission)
half-wave, 294 Space charge, cloud, 9
L-section filter, .301 in glow discharge, 28
multiple L-section filter, 305 limitation of current by (see Space-
pi-section filter, 305 charge limited current)
in plasma, 28
S positive ions neutralize, 29
Space-charge flow (see Space-charge
Saturation, in phototubes, 4 limited current)
in screen-grid tubes, 21 Space-charge limited current, factors in-
space charge limited current, 11 fluencing, 11
temperature limited current, 2 parallel planes, current between, 9ff.
tungsten filament, 3 three-halves-power dependence, 11
Saw-tooth generators (see Sweep gener- Sparking potential, 27
ators) Square-law modulation (see Modulation,
Scaler circuits, with trigger circuits, 422 square-law)
ring counter circuits, 423 Stability of oscillators, amplitude, 254
Schedule method in Fourier analysis, 49 frequency, 379
Screen grid in tubes, 18 reactance-tube, 379
Screen-grid tubes (see Tetrode) Stabilization of amplifiers by feedback, 88
Screen supply voltage, 65 Stage, driver, 187
 INDEX 527
Stagger-tuned amplifier (see Tuned am- Temperature, anodes in diodes, 36, 174
plifiers, stagger-tuned) arc cathodes, 31
Starter probe, 34 glass envelope of tubes, 13
Static characteristics (see Plate char- mercury pressure, 30
acteristics) rating of diodes, 12, 35
Supercontrol tubes, 25 thermionic cathodes, 3
Superheterodyne (see Receiver, super- Temperature limited current, 11
heterodyne) Tetrode, 18
Suppression of carrier wave, 325 (See also Beam power tubes)
Suppressor grid modulation, 337 coefficients of, 23
Surge current rating, 12 disadvantages of, 24
Sweep generators, for cathode-ray tubes, equivalent circuit for, 62
capacitive, 443 gas (see Thyratrons, shield-grid)
characteristics of, 442 input admittance, 62
electrostatic deflection, 442jJ. interelectrode capacitances in, 62
free-running, synchronization of, modulation with, 337
445 mutual conductance of, 23
thyratrons in, 444 negative resistance in, 22
vacuum tubes in, 446 plate characteristic of, 22
linearizing methods, auxiliary time screen-grid, 22
constant, 450 secondary emission, 22
constant-current generator, 448 Thermionic cathodes, carbonization, 3
feedback, 451 directly heated, 4
high-voltage charging, 448 efficiency, 3
inverse curvature, 450 heat-shielded, 4
push-pull, 452ff. indirectly heated, 4
(See also Push-pull amplifier, oxide-coated, 3
driver stages for) (See also Oxide-coated cathodes)
triggered, 447 temperature of, 3
circular, 468 thoriated-tungsten, 3
electromagnetic deflection, 460ff. Thermionic emission, 2ff.
characteristics of, 460 constants of (Ao, bo, Ew), 2
effects of distributed capacitance in, Dushman equation of, 2
465 efficiency of, 3
linearizing methods, exponential, power for heating cathodes, 64
460 saturation, 2
feedback, 466 work function, 3
oscillatory, 461 (See also Work function)
trapezoidal, 462 Third harmonic (see Nonlinear distor-
rotating radial (PPI), 470 tion)
Swinging chokes for filters, 303 Thoriated-tungsten cathodes, 3
Synchronization of blocking oscillators, Three-halves-power law (see Space-charge
411 limited current)
multivibrators, 408 Threshold frequency, 5
Threshold wave length, 5
T Thyratrons, 32ff.
argon-filled, 33
Tank circuits for tuned power amplifier, cathode for, 4
214 (See also Oxide-coated cathodes)
Taylor series, for triode, 151 cathode-ray oscilloscope sweep circuit,
(See also Power series) use in, 444
528 ELECTRON-TUBE CIRCUITS

Thyratrons, control of average current, Trapellium distortion in cathode-ray

282 tubes, 453
critical grid control curves, 32 Trigger circuits, 419ff.
cutin, 281 Eccles-Jordan, 419
deionization time, 36 pentode, 422
filters for use with, 304 scale-of-two, 420
grid glow tube, 30 pentodes in, 421
ignitron controlled by, 289 Triode, gas (see Thyratrons)
ionization time of, 36 vacuum, 14ft.
mercury vapor pressure in, 30 amplification factor (see Amplifi-
positive control tubes, 33 cation factor)
pulse generators, use in, 435 characteristic curves (see Character-
line-controlled, 438 istic tube curves)
shield-grid, 34 coefficients of, 18
tube drop, 282 grid currcnt in, 15
Time, deionization, 36 (See also Grid current)
Time delay, generation of (see Delay, grid heating in, 231
generation of time) input admittance of, 60
in amplifiers, 54 interelectrode capacitances in, 58
Transconductance, grid-plate (see Mutual mutual characteristics, 16
conductance) mutual conductance (see Mutual
Transfer characteristic (see Mutual char- conductance)
acteristics) plate characteristics, 16
1'ransformer, interstage, 76 plate resistance of, 17
capacitances in, 77 space charge in, 15
output, push-pull, 180 square-law detection with, 343
r-f air core, 192 square-law modulation with, 323
Transformer-coupled amplifiers, 76ff. Taylor series for, 17
gain, high-frequency, 79 Tube constants, multielectrode tubes, 23
low-frequency, 78 relation among, 24
mid-frequency, 77 triodes, 18
impedance matching, 170 Tube drop, in arcs, 28
push-pull amplifier, 178 in glow tubes, 28
tuned, in amplifiers (see Tuned ampli- Tubes (see Characteristic tube curves)
fiers, class A, single-tuned, trans- Tuned amplifiers, class A, 192ff.
former coupled) coupling in, 192
Transformer insulation stress, full-wave distortion in, 192
rectifier, 277 double-tuned, 199ff.
half-wave rectifier, 274 band width, 204
Transformers as impedance matching de- cascade, 205
vices, 170 band-width reduction, 206
Translation, frequency, by modulation gain, 200
(see Modulation, frequency trans- gain-band-width product, 206
lation by) maximal flatness in, 201
Transmission line" pulse generators, 436 transitional coupling in, 203
Transmitters, amplitude-modulated, 340 single-tuned, 192ff.
frequency-modulated, 373, 381 cascade, 204
using phase-shift modulators, 381 band-width reduction factor,
using reactance-tube modulators, 205
373 direct-coupled, 193
stabilization of, 379 band width, 195
 INDEX 529
Tuned amplifiers, class A, single-tuned, Tungsten, filaments, 3
direct-coupled, gain, 194 work function, 3
gain-band-width product, 206
transformer-coupled, 196 U
band width, 198
gain, 197 Universal resonp,nce curve, 196
optimum coupling in, 197 y
stagger-tuned, 207ff.
comparison with double-tuned, Vacuum-tube voltmeter, 473ff.
208 average, 476
class B power, 216ff. d-c,479
analysis tabulation for, 229 logarithmic, 479
analytical treatment of, 219 peak, diode, 476
applications of, 213 grid leak, 477
optimum conditions in, 222 negative feedback, 478
phase relations in, 218 slide back, 478
semigraphical analysis for, 229 rms,474
class C power, 223ff. Van der Bijl modulation, 323
analysis tabulation for, 229 Variable-mu tube, 25
approximate analytical solution, 234 Video amplifier (see Amplifier, broad-
cathode-modulated, 337 band, compensated)
conditions in, for high efficiency, 233 Volt-ampere characteristics (see Char-
design considerations for, 231 acteristic tube curves)
grid-bias modulated, 332 Voltage, disintegration, 31
grid current in, 226 Voltage comparator, 430, 434
grid driving power in, 230 Voltage-doubling rectifier, 280
plate current in, 226 Voltage gain, cascade amplifier, 70
plate dissipation in, 230 decibels of, 56
plate efficiency of, 229 with feedback, 83
plate modulation of, 327 (See also Amplifier, voltage-gain)
Q of load for, 213 Voltage-quadrupling rectifier, 281
semigraphical analysis for, 223 Voltage-regulated power supplies (see
neutralization, grid, 239 Power supply, voltage-regulated)
plate, 239 Voltage-source equivalent circuit, 43
Tuned circuits, class B and class C ampli- Voltmeter, vacuum-tube (see Vacuum-
fiers, 213 tube voltmeter)
half-power frequencies in, 196
in oscillators, 248 W
single-tuned, direct-coupled, 193
single-tuned, transformer-coupled, 196 Wattmeter, electronic, 488
double-tuned, 199 Wave length, threshold, 5
Tuned-grid oscillator (see Oscillators, Wide-band amplifiers (see Amplifier,
tuned-grid) broad-band; Tuned amplifiers)
Tuned-grid-tuned-plate oscillator (see Work function, 2
Oscillators, tuned-grid-tuned plate) definition, 2
Tuned-plate oscillator (see Oscillators, Dushman equation involving, 2
tuned-plate) oxide-coated cathodes, 3
Tuned power amplifier (see Tuned ampli- photoelectric emission, 5
fiers, class B power; Tuned amplifier, table of values, 3
class C power) thoriated tungsten, 3
Tungar tubes, 31 tungsten, 3