1

yPete; H..Sydenham, M.E.

F.I.I.C.A., M.I.E.A
Professor and Head of School of
Electronic Engineering, .
South Australian Institute of Technology.

egEWTRONICS TODAY RITERfiATIONAL PUBlip TION

r
THE COMMONWEALTH COPYRIGHT ACT OF 1968 EXPRESSLY FORBIDS THE REPRODUCTION OF ANY
MATERIAL WITHOUT THE COPYRIGHT OWNER'S PRIOR PERMISSION.
ALL RIGHTS RESERVED. NO PART OF THIS PUBLICATION MAY BE REPRODUCED IN ANY FORM OR BY
ANY MEANS WITHOUT THE PERMISSION OF THE COPYRIGHT HOLDERS - ELECTRONICS TODAY
INTERNATIONAL. UNDER NO CIRCUMSTANCES WILL PERMISSION BE GRANTED TO PHOTOSTAT ANY
PART OF THE CONTENTS AND ANY PERSON OR ORGANISATION FOUND SO DOING WILL BE
PROSECUTED UNDER THE TERMS OF THE COMMONWEALTH COPYRIGHT ACT 1968.
Electronics
It's Easy
Vol.2
by Peter H. Sydenham, ME., Ph.D., F.Inst.M.C., F.I.I.C.A., M.I.E.Aust. Professor
and Head of School of Electronic Engineering, South Australian Institute of
Technology.

Introducing Digital Systems 5

The Algebra of Logic 10
Integrated Circuit Forms of Logic Functions 16
Digital Sub- Systems Counters and Shift Registers 24
Digital Displays — Their Development and Forms 32
Code Converters and Display Systems 39
D-A and A- D Conversion 49
Digital Instruments and Test Equipment 58
More About Digital Instrumentation 65
Digital Computers 74
Computer Peripherals, Stores and Microprocessors 84
Transmission Links and Coupling 93
Oscilloscopes 102
Oscilloscopes — The Refinements 109
Chart Recorders 116
The Control of Power 124
Index 130

41)
Special publications editor: Jan Vernon
Layout: Bill Crump
Managing editor: Collyn Rivers

Published by the Scientific Division of Modern Magazines ( Holdings) Ltd,

15 Boundary Street, Rushcutters Bay, NSW 2011 in May 1981.
Printed by Offset Alpine Printing Pty Ltd, Silverwater, NSW.
Distributed by Gordon and Gotch.
Copyright: Electronics Today International. * Recommended retail price only.
 Acknowledgement

Electronics it's Easy was first published as a monthly series in Electronics Today
International. It was republished as three separate volumes during the period
1975-1976.

A second edition was published in 1977-1978 and arevised version of Volume 1

was produced especially for the TAFE School of Electrical Engineering in Sydney in
1980.
This, the third edition of the complete series, has been totally revised and new
material added where relevant. At the same time the material has been re-arranged
so that the complete work now appears in two volumes rather than the original three.
Apart from a card-cover edition, the work is also available in hard- back form —
details at the end of both volumes.
The author would like to thank Brian Chapman (technical editor of Electronics
Today International from 1973-1976) for editing the original edition. Thanks also to
Ron Farley and Maree Breen for turning hundreds of my rough pencil sketches into
clear and concise engineering drawings.
The very considerable work of turning my original individual articles into book
form, and of editing and producing this latest revised edition, has been done by
Janet Vernon B.A., special publications editor of Electronics Today International.
Iwish to thank the following for use of illustrations:—
Philips; BWD; Science Museum ( London); Advance Instruments; Electronics
Australia; Hewlett-Packard; Teledyne Philbrick; Siemens; Tektronics; Non- Linear
Systems; John Fluke; Heidenhein; Lloyds Bank; AEI Associates; Kirklands; Novus;
Gerber Scientific; Honeywell; Motorola; National Semiconductor; Levers Rich;
Andrew Antennas; Vu- Data; Data- Lab; Marconi; Australian Post Office; Kent Instru-
ments' Record Electrical Company; Goerz, Hathaway Instruments; Devices; and
Kepco.

Peter H. Sydenham. Adelaide, April 1981.

4
 Introducing

21 digital systems

WE BEGAN this course with a electrical- input signal is tnen Other factors that decide the choice
discussion of electronic systems in conveyed, through the system being of signal form are the cost of
general; what they do, how they do it modified, and converted in different components needed, the size of
and how we can progressively break ways as required. The output signal equipment and power supply
down a complex system into from the system is fed to actuators demands. Today, the enormously
fundamental building blocks. The which convert this signal back to large- volume production of digital
example chosen then, aTV system, uses, real-world variables at the output. It is circuits, especially when marketed as
in the main analogue signals. These we the differences between two basic large-scale integrated systems, coupled
know from other parts of the series are means of transmitting and converting with the tremendous effort that has
those signals that contain information information that we are concerned been expended on digital techniques
with now. for computing markets, has now
in the form of many continuously
changing levels of an electrical voltage tipped the balance heavily in favour of
We have seen in the earlier part
(or current). using digital methods. This is now true
dealing with information that both
Although we have already even for what have traditionally been
analogue and digital signals can convey
analogue applications. It may well now
introduced the concept of the on- off, the same information between two
or digital kind of signal, the course so be cheaper to use a mass-produced
points. It is a matter of how the
far has concentrated almost entirely digital assembly for a more unusual
information is coded on the signal. It
analogue requirement, even when
on the linear, analogue circuits used in is not possible to state categorically
electronic systems. The time has now analogue circuits could easily supply
that one signal form is better than the
come to study an alternative the need.
other. Each has its advantages
Take, for example, the choice
philosophy and practice, by which depending upon the application.
tasks can be accomplished in another confronted when purchasing a good
Analogue systems can process the
manner — the digital electronic quality multimeter. The traditional
same information using far less
approach. multimeter can be represented as a
components, than their digital
resistive network driving a display
counterparts, but they are unable to
meter — see Fig. 1. The signal level can
DIGITAL OR ANALOGUE provide anywhere hear the same
be ascertained by the degree of pointer
SIGNAL APPROACH? ultimate accuracy, precision and
deflection seen on the meter. High
By itself a purely electronic system long-term stability. In some uses, such
input impedance units incorporate a
has no real value until it is applied to as precise mathematical computation,
linear amplifier to buffer the signal
the real world we exist in. At the input digital techniques are a must. The
source against a relatively low-
of a system physical variables are same holds true for measuring
impedance meter movement. Apart
measured by sensors that convert the equipment needing better than around
from the selector switch which has
information, from the original form of 1 percent, or perhaps 0.1 percent,
discrete settings, all components work
energy, into an electrical signal. This accuracy.
with analogue signals and this means
they must be linear in operation and
adequately stable with time. Some
SELECT RANGE
components — the ballast and shunt
resistors, for instance — must be made
to tolerances that require expensive
hand- made manufacture. We can
summarize the situation as one where
RESISTIVE
only afew components are needed but
NETWORK OUTPUT
ANLOGUE they are inherently expensive.
INPUT SIGNAL The alternative is to use a special
TO BE BUFFER
AMPLIFIER DISPLAY METER circuit that we will discuss in detail in
MEASURED
a later part. This is called an
Fig. 1. Schematic representation of analogue multimeter system. analogue- to-digital converter ( or just
A- to- D converter). As represented in
AUTOMATIC the schematic of a digital multimeter
RANGE given in Fig. 2, it converts the
SELECT
analogue input level into a digital
signal form that is then used to drive a
digital readout display. These units
ANALOGUE
INPUT SIGNAL display the output value as adecimal
ANALOGUE1 0.35
TO BE TO number rather than as the position of
OUTPUT
MEASURED ANALOGUE DIGITAL DIGITAL
DIGITAL a pointer as is used in totally analogue
NETWORK SIGNAL COUNTER
DISPLAY
CONVERTER systems. We will see, as we delve more
deeply into how such a system works,
that the digital alternative uses literally
Fig. 2. Digital multimeters contain black boxes that operate with digital rather than
analogue signals. dozens of active elements and many
5
 CONTACTS REMAIN OPEN STATE
CLOSED FOR THIS TOLERANCE

V /INSULATION
POSITION TOLERANCE

ELECTROMAGNET
CLOSED STATE OF
PROVIDES MOVEMENT
NORMALLY-OPEN OPEN STATE OF
UPWARD TO CLOSE
CONTACTS FORCE REMOVED NORMALLY OPEN
CONTACTS
CONTACTS

Fig. 3. Mechanical switches

are designed to allow con-
siderable lattitude in the
open and closed contact
positions, thus ensuring
reliability. Electro-
mechanical switches come
in many shapes and sizes.

many more passive components than Yet, somewhat strangely, they can be with each state can be very wide.
an analogue type of multimeter. Yet, far less expensive, much more accurate Consider the basic mechanical switch
today, there is little difference in the and more reliable. Undoubtedly the having two contacts as shown in Fig.
cost of either alternative for the same trend in electronic systems is toward 3. When the contacts are disengaged it
accuracy. The digital scheme, however, more use of digital solutions — but this matters little how much further the
can be made considerably smaller, may does not mean that analogue systems designer separates them; the further
be made more accurate, uses no more have no place in electronics. they separate the less the chance of a
power and may even have circuitry One dominant reason why digital spurious make-condition occurring.
that automatically selects the most systems can be so reliable and positive Conversely, when closed the spring
appropriate range for itself. to design is that the signal operations action will ensure contact over a wide
Another example is found in involve switching rather than range of relative positions. The harder
computing. We have seen how continuous- mode action. We, there- the two contacts are pressed together
operational amplifiers — those that fore, begin our study of digital systems the better the reliability, but there will
perform linear arithmetic inherently by looking at the design merits of be negligible electrical change in the
can be used to solve equations and do various switching devices, starting with circuit- made state.
complicated arithmetical operations in the mechanical kind. Continuing with the mechanical
what are called analogue computers. MECHANICAL SWITCHES switch example we can also easily see
These can provide extremely powerful The ON-OFF switch has only to that a switch with heavily
solutions of mathematical problems define two states of circuit operation over- travelled contact pressure or
for quite small outlays. But only if the and hence the tolerances associated excessive opened distances will be slow
problem does not require high
accuracy — then digital computation is Vs
needed. Another instance where digital
method is a must is when the problem
Ris HIGH
involves logical type operations where
0—ic/eo-0
'
yes- no decisions are needed. Digital
VioVout Is
computers can sort information into (a)

groups and decide which way to o o o

Fig. 4. Switches can SCHEMATIC
proceed at a decision junction. Tliis EQUIVALENT CIRCUIT
handle large power signals Vs
will become clearer when we discuss
with little loss (a) open I
the mathematics of logic which is (b) closed. R,,LOW
quite unlike normal algebra. OAvWO
As with the multimeter example,
Yin Vout
digital computers also involve many
more components than the analogue o o o
units that would perform similar tasks.
6
 the development of workable
ELECTRONICS - it's easy! solid-state switching systems has been
influenced by the need to make-do
with this shortcoming. A relatively
recent newcomer to the solid-state
j SWITCHED
\te
switch, which overcomes the last
SVVITCH CIRCUIT disadvantage, is the solid-state
SWITCHED
ACTIVATING
SWITCH CIRCUIT opto -electronic isolator, shown in Fig.
SIGNAL
ACTUATING 6. This uses the actuating input to
INPUT
SIGNAL energise a solid-state light- emitting
ACTUATION diode ( LED); this, in turn, reduces the
MECHANISM resistance of a light sensitive detector
NO ELECTRICAL ELECTRICAL that acts as the ' contact'. This device is
CONNECTION EXISTS CONNECTION EXISTS
used in a minority of switching
(a) ISOLATED SWITCHING (b) NON- ISOLATED SWITCHING operations involved in digital circuitry
where extremely high isolation is
Fig. 5. Unlike electro-mechanical switches, most solid-state electronic switches do not
required between the switching and
provide ideal isolation between the actuating signal source and the controlled source.
the switched circuits.
The two most commonly used
by the made- state switch resistance. solid-state switching techniques are
to change to the opposite mode
What is high or low is purely relative, those using two- terminal diode designs
because greater force, or more travel,
depending upon the impedances of the and three- terminal active element
is needed to effect the change.
circuit elements connected to the designs based on devices such as the
Another feature of the two- state
switch. A perfect switch provides transistor and other solid-state
switching circuit is that the switch's
infinite open- circuit resistance and amplifying devices.
power rating can be very small
zero closed-circuit resistance. Typical Let us first look at adiode wired to
compared with the load power being
resistances encountered in a small provide a switching function. In Fig. 7
controlled. This arises because in each
relay are from many megohms adiode is connected to abias supply V
of the two states the switch has only
(contacts open) down to mere and to the input as shown. When the
to dissipate very small power losses.
milliohms ( contacts closed) thus giving input voltage Vin is more negative that
When open, see Fig. 4(a), the voltage
across the switch is maximum but the excellent switching characteristics. the bias voltage V the diode is
Solid-state switches normally do not forward- biased providing a quite low
current minimum. The power rating
provide such large resistance ratios resistance path between the input and
needed of the switch ( neglecting arcing
(some special devices come close) the output terminals. In this state
effects in this case) whilst open is,
giving around a megohm to ahundred Vout will be closely equal to Vin. If
therefore, the product Vs.ls, and this
ohms change which is adequate for the bias voltage ( or the input voltage)
is always very small, for only leakage
most logical tasks performed by digital are changed to make Vin more positive
currents flow when the switch is open.
systems. than V the diode becomes
When closed, the situation is reversed;
reverse- biased placing a high-
the current is now of the maximum Another disadvantage of most
resistance between input and output.
value but the voltage drop is merely solid-state switches is that, as we will Thus, by changing V from positive to
that due to resistive losses in the made see below, the circuit connected to the
negative we have produced aswitching
contact ( which can be very small). In switching part of the solid-state switch
action between input and output
practice the change of state from one is not completely isolated from the terminals.
condition to the other is so rapid that circuit actuating the switch mechan- A similar action is provided if the
we can consider the switch as only ism. This concept is shown in Fig. 5.
diode is wired in shunt across the line
ever being in the fully- off or fully- on At times this is most inconvenient and rather than in series as shown in Fig. 8.
case. This low- loss feature is used to
The state of V decides whether the
effect in power- supply switching
diode shunt 3 the line ( when forward
regulators where the " made" to
biased) or not ( when reverse- biased).
"not- made" times of a vibrating
ACTUATING SWITCHED In either design it is important that
contact are varied to pass the required INPUT SIGNAL
the diode resistances in the two states,
amount of average power.
the output impedance ( Rs) of the
preceding stage connected to the
SOLID-STATE SWITCHES
LIGHT EMITTING PHOTO-SENSITIVE input, the load impedance ( RL)
Originally digital circuits did indeed DIODE ( LED) SILICON DETECTOR
use electro -mechanical switches; the (BANDWIDTH 20 wiz) connected to the output and the
relay as we know it. These are still Fig. 6. Opto -
electronic switches such as
HP5082 series can provide a very close
used in some circumstances today but
approximation to the low-power mechanical V ( BIAS SUPPLY)
their size, cost, slowness of switching switch and are much faster in operation.
and possible unreliability now make
them a poor choice, for logic
V
applications, compared with solid-state (BIAS SUPPLY)
switching alternatives.
A switch by definition, is a device
that provides either a satisfactorily
high or low resistance between two
points, with the state being rapidly
reversed by an external control input. Vin Vnut
It can be used in series or shunt to
o o
effect control. The degree of isolation
provided is decided by the open-state Fig. 7. Diode wired to provide switching Fig. 8. Diode wired to provide shunt
switching action.
resistance; the power rating is decided action between input and output.
7
 FORWARD BIAS circuits may apply areverse polarity to
ensure this); the transistor is switched
off with VE being virtually at the
Ge
supply voltage. In the on-stage the
transistor provides a low- resistance
path between its collector and emitter:
when off, ahigh- resistance path.
The transistor switch, unlike
REVERSE 1.0V non-amplifying diodes and mechanical
BREAKDOWN contacts, does not directly pass the
-v R ,VF
input signal but instead replicates a
signal current in its base by providing
0 V 10.6 V 11.0 V
.3 an equivalent change in collector
current or voltage. In reality a large
proportion of digital circuits
regenerate in this way with the output
signal change closely following that of
the input.
Fig. 9. Typical characteristics of silicon and At either of the circuit operating
—IR points A or B the transistor is operated
REVERSE BIAS germanium diodes.
well within its allowable power
dissipation. As we should expect, a
bias- resistor value ( R) are all chosen to The speed at which diodes can given transistor used in a switching
have the right ratios in the two switch is a function of circuit values mode can handle a greater power than
switching states. Adequate switching and the characteristics of the diode. It if operated as a linear amplifier. A
action will not result if the bias input is routine practice with diodes to little thought will also show that the
is allowed to dominate the values switch at tens of megahertz rates or
load line can, in switching use,
being switched. higher. Mechanical switches are limited
intercept the maximum dissipation
It is important that the bias voltage to less than 1kHz at the very best.
curve, the reason being that the
actually developed across the diode is Later in the course we will see how
transistor does not dwell long enough
sufficient to produce a diode these basic diode switches can be used
in states other than A or B to produce
forward-current greater than a value to perform logical operations by
deleterious heating. It is vital, however,
around the knee of the IF against VE connecting more than one diode to the
in such designs to ensure that the
characteristic — see Fig. 9. For a same bias source. Such connections are
switching action is rapid between
silicon diode this requires at least called gates.
states, and that the device never dwells
700mV, a germanium diode at least Now to the use of three-terminal
350mV: these values vary little with on the way through. A ramp input
devices, transistors for instance, as signal may well destroy a stage
make or shape of particular device, switches rather than as linear designed to switch!
being a parameter of the amplifiers. This can be explained using The above explanation is most basic
semiconductor material itself. Note the l c versus Ec characteristics of a
how a quite large change in forward — reality requires other criteria to be
typical transistor, as is given in Fig. 10. recognised to obtain more ideal
current hardly changes the dynamic The two switching states occur when switching. Like the over- travelled
resistance once the knee is passed. lb is either large or small. A chosen mechanical switch, a transistor switch
(Dynamic resistance is the slope of the collector resistance value ( in common
characteristic which is reasonably with too much reverse- bias base
emitter configuration) establishes the current ( off- state) or too much on
constant beyond the knee). This load- line on the characteristic. In a state base current will be slower
reliable and constant loss switching switching- mode the transistor operates to operate than one not driven so hard.
(but not zero- loss) results over a very
around points A or B. At A, lb is large; This is because the charge associated
wide range of bias current conditions.
the transistor is, therefore, switched with the base current must be removed
In its reverse- biased state the diode on with VE being very close to zero to alter the state and the more the
provides a larger resistance. Fig. 9 volts. At B lb is small ( practical charge there is to move, the slower will
shows that germanium diodes do not
provide as high an ' open' resistance as
do silicon diodes — this is because the
slope of the germanium characteristic
is not as horizontal as that of silicon.
Nevertheless both slopes represent IC
higher resistance than in the (ON STATE) /
A
forward- biased case, proving that • ,
resistance of the diode changes
1
markedly. Again, we see that both %
1
reverse- biased curves are closely linear
meaning constant resistance or, in LOAD LINE --
other words, constant " open circuit"
switch resistance.
r
When selecting the value of MAXIMUM —
POWER -..
switching bias to apply it can be seen DISSIPATION -_
/ , 1- 1
from Fig. 9 that too high a value for CURVE
Ec
silicon devices will cause breakdown at g

the zener point, providing instead, a (OFF STATE)

made-state that could cause total

Fig. 10. In the switching mode the transistor is operated at either end of the chosen load
failure of the device. line.
8
 value capacitor placed across the
resistor feeding the input of the
switching stage. When fast switching
Di ( FEEDBACK) signals occur the capacitor provides a
low impedance path around the
resistor which must be of areasonably
high value to supply correct dc signal
OUTPUT
LEVEL level requirements. Yet another
V IN ( TO CHANGES technique is to use feedback between
SWITCH IT GOES
POSITIVE) Fig. 11. High-speed switches the collector and the base to speed up
are designed to remain non- the switching transition yet hold the
saturated. This circuit em-
stage in a non-saturated state once
ploys feedback 01 with D2
R ( REVERSE BIAS TO providing a voltage supply switched. Fig. 11 is a non- saturating
TRANSISTOR) needed. The speed-up switch circuit — one of many
capacitor is C. Diode D3 possibilities. It shows how the basic
assists reduce the delay transistor needs the addition of more
time.
components to realise fast switching in
discrete designs.
The integrated-circuit revolution
has provided us with inexpensive,
be the switching time. Solid-state be needed to just turn it on. This ready-made digital circuits of great
switches operated very positively by speeds up the charge movement but sophistication. These are extremely
use of large drive currents are said to would take the device into deep basic yet super fast — see Fig. 12.
be working in asaturated state. saturation unless clamps are added Rarely does one now have to consider
that hold the circuit nodes at given the in-depth design of switching
Certain circuit devices can be added values. Diodes acting as switches are circuits. The task is usually one of
to the basic solid-state switch to speed often used to hold a point at a given devising a system using a few basic,
up the response. The first is to supply voltage. A second circuit addition is digital system building blocks which
a much larger input signal than would the speed-up capacitor. This is a small have been so developed as to facilitate
+5V their ease of connection into systems.
The reliability of the switching
state of an electronic circuit is one
reason for the widespread use of
digital techniques. There is another
equally important reason for the use
of digital signals and that is that
philosophers and mathematicians of
the past have developed powerful ways
to process logical information by way
of special algebra and techniques. This
is employed to design complicated
switching circuits and other digital
Fig. 12. Transistor- transistor logic,
MULTIPLE- systems with the simplest possible
TTL is acommonly accepted in-
EMITTER tegrated circuit manufacturing circuitry. In the next part we look at
TRANSISTOR
method. This gate combines up these philosophical concepts in
to three inputs switching with a readiness to return to a discussion of
delay of only 10 ns. The whole
circuit is integrated on a common
the basic, digital- system building
OV chip. blocks. •

Notes

9
 The algebra

22
of logic

MATHEMATICS is a kind of contributed to the first systematic milestones in digital computer

shorthand language which enables us arrangement of Aristotle's logic. real isat ion.
to present a physical process, on Boole took the concepts further Shannon's 1938 paper "A Symbolic
paper, with symbols which may be than the Ancients by substituting Analysis of Relay and Switching
manipulated in order to gain a better mathematical symbols in place of the Circuits" was a paper of very practical
understanding of the process. It is thus basic logical situations. This symbolic relevance for it described how to put
atool which aids understanding. logic became known as Boolean Boole's rather abstract logical albegra
The familiar kind of algebra which algebra. to work in engineering and computer
relates two variables, x and y, in
Little was achieved with Boole's design. But this was not the first
combinations such as x+y, x-- y, x.y, work for the next few decades. The recorded use of electrical logic circuits.
x/y, xY and others is a linear process
first machine to utilize his algebra to In a letter Charles S. Peirce wrote to
because the two variables can hold any solve logic problems, faster than by his former student, Marquand, around
value. It is this kind of algebra that is hand, was William Jevons' logical 1890 he expressed, in the words and
performed by analogue operational piano of 1869. Boole's contribution, circuit diagrams shown in Fig. 2, that
amplifiers.
however, had to wait until the early logical algebra could be performed
However, if x and y can only have 20th Century to find extensive with three switches in parallel or in
two possible states, such as a voltage application. One by one, logicians series, also stating that he felt
which is there or not there, we can advanced the techniques of logical electricity to be one of the best ways
ignore the actual value of the voltage
algebra: Pierce, Venn, Dodgson, to implement logical equipment.
(or whatever) and regard the variables Marquand, Pastore, Bollee. The Later theoretical studies con-
as behaving according to atwo- state or "Principia Mathematica" of Whitehead centrated on ways to ensure that
binary number system. Just what the and Russell ( 1910-1913) and the switching networks contained no more
two states are is of no importance Hilbert and Ackermann work " Math- 'switch contacts than were absolutely
whatsoever — they can be high or low, ematical Logic" ( 1928) were further necessary. Unnecessary contacts can
positive or negative, there or not there easily be unwittingly designed into
and even true or false.
complex switching networks — the
A mathematical algebra has been "spares" are called redundant
developed to cope with such binary switches. Shannon, in his M.Sc. thesis
systems: It is known as Boolean (Fig. 3) prepared at the famous
algebra — the algebra of logic, and it's Massachusetts Institute of Technology,
rules are somewhat different to those realised ways to systematically set
of linear algebra. Before delving into about analysing a given switching
the operation of Boolean algebra, it is network in order to reduce the
worth tracing it's historical develop- contact requirements to a minimum.
ment. Thus it was realized in the early 1940's
that really powerful digital computers
HISTORY OF SVVITCHING could be built using entirely electronic
MATHEMATICS components.
Later in the course we will be
Philosophers, those people who
dealing specifically with computer
apply special skills to resolving
systems. They are, however, but apart
paradoxes by the use of logic, have
of the total use of digital electronic
existed since the earliest civilisations.
methods — digital electronics finds use
The Ancient Greeks were so impressed
in an ever increasing number of
with logic that they wrote plays
instruments and devices.
around Aristotle's formally arranged
rules of logical deduction. The rules
for this process of reasoning were BASIC LOGIC GATES
handed down, largely by word of
A quite satisfactory way to begin to
mouth, through the Dark Ages, with
comprehend basic switching algebra is
little, if any, recognition of their value
to think in terms of mechanical switch
for logic in computation. It was not
contacts arranged in various different
until the early 19th century that the
configurations. That we draw them
use of logical rules in calculation was
and consider them as mechanical
established. This work was very much
contacts that are either open or closed,
the result of George Boole's 1854 work
does not imply that the contacts
(see Fig. 1) entitled " An Investigation
necessarily need to be mechanical —
of the Laws of Thought on which are
they are, today, more often than not
Founded the Mathematical Theories of
Fig. 1. In 1854 George Boole, an English the solid-state switches we discussed in
Logic and Probabilities", Augustus de logician, showed how ordinary algebra the last part.
Morgan, a contemporary, also could be applied to logic situations.
Groups of switches combining
10
 Fig.3. Claude Shannon published details
of "modern" digital computing design in
1937.
.A ,
C (.--4

t; 7k4.- A .

e•ce- s. • - 4:1.1 „— -
4-1( «
M IS which to communicate the idea. And
few digital systems are that easy:
C•,. t- .
1 alt fir., •
many contain literally thousands of
ç .• .{ Fig.2. The first known AND, OR and NOT gates.
T h•.• 1:.• 1(..A.
.
description of electrical We designate an OR function by
switching carrying out logic
LIT". .•••••. C . • means of the '+' symbol. This does not
• C. *•••.< is in this letter of C.S. Peirce
written around 1880. mean the same as our normal
understanding of addition. When
applied to decimal numbers it means
case either switch A or switch B will addition as we normally understand it.
digital signal levels are known as gates.
provide transmission. With binary numbers, however, it has a
We begin by considering the simplest
The AND and OR are basic logical different meaning and still another
possibilities where there are just two
functions. They need not necessarily meaning when designating an OR
contacts to build with.
They can be placed in series or be used only to describe electrical function. For eliample:-
in parallel, as shown in Fig.4. In each circuitry. They did, in fact, as we have In decimal addition 1+1=2
case different conditions exist between seen, arise originally from phil- binary addition 1+1=10
the transmission made through them osophical study of truths and falsities. OR addition 1+1=1
for the two positions of each of the Note that switch contacts are In Boolean algebra the OR meaning
switches. We denote the switch inputs always shown in their non-actuated of addition is the one that applies.
as A and as B ( and C.D. etc., if more condition and this brings us to another Thus A+B=Z means that A OR B
are involved) and the transmission as basic gate function which can be switch closed . will produce a
Z, thus using mathematical symbols to realised using only one switch. If, as transmission Z.
represent a physical situation. Imagine shown in Fig. 5, the switch A is We designate an AND function with
that the switches are wired in series actuated, Z is NOT enabled. If A is not a dot. The dot means logical
with a lamp: when a circuit is made actuated Z is enabled. A single switch, multiplication and is not to be
the lamp lights. therefore, can provide a NOT function confused with normal multiplication.
In the series case we need switch A if its contacts are closed in the However the truth tables for AND
and switch B to be made to obtain a non- actuated state. multiplication and normal multi-
transmission function Z. In the parallel Attempts to explain switching plication are the same. Thus when we
circuit action in words, as above, only give the Boolean equation A.B=Z we
applies for the simplest of situations. mean that if switch A and switch B are
The descriptive method becomes both closed there will be a
{›3.
prohibitive when, say, we have two transmission Z.
switches in series, in series with two The NOT function is designated as
switches in parallel, as shown in Fig. 6. a line over the switches algebraic
SERIES
Describing the action of all possible symbol giving Z = T1 /to mean Z is NOT
4
AND SWITCHING
Z = A.B. switch combinations on the lamp Z transmitted when A is actuated.
using words, is an inadequate way with Each of these functions have a

INVERTED
SWITCHING
PARALLEL
Z SWITCHING NOT
Z = A
Z = A+B
Z = A.B. ( C+D)
Fig.5. The NOT function is obtained by Fig.6. More complex switching functions
Fig.4. Two contacts wired in series or in
reversing the state of the switch operation. are best described in terms of logic algebra
parallel provide the basic logic functions
than by words.
of AND and OR.
11
 ELECTRONICS - it's easy! matching
levels.
voltage and impedance
However, when transferring
logic signals between devices from
different logic families translator
circuits will la, needed to make voltage
levels compatible.
Occasionally, but not commonly, it
is more convenient to reverse the levels
calling a 1 the lower voltage and 0
the higher. This is denoted negative
logic. Such asystem is however seldom
used in modern integrated- circuit logic
families.
Consider then the series contacts of
Fig. 4. Assuming we use the positive
logic convention where 0 represents an
open contact and 1 a closed
contact; it is easy to draw up columns
as given in Fig. 8.
When A and B are both 0 then so
also is Z, for no contacts are made.
Similarly, if either A or B are open.
When both A and B are closed, that is
a 1 each, then Z is made. This is called
atruth table.
Fig. 9 is the truth table for the
parallel contacts of Fig. 4. In this case
Z is 1when A or B are 1.
An interesting property of the AND
and OR functions is their dual nature
when negated. For example, if we
negate the inputs of the OR gate the
truth table becomes that of Fig. 10,

INPUTS OUTPUTS
AB Z ZAND= A.B
00 0 A
01 0 11Y
10 0 B -
symbolic representation as black- boxes negation) is basically an amplifier
11 1
with inputs that act in certain ways to providing 180° phase shift so its Fig.8. Truth table for 4.8. Z, the AND
give the output. The shape of the box function.

AB Z
symbol is that of an amplifier with the
(or the designation within a square circle added.

00 0
box) tells the viewer the function of
the box. TRUTH TABLES
Unfortunately there still exists
more than one conventional way to
Before we discuss more complex
01 1
10 1
gate networks by studying their
draw these symbols. For this course
inter- connection, we need to under-

11 1
we will use those given in Fig. 7, which
stand the concept of a truth table.
are also those used in projects in
This is asimply drawn table that lists
Electronics Today.
the output state for the various
The NOT function bar can be combinations of input states. Fig.9. Truth table for A-1-8=Z, the OR
applied to any function to signify that Rather than write on and off, or function.

POSITIVE LOGIC
NEGATIVE NAND AND ONLOGIC)POSITIVE
it is negated. For instance, an OR such high and low, true or false, it is simpler

LOGIC
as A+B=Z becomes . A-113
- which is to express the two states merely as '0'
called a NOR function. Similarly so and ' 1'. The positive logic convention
A.B is a NAND function.
The OR, AND, NOR and NAND
functions can each have more than
considers a high-voltage level as a ' 1
and the low level as a ' 0'. Fortunately,
today, just about all logic circuits used
ABA -B Z-Z-
0 0 1 1 1 0
two inputs, for example, A+B+C+D-Z. are now in integrated circuit form and
When a function is negated its they 1 rly all work between just two
- 0 1 1 0 1 0
graphical symbol is also altered in levels - which are the same for any
some way to signify this. The devices from a.particular logic family. 1 0 0 1 1 10
convention used is the convention of This provides a compatible arrange-
the addition of a small round circle. If 1 1 0 0 0 1
ment whereby gates and other logic
the circle is at the output the output is system boxes ( that are yet to be
Fig. 10. Truth table showing that negative
negated; if at the input the inputs are introduced) can each be intercoupled logic (or negated positive logic) input to an
negated. The inverter ( that provides without having to worry about OR gate provides NAND output.
12
the output of which is the NAND
function. Hence a negated input OR
gate is a NAND gate. Also, by similar
reasoning, a negated input AND is a
NOR. Put another way, in negative
logic an OR becomes an AND and vice
versa.
Z = ( A.B ) + ( C.D.E ) + ( F.G.H)+ ( 1.J)
UNIVERSAL GATES
Using the basic gates, AND, OR
and NOT, we can build a logic circuit
for any given Boolean expression.
Where there is aplus sign (+) we use an
OR gate, where there is a dot we use Fig. 13. (a) Logic network for function
an AND gate and we use an NOT gate Z=IA.B1 + + +
for those functions that are negated. (b) same logic packaged in IC flatpack. The problem might be to realise a
However it is inteiesting that the Vcc J I HG F NC
logic network that performs this
NAND gate can be used to obtain any logical task — imagine trying to.
desired function. It can be used to describe it in words! Brackets are used
build AND, OR or NOT gates. In other to ensure that sub- connections are
words it is a universal building block, made in the correct way; as in linear
as is the NOR gate also. algebra operations in brackets 2i edei.lt
Thus the majority of gates used in with first as individual units.
modern logic systems are NAND gates The first step in realising the
with the occasional use being made ot network is to form the dot AND
NOR gates and inverters ( NOT) to functions of Z. We need two two- input
minimize complexity. The use of one AND gates and two three- input AND
major form of gate simplifies gates. ( It matters not if a gate has
manufacture and reduces costs. more inputs than needed — the unused
terminal is ignored). The outputs of
FAN OUT these four AND gates are then fed into
the inputs of a four input OR gate so
There exists a finite number of
Fig 13(b). that the function under the negation
circuits that can be safely connected
bar is achieved. At this point we could
to the input, or the output, of logic
select an OR gate followed by an
elements. This number is called the
is given in Fig. 11. In this variation of INVERTER or make use of a NOR
fan- in and fan- out respectively, and
the basic OR gate the output is 1 for gate direct.
gives the number of standard loads
either A or B but not when both are 1 When drawn as a syt-,-m ot
that can be accommodated. Fanouts
simultaneously. Written in Boolean interconnected schematic blocks it
of 10 and 30 are typical load factors.
algebra symbols this gate performs appears as in Fig. 13a. Also given in
A.B+À.B -Z. ( Symbols written as AB Fig. 13b is how a 14 pin dual- in line
EXCLUSIVE OR
imply that a dot exists between them; IC would appear that performs this
One other important gate is a it is common practice to omit the function.
special class of the OR — the exclusive AND dot). As a third example the exercise is
Oh. The logic action of this gate is to devise a logic network that will
seen by studying its truth table which MORE COMPLEX LOGIC add ( in binary system) two binary
inputs producing the binary sum
The exclusive OR gate is more
output plus a carry output. This
complex than the other gates discussed function,called the half- adder, forms
above because it contains more than
A the basis of digital computation with
EXCLUSIVE — OR one basic gate — it is a small logic
binary numbers.
system in itself. Fig. 12 shows how
Z = A.B 4A.B Back in Part 5 the concept of the
tvvo inverters, two AND gates and one
binary number system was introduced
o 1 1 OR gate can be interconnected to
showing that the counting base is 2
achieve the exclusive OR reat.irernc;it.
0 1 instead of the more commonly
1 A second example is given by
encountered 10 of the decimal system.
considering afunction
1 1 o At any digit position in the binary
(Z = A B)
number, the value can be only 0 or 1
Z=1A.B1+1C.D.E1+(F.G.H.)+(1.1)
Fig. 11. Truth table for two input exclusive so addition of two binary numbers
-- OR gate. gives a value at each digit position that
alternates as 0 1 0 1, etc., as counting
progresses. When 0 and 0 are added we
obtain 0; when 0 and 1 are added we
get 1. When 1 and 1 are added we
cannot have 2 in a binary system so it
— _
returns to 0 with a carry of 1going to
A.B + A.B = Z
the next higher digit position. Fig. 14
illustrates this idea — try adding the
two numbers! A half- adder does this
operation for one digit position. The
A.B Fig. 12. Logic network providing exclusive truth table for the half- adder is,
-OR function. therefore, as given in Fig. 15;1.
13
 ELECTRONICS - it's easy!

A 1 0 1 1 0 0 1 Fig. 14. Addition of two binary numbers proceeds

with a carry as for decimal arithmetic but with only
two states 0, 1in each digit.
0 1 0 1 1 1 0
k k k
K ç
1 0 0 0 0 1 1 1
f I CARRY

MOST SIGNIFICANT LEAST SIGNIFICANT (b) IMPLEMENTATION OF

DIGIT M.S.D DIGIT HALF ADDER

RECORD
\it CARRY OF 1
\ TO NEXT HIGHEST
DIGIT
Fig. 15 (b). One form of half- adder logic network.

INPUTS OUTPUTS

A B CARRY RECORD

0 0 0 0
(a) TRUTH TABLE OF
0 1 0 1 HALF- ADDER

1 0 0 1

1 1 1 0

AND CONDITION EXCLUSIVE- OR- CONDITION

Fig. 15. (al Truth table for half-adder logic. Fig. 15 (c). The half adder using NAND gates only.

The sum column shows we need an between Boolean statements. There is development — for that see the reading
exclusive - OR to provide the sum value little point in dwelling on their list. The following relationships are
— hence its importance in computer individual proofs and historical summarized to assist when needed:
design. A carry is to occur when both
A and B appear so an AND gate is de Morgan's rule 1 : =
A.
needed. From these we can develop
one form of the half- adder system — de Morgan's rule 2 :
given in Fig. 15b. Note how the Commutative laws : A+B=B+A
complexity is growing. Such a circuit A.B = B.A
requires around 30 or more passive
and active components and hundreds Associative laws : A.(B.C) = ( A.B).0 A.B.0
of such circuits are needed in adigital A+ ( B + C) = ( A + B) +C=A4 8+C
computing circuit. A version of the Distributive laws : A.(13 + C) = A.B + A.0
same circuit only constructed using A.0 + A.D + B.0 + B.D = ( A + B) . ( C + D)
NAND gates is given in Fig. 15c. Note This is as for linear algebra but with
that NAND gates 1 and 2 have both extra cases:—
inputs tied together, they therefore
perform the NOT function. Try yourC) A + B.0 = ( A + B) . ( A +
BooLean on this as follows-- . and ( A + B) . ( A + C) . ( A + D) = A + B.C.D
Absorption laws: A + ( A.B) = A
SOME LAWS OF BOOLEAN A. ( A + B) = A
ALGEBRA
Double negation : not A = A
When devising systems of logic the
situation soon arises which calls for Universe class laws : A + 1 = 1
knowledge of the rules for A.1 = A
manipulating Boolean expressions. Null class laws : A+ 0 = A
Possible reasons for this may be that a A.0 =
limited range of logic functions are
Complementation laws : A + A = 1
available, so conversion of an
A.A=0
expression is needed, or that a large
expression may not be in its simplest Tautology laws: A + A = A
state. Reduction to its non- redundant A. A = A
state means use of less elements. Expansion laws : ( A + B) . ( A + = A
A number of axioms ( truths based (A.B) + ( A.B) = A
on experience) exist for relationships
14
 A.B.
MINIMIZATION
To save components the network insight into the function of
Z = A.B
first realised by inspection from avalid straightforward gate circuits.

.•
••
•
In the next part we will look at
truth table may well not be in its
practical circuitry of logic gates and
simplest or so-called minimal form. In
simpler cases, apolication of the above introduce several other basic digital
e'lre circuit building blocks. We will then be
Boolean algebra laws by a well
Z = ready to discuss digital systems in
practiced person can often come up
some degree of depth.
with simplifications.
Beware, however, of applying linear
REFERENCES
algebra rules of factoring. It is quite
Most books on digital computer
wrong to cancel or subtract equal
design include a chapter on Boolean
terms in both sides of a Boolean Z = AC)B
algebra and binary arithmetic.
equation.
Unfortunately, no direct way is
"Electronic Computers — Made
known with which to arrive at a
Simple", H. Jacobowitz and L.
minimal network by a routinely j ei 'V: • 1:47•;•;., Basford, W.H. Allen, London,
declared simple procedure. The nearest •
I •
Z = A
— 1967.
we can get to this is by means of a •
•
4 :1:
1 :::
• •efrle
••

Karnaugh mapping procedure which •••••f•4

b • . • "Electronic Instrumentation Funda-
•7•1›...4
we do not discuss in this course as few
mentals" A.P. Malvino — McGraw-
readers will be required to be expert in
Fig.17. Venn diagrams represent logic states Hill, 1967.
this facet of digital electronics. in topological form. Some people find these
An example will show how asimple easier to use than truth- tables. "Numbers" R. Froom, Electronics
system can be minimized by
Today International, July 1973, p.
inspection. Consider the expression Z
84-89
= ( A + B) . ( A + C) . ( A + D). This is chosen) the appropriate area of the
readily seen to be the logic network circles is shaded. The rules are that For the historical development of
given in Fig. 16a. From the inside a complete circle its variable is
distributive laws given above this can computers and other data processing
not negated, outside it is negated.
equipment see
be rewritten as Z = A + B.C.D which Overlapping area of common circles
represents the logic configuration of represents their AND combination. "A Computer Perspective" C and R
Fig. 16b. This minimal form requires The examples given in Fig. 17 Eames, Harvard University Press,
two less gates ( provided a three input illustrate the use of Venn diagrams in
AND gate is available). Massachusetts, 1973.
various simple logic situations. The
concept extends to as many circles,
that is, inputs as are needed.
THE VENN DIAGRAM
In the early days of logical algebra LIMITS OF BOOLEAN
development, John Venn developed a
There are a number of limits to the
system of overlapping circle diagrams
use of Boolean algebra. In the logic
as an alternative way with which to
express the concepts contained in the combination we have considered so
truth table. Venn's diagrams consist of far, there has been no mention of time
or of any feedback around the circuit.
overlapping circles contained in a
rectangular box. Each circle represents In practical systems, time delays always
one of the required number of occur and, further, other elements
independent input variables — A, B, C, such as counters, multivibrators and
etc. If the output variable Z is a 1 memory devices are generally present
(assuming that is the convention whose state depends, not only on the
logical inputs at any given time but, on
what has happened previously!
Boolean algebra is unable to deal with
such situations.
In addition, if a function is
minimized by means of Boolean it
does not follow that the derived
circuit is the cheapest possible. The
minimized circuit may call for 3- input
AND gates, say, but it could well be
cheaper to use the more readily
available NAND gates — even if more
gates are required to achieve the same
function.
Thus it can be seen that Boolean
A
algebra is far from an infallible means
A + B.C.D of arriving at the cheapest possible
solution. In fact it may not give any
D (b) solution at all! Engineering skill and
ingenuity are still the most important
Fig. 16. Logic network realising Z = (A + H).
(A + C). (A +
factors in efficient logic design. It is of
(b) Simplified network. value however, and does give a good
15
 Integrated
circuit forms
of logic functions
economic family to use but needs
more space than the alternatives
developed since. This form of logic
was very much in vogue in the early
+V 1960s but, although still manufactured
ty some companies for replacement
purposes, is an obsolete type not used
in new design.

DIODE-TRANSISTOR LOGIC
- DTL:
This was the next family developed.
The devices of the family use resistors,
Z = A.B.C.
diodes and transistors. Initially DTL
logic was constructed with discrete
components. These designs were then
integrated as shown in Fig. 3. Later
devices used transistor input logic
instead of diodes, thus reducing the
input current requirement and
00 o allowing higher fanouts. Typical noise
immunity ( for a5 V supply level — the
Fig. 1. The implementation of AND and OR oates using diode logic. standard used) is around 1 V. The
delay time for a pulse signal to travel
through, that is, the propagation delay
NOW that we have a basic bias node the circuit becomes a gate.
understanding of switching circuits between input change causing output
Three input AND and OR gates are
change, is around 30 ns. Output is
and the algebra used to shown in Fig. 1. The circuits are the
> 3.5 volts for a ' 1' and < 0.4 -
volts for
mathematically describe logical same; the different class of gate arising a '0'.
operations we can look at the modern because of different bias conditions in
It has a generally lower speed and
methods used ir solid-state circuitry to each case.
lower noise immunity than other
produce the various logic functions in Diode logic is the simplest form of
families. The advantages of DTL are
integrated circuit packages. solid-state logic, it is not available in
the reasonably high fanout of 10 and
integrated form but is often used in
the ease of interfacing or coupling a
LOGIC FAMILIES discrete designs to obtain logic
stage to the TTL family to be
DIODE LOGIC - DL: functions at high, or unusual
considered next.
(compared to standard IC logic)
In part 21 it was explained how a A similar family is HTL ( high
voltage levels.
single diode could be used as a switch threshold logic) which uses 15 V
by altering the bias or the input signal supply lines and zener diodes. This is
level. If more than one diode feeds the RESISTOR-TRANSISTOR useful in situations where high noise
•5V
LOGIC - RTL: levels occur because this logic is more
Logic gating operations can also be immune to noise effects than is DTL.
obtained using transistors acting as •
5V

(i4D switches in various ways. Fig. 2. shows

a typical RTL NOR gate for which
OUTPUT Z = A + B + C. Base current appearing
INPUTS
at A, B or C will cause the respective
A -
transistor to switch to the ON state
150".2
taking the output to ground which is a
Z A ,•8 • C '0' in apositive logic system.
This family was the first to be used
in the now more usual integrated-
circuit form based on the planar
manufacturing technique ( one in
which a mask is used to selectively
diffuse impurities into apure substrate
in order to produce and separate active
device junctions).
RTL is based on a supply of 3.6 V.
Fig. 2. Typical RTL circuit of a NOR gate: Propogation delay is 12 nS for a
any logical ' 1' input will give a logical '0' medium power gate and 40 nS for a Fig. 3. An early integrated circuit a. ,^ign
output. for a NAND gate in diode- transistor knic —
low power gate. It is a reasonably
16 OTL.
 HIGH
THRESHOLD
— 24V

LOW
THRESHOLD
— 4V

Fig. 5. Typical output signal from TTL, note the transient ringing at
each transition.

FIELD-EFFECT TRANSISTORS ( FETs)

OUTPUT Z
n- channel p-cnannel
junction gate (JFET)

Fig. 4. In the TTL (transistor- transistor logic) form of NAND gate

amulti-emitter transistor replaces the diodes of DTL.

three terminal depletion-type

Z AB
insulated gate ( IGFET)

Fig. 6. Emitter- coupled

three terminal depletion- type IGFET.
logic (ECL) operates in a substrate tied to source
non-saturating mode and
is therefore capable of
high-speed operation.
VFE

TRANSISTOR-TRANSISTOR EMITTER- COUPLED LOGIC

LOGIC - TTL: - ECL:
tour terminal depletion-type IGFET
A typical ECL stage is shown in Fig.
This is the most popular logic family
6. As this operates in the linear mode,
in current use. It has higher speed and
that is, without allowing the active
driving capability than DT L. The
devices to go into saturation, it gives
propagation delay is around 12 ns
high-speed 2-3 ns switching times. It,
which is quite fast enough for the
however, needs a moderate power
majority of computing applications.
requirement, is not particularly noise
A typical TTL gate circuit for Z -
immune and needs an extra power four terminal enhancement- type IGFET
A.B.C. is shown in Fig. 4. Note how
supply line. Supply voltages used for
the diode input gate has been replaced
ECL vary but when typical supply rails
with a multi- emitter transistor. This
of 0 V and — 5.2 volts are used the
multi- emitter technique reduces the
output is — 1 volt for the ' 1' state and
input capacitance thus speeding up the
—1.6 volts for the ' 0' state.
switching time, as well as simplifying
Each of the above logic families is
manufacture. In TTL the supply is
based on the use of the transistor five terminal dual-gate depletion-type IGFET
+5 V, the output switching between semiconductor junctions — the
around > 2.4 V for a 1 to < 0.40 V so-called bi-polar technique. Around 02
for 0 ( in positive logic). Fig. 5shows a
1970 ECL emerged as apossible future
TTL signal switching state with time. G1
contender to TTL and at the same
For all of its popularity TTL is not
time another quite different kind of
ideal, especially for the fastest circuit
semiconductor active device became
operation or where the lowest power
freely available — the field effect
consumption is required. Another five terminal dual-gate enhancement- type IGFET
transistor FET. A variation of this is
difficulty with TTL is that switching
the insulated- gate field-effect
transients occur ( see Fig. 5) at the
transistor IGF ET. Fig. 7 lists the
transitions. It is also not particularly
symbols of the basic FET structures
suitable for large-scale integration by
used in logic. This technique is
virtue of the relatively large amount of
manufactured using Fig. 7. The symbols used for the various
space and power required by each gate
metal- oxide- semiconducting materials; types of FET device.
function.
17
 ELECTRONICS - it's easy!
1— DRAIN
SUBSTRATE IP TYPO
v GG ,
CUTI I— SOURCE
current drain of a CMOS inverter is in
the low nano-ampere range as
compared to 0.2 milliamps for a
low- power TTL inverter. At the
system level aCMOS system typically
requires one-twentieth to one-thirtieth
the power required by an equivalent
TTL system. In fact torch batteries
are adequate to run quite complex
CMOS devices.
VIA r,
Because of the complementary
la/ 11,1 (61 configuration CMOS has high
Fig. 8. A comparison of PMOS and CMOS inverters. common- mode noise immunity ( such
as power supply variations). It will
operate with supplies from 3V to
16 V and needs only asingle positive
supply. In addition the high packing
*10 density has allowed the building of
+10V low cost large-scale- integrated ( LSI)
o—
Inp(J t 1
packages. Interfacing to conventional
transistor logic is easy because it has a
Input 1
Q1 02 low output impedance. It generates
low noise because of nicely
Output
Input 2
conditioned rise and fall times. The
fan- out factor of 50 is the highest of
03
all logic because of the extremely high
input impedance. Its speed is better
Input 2
than PMOS but not quite up to that of
TTL — propagation delays being from
04 12 to 60 nanoseconds depending upon
supply voltage used.
2- Input NOR Gate
We have merely glossed over these
2- Input NAND Gate
various kinds of device because we are
mainly concerned here with digital
systems in general. To design systems
r
.. .0'
„ 0y _ -1-
_ . 10 V ' requires little in-depth understanding
- 0 V — I" = * 10 V
1 Input 1 1_ Input 2 Output of the manufacturing method used for
I
--- . Input 1 Input 2 Output
0 0 .1
the actual logic element. The logic IC
I 0 1 0
0 1 1 is merely a black- box with certain
1 0 0
t___ 1
1 0 1 input-output characteristics as stated
1 1 0 on adata sheet.
Why do new families keep emerging?
Fig. 9 (a). A CMOS 2-input NOR gate. The facts are that there is still acost
(b). A CMOS 2-input NAND gate. saving to be had and the market is
huge. The estimated value of the total
market for IC devices in 1980 runs to
around $1,000,000,000. CMOS offered
new horizons in cost savings in
manufacture. As a bonus from the
abbreviated to MOS. Hence, the power requirements of CMOS systems,
THE CMOS FAMILY
MOSF ET is a metal- oxide- the so-often neglected power supply
IC device designers went a step
semiconductor field-effect- transistor. further in the early 70's to produce cost drops remarkably. One example
They can be made in the two published in 1973 gave the comparison
yet another family — the
complementary ways, Por N, of which that a given transistor- based logic
complementary metal oxide
the first is most conventional giving (called bipolar to distinguish it from
semiconductor logic — CMOS. This
the PMOS technique. With MOS) would have apower supply cost
combines both the P and N, MOS
improvements in technology the of 33c per MSI bipolar function
technique in acomplementary manner
NMOS technique is also being used for compared with 2c for CMOS. In fact
to produce a more ideal switching
very complex circuit blocks. one watt will power 50 CMOS devices.
action than PMOS — Fig. 8 illustrates
The attraction to manufacturers and We have all experienced the
the difference between the two. Fig. 9
users of MOSFET devices is that the remarkable increase in the use of
is the schematic of a CMOS two- input
packing density of the active devices is active devices in the last decade or so.
NOR gate. One of the most significant
the highest of all types — much better A transistor radio now costs $1.50 to
advantages of CMOS is its low power
than TTL. It is, however, not as fast as make commercially — and there the
dissipation. Because of the
TTL but adequate for a large part of semiconductors are but asmall part of
extremely-high off- resistance of MOS
the consumer market. FETs have the cost. ( Just one common
transistors, and because only one
extremely high input impedance, thermionic valve costs more than this
transistor of the series- inverter CMOS
1.0 Ti2 ( 1000 000 M S2) is common. today!) A calculator using hundreds of
pair is ever on at the one time, the dc
18 devices can be bought for $5.
 P positive supply voltage
LOGIC DIAGRAMS AND TERMINAL CONNECTIONS FOR SN7400 SERIES
0 = earth

FlIGIRIFIRIfil
us Q. Q. Q. QS 0 .

F.P1111/71.00

C
ia Ga 4, Oa 0
Basic gate circuit W WW 15) UI Le 12.1

,CLEil PAM girl PUB

Oa, 0, 0, Oc 0 0. 0, 0,
P O G, ne Oa 4, Oa
ne G Ca, n nc
D Fmoocc,
G,
FM 111/7420 P1M10111.30
o C. I, I, P I I
2 G2
o 6 Ga G nc G 15 0, Or 0144 On040
..
d. G
5 G..
s Gs
10 Gsj o
Gi 0. P. Gn 01

i3 G0 F1.1141/ 54.0

0, 0 nc G, G 0
‘';
n, .1 et e,

Fig. 10. The basic logic gates of the TTL family.

Minicomputers and microprocessors

are rocketing down in price — tens of
2N5459
thousands of elements for a few
hundred dollars. It might be argued C6 CA3130
10p 401
that reducing the price will soon 400 RESET
STROBE
reduce the makers' total income but as 4016
VIEW FROM o
prices fall, applications for digital ABOVE
circuitry widen at an even faster rate, CB
thereby keeping up an expanding 330p
demand.

THE DUAL- IN- LINE D4

1N914
PACKAGE
The most commonly used form of IC 0V
logic package in small batch
production is the dual- in- line
arrangement with 8, 14 or 16 pins
Large-scale integration LSI used by
specialist manufacturers will vary in
number of connection, but systems C7
based on these require very large -0CK 330p
volume sales to make a large special
system economic. Thus LSI chips are
largely restricted to computers or very
high volume things such as cPlculators
o
and digital clocks. OVERFLOW
The number of connections decides
Fig. 11. In aschematic diagram the individual gates in acommon IC package are drawn to
the available combination of functions
suit the system layout — not the actual wiring layout. This is illustrated above by part of a
or inputs and outputs. Assuming a digital voltmeter circuit.
need for two power- supply
other basic digital-system building
terminations, of which one is the as a more powerful dual four- input
blocks. These are the flip-flop, ( more
common for all functions, a 14 pin NAND buffer and others to be
correctly called the bistable), the
device will have 12 pins available to discussed below. Each gate function is
monostable, the astable and the
produce various combinations of a quite separate entity on the
Schmidtt trigger. Let us look at each
input. Fig. 10 shows the main units substrate; when schematic circuits are
drawn the individual gates can appear in turn.
that are marketed. Available are the
sextuple inverter ( 6 inverters with one anywhere on the system schematic, as
shown in Fig. 11, where part of a THE FLIP-FLOP
input and one output each); the quad
2 input NAND ( four two- input NAND digital-voltmeter circuit is given. Fig. Gates are used to perform logical
gates); the quad two- input NOR, triple 12 shows a component overlay and arithmetic, such as, allow event A to
three- input NAND, dual four- input actual circuit boards for the same occur when B or C have operated.
voltmeter. Digital systems can be greatly
NAND, single eight- input NAND plus
other more special combinations such As well as gates there are several enhanced by the addition of blocks
19
 ELECTRONICS - it's easy!
9 15V cic

C11

INPUT

COMMON

Fig. 12 Wiring layout and actual circuit boards of the DVM shown in Fig. 11.

that remember and count the digital In essence the characteristic behaviour effort was expended to provide afast,
numbers. It is quite possible to use the is that one stage will be held ON when reliable flip-flop action. Today a
gates we have discussed to form the other is OFF, or vice versa. To typical IC equivalent, see Fig. 14, uses
counting stages but it is more reverse the situation a pulse or change many active elements for less cost than
economical to build specific circuitry in input signal level ( which can be two discrete transistors. As discrete
that will count pulses and/or store applied to either the base, emitter or designs play no real part today we will
values in a binary form The basic the collector) will cause the system to only give the characteristics of the
block of modern counting technique is toggle over to the other state. Because flip-flop, not its design details.
the flip-flop — FF or bistable — one of of the heavy degree of positive
the multivibrator family. Flip-flops provide a counting action
feedback provided, the circuit does not because each pulse at its input causes
The most elementary flip-flop circuit dwell in the in-between state. State the system to switch over one state —
is repeated from Part 18 as Fig. 13. It change is very rapid — nanoseconds.in
the output, therefore, switches state at
consists of two-transistors and several well designed IC flip-flops.
each second input pulse providing a
resistors which are dc coupled in such A decade ago flip-flops were built
divide by two action. The output is a
a way that each stage provides asignal from two, or in the faster toggling switched level which can then be used
which controls the state of the other. circuits, four transistors. Considerable
to pulse a following stage dividing by
two again and so through cascaded
fl ip-flops.
The most commonly used flip-flop
symbols are given in Fig. 15. Outputs
are denoted by the symbol O. A flip-
flop has two outputs, one of which is
the inverted value of the other, that is,
O and Q. The pulsing input is denoted
T for trigger. As well as these
connections we need a set and reset
input denoted S,R. ( Although often
only R is provided). These enable the
flip-flop to be set up on demand with
the output Q set to either 0 or 1state
as is needed. This is essential firstly
because a flip-flop can come up in any
state when the power is energised, and
secondly because it may be necessary
Fig. 13. A basic flip-flop built with discrete components. to set counter stages to a given binary
number.
20
 FLIP FLOPS DIP ITOP VIEW) SCHEMATIC DIAGRAM

R-S ( set- reset) flip-flop

R17719 il7117 1

sr,
Co
Oep o

"

.2s

FF

Fig. 14. Schematic and

logic diagrams of an
integrated circuit flip-
Toggle flip-flop flop using bipolar
devices.
Component values shown are tVO cal

LOGIC DIAGRAM

o à

PRESET CLEAR
(RD)
(SD'

K1

K2

K3
J- K flip-flop
CLOCK
I FF
ICPI

Some flip-flops provided in IC form saw in Part 18, an output pulse having
also may contain three- input AND a designed length and height which
gates that feed the set and reset inputs. remains the same irrespective of the
These are known as the Jand K inputs input pulse shape. It, therefore, finds
giving the name J- K flip-flop. The JK application as a pulse reshaper. As the
flip-flop overcomes an ambiguous duration is fixed it can also be used to
FF
output state, when both inputs are generate a pulse that is delayed from
high, that occurs with the RS flip-flop. the triggering pulse by the length of
We will see, when counting systems are the monostable pulse. Nlonostables are
discussed, why the reset ability is vital available in IC form, and can provide
in many counters. Fig. 16 gives the pulses of duration from 20 ns upwards
D-type flip -flop schematics of some of the available to minutes or more by appropriate
choice of values.
fl ip-flop I
Cs.
FF
Dc

THE MONOSTABLE THE ASTABLE

If one of the bias resistors of the If both feedback paths use capacitive
bi-stable circuit is replaced by a coupling the circuit becomes
capacitor, as shown in Fig. 17, the self-toggling with the stages alternating
in state without being externally
circuit provides adifferent action from
the normal flip-flop. When triggered driven. We considered this circuit in
the circuit changes state, but only stays Part 18 when discussing signal
D toggled in the changed state for atime generation. The astable is important in
FF decided by the product of the values digital systems for it provides the
CLK 6
- — of the capacitor and its "charging" square wave signal that increments the
resistor R. ( T = 0.7RC). Hence a pulse digital system along pulse by pulse. It
input will cause the output to change acts as the 'clock' regulating adigital
state and remain there for a chosen system's sequential operations.
time interval before it triggers back to Astables are not usually produced
the original state. directly in IC form, for the same action
The monostable, also called a is obtainable with other elements, for
Fig. 15. Schematic symbols used for
one-shot or single-shot provides, as we instance, with the next element to be
different types of flip-flops.
21
 1, 0, TT 1
1,
0 P s t It, K K 0, P S 0 T, S, 0, 0,
F.11121/7473
3,111/7473 Fuj131/7474
T, S, 0, P T, 5,
0, 0 5 0, T, 5, a,

Si f K, K, K O. 11-
Pm101//4.70

.r. 52 Ii h 3 0, 0

considered, th q Schmidtt trigger. Fig.

F./1141/74.71

1), T, P 0, O G,
18a shows a clock source based on a
UrCI-U=T: Schmidtt trigger IC. Fig. 18b shows
one based on two NAND gates.
• e

o
THE SCHMIDTT TRIGGER
Fig. 16. A variety of flip-flops This unit, also introduced in Part 18,
are available in integrated form. toggles over from one state to the
other at acertain input voltage level. It
remains in the opposite state until the
voltage falls below the threshold level.
The Schmidtt trigger is used to
produce digital signals from analogue
signals providing the two necessary
binary levels at the output which
indicates whether the analogue signal
is above or below the threshold and,
ou I"PUT PULSE
which are compatible with the rest of
OF 20pS LENGTH the digital system.

_FL The Schmidtt function is available in

IC form as a dual- in- line pack. It has a
four input AND gate feeding the
actual trigger circuit and is buffered
with an inverter. The Schmidtt trigger
is readily identified in Fig. 18. Its
preferred symbol is given in Fig. 19.

YOUR LIBRARY
There are a bewildering number of
digital ICs and to identify them
correctly it is wise to have a good
Fig. 17. A basic monostable constructed with discrete components.
range of manufacturers' catalogues and
application notes.

REFERENCES
I.C. SCHMIDTT TRIGGER The reading list given in Part 22 is
,5V
PACKAGE also relevant to this part.

Fig. 18(a) LEFT: A Schmidt trigger com-

O bined with a transistor provides an astable
multivibrator action.
Fig. 18(b) BELOW: Astable action may
also be produced with two NAND gates.
2200

+5V BC 108 V\AA

47k ,

BC 208
2N 3565 CLOCK RATE
IírC1
470pF

1
11 3 12
2
1 11
2 CLOCK OUT
330pF C2
47012
4700F

1
22
 BASIC DIGITAL ELECTRONIC TERMS...
test yourself on these. Each is explained in this chapter.

AND- gate Circuit that performs logical AND operation.

Bi -
stable ( function) Two state circuit that is used for binary counting.

Boolean algebra Mathematical algebra for logic operations.

Boolean statement Mathematical way of writing alogic situation.

Clock Source that causes adigital circuit to step on bit by bit.

Digital circuit Electronic circuit that processes information in digital form.

Digital IC Integrated circuit unit using digital techniques of operation.

Exclusive-
CR gate Special kind of CR gate: basic to digital computing.

Fan- in Number of inputs that adigital circuit can have and operate correctly.

Fan- out Number of output circuits that can be driven by adigital stage without
altering operation.

Flip-flop Same meaning as bi-stable but more generally used.

Gate Any circuit that performs logical algebra AND, OR, NAND, NOR function.

Inverter Circuit for changing state of alogic level to its negated form.

Indicator Device used to show logic state.

Karnaugh map Matrix representation of logical statement; used in minimising alogic circuit.

Logic ( circuit) Collective circuitry for performing logic operations.

Minimization Process of reducing degree of redundancy in aprototype logic system.

Mono- stable Circuit for generating aknown interval; square pulse; used to regenerate
digital signals or to provide adelay.

Multi-vibrator A general name for the family of bistable, monostable and clock circuits.
Also used to describe the clock source itself.

NAND- gate Circuit that performs logical NAND function.

Negative logic Circuit in which logical 1is represented by zero or negative polarity voltage.

NOR- gate Circuit that performs logical NOR function.

Positive logic Circuit in which logical 1is represented by the positive polarity signal;
usually 5 V in TTL.

Redundancy Part of logic circuit that could be removed without altering logic operation.
Often used to ensure total reliability — sometimes there by accident!

Schmidtt trigger Circuit used to provide digital signals from analogue levels. It changes state
at apredefined constant level.

State of circuit In digital terms, the logic level at apoint in the circuit.

Truth table Matrix table displaying states of logic circuit in terms of input and output
states.

Venn diagram Overlapping circles drawn to represent alogical situation.

23
 Digital

24 sub systems
counters and
shift registers
N THE preceding articles of this series We have already become involved in interrogation period) the outputs A. B,
we have described the basic building small systems — the exclusive OR and C, etc. will either be at a 0 or 1, as
blocks of digital systems. To the half- adder of Part 22, for instance. shown in the truth table. For example,
summarize, there are the various gates, The next step to take is to form a count of five ( decimal) will be
the flip-flop, the monostable, the sub- systems with the fundamental registered at 101 for CBA respectively.
Schmidtt trigger, the inverter and the blocks that provide us with the facility Note that the truth table appears to be
square-wave clock source — a to form and manipulate digital written back to front — the reason is
surprisingly few basic elements from numbers, because many ( but not all) simply that we write numbers ( by
which all the countless different forms digital systems operate with numbers convention) with the most significant
of digital equipment are constructed.
either to provide means of calculation, digit to the left hand of the number
We are now in a position to examine or to provide a display of numbers. and this corresponds with the furthest
how digital signal systems are put Thus we need to know something of right-hand flip-flop, its position on the
together using these basic building digital counters and the somewhat schematic arising from the drawing
blocks. similar units known as registers. convention used for the signal- flow
DIGITAL NUMBERS INTO DIGITAL We saw in the previous part how through information systems. Thus a
SIGNALS — COUNTER SUB- the flip-flop provides acounting action stream of input pulses with time are
SYSTEMS by virtue of its ability to switch states converted into a multi- element digital
Not long ago digital systems were for each pulse appearing at its input. number. This form of input is often
invariably built up from individual However nowadays it is a level change referred to asá " crazy-digital" number
blocks where each of the above rather than a complete pulse which system when applied to systems
functio: scould be clearly identified in causes the transition. incorporating measuring sensors. Such
the system. But not so now. Many of sensors generate pulses not having any
t.le blocks now marketed as basic BINARY COUNTING clearly obvious relationship with time
building elements are complex systems Cascaded flip-flops, such as shown — examples are digital- position sensors
in themselves. The most extreme in Fig. 1, form the simplest type of wherein a pulse is generated for each
example is probably the micro- counting system. Each time the input unit displacement occurring.
processor system ( it provides the bulk changes state, the A flip-flop toggles Clearly, if the state of A, B, C: etc.
logic requirement of a powerful back and forth delivering a state is to represent a number the count
computing system) which is now
change to flip-flop B for each second must start from some clearly defined
available as a 'throwaway' element for input change — and each fourth initial condition for each stage —
around $ 10 or less. It would take to change to C and eighth change to D often, but by no means always, stages
the end of this course to begin to and so on. are reset to 000 — by applying a level
appreciate the complete system — At any instant ( where the input is to the reset R line. In some
operation of such abuilding block!
presumed stationary, at least for the applications the number must begin at

T
LAMPS
INDICATOR
COUNT
IN
CLOCK CLOCK CLOCK
IR i
R INPUT CBA
RESET LINE 0 0 0 0
1 0 0 1
AINPUT 2 0 1 0
3 0 1 1

A
OUTPUT 4 1 0 0
5 1 0 1

c
71 r---- B
OUTPUT 6
7
1
1
1
1
0
1
8 = 0
C
OUTPUT
0 0 0
67 TRUTH TABLE
TIME
Fig. 1. The asynchronous or ripple- through binary counter is the simplest to implement with flip-flops.
24
 a specific value. This is achieved by
ELECTRONICS - it's easy! operating the preset lines accordingly
for each stage, the value being
arranged, say, by the setting of
numeric dial switches usually oper-
ating through solid-state gates. Once
PRE-SET
these are set, a single command signal
ENABLE VALUES 0OR 1AS
LINE can preset the whole counter to that
REQUIRED FROM
chosen number. Figure 2 shows how
SWITCH OR
PERMANENT this is achieved using gates.
CONNECTION It is often desirable to display the
digital value for visual interrogation.
An indicator lamp ( light emitting
diodes LEDs are now used) driven
from the A, B, C, etc., will show the
positive logic binary number. Driven
from A, B, C, etc., it would show the
negative logic number.
RESET ( TO 0) Waveforms for this form of counter
LINE (called the ripple- through type) are
Fig. 2. Counting stages require their initial conditions be set — either to Os or to desired shown in Fig. 1. Although we refer to
values. pulses flowing through, the waveforms
are actually square- wave trains in the
dynamic state, or levels in the static
state, that have frequencies divided
down in the ratio 1:2:4:8 etc. This
'pulse' form of expression is a
hangover from the early days of digital
technique ( just a decade ago) where
actual pulses, not levels, were used to
trigger flip-flops. Each square- wave
transition was differentiated by a
CLOCK CLOCK
--- CLOCK simple RC circuit to provide a pulse of
energy that triggered the next stage.
Today, this method is unnecessary and
COUNT IN is seldom used.
RESET
ASYNCHRONOUS AND SYNCH-
Fig. 3. Synchronous binary up -
counting with JK flip -
flops. RONOUS COUNTING
Once an input pulse has occurred, it
sets the chain in action, each flip-flop
passing each second input pulse on, to
the next stage. As each stage has a
finite delay time — nanoseconds with
TTL, microseconds in older forms of
logic — each stage triggers at a later
time than the one before it. Hence the
COUNT
form of action of each stage is said to
be asynchronous with the others. Thus
whilst the pulse is rippling through the
RESET counter the outputs are in an
undefined and changing state between
the previous and next correct states.
Fig. 4. Reverse counting is simple — use complement outputs instead to trigger the
The output cannot therefore be read
next stage. (al Ripple- through count- down. until the whole thing has settled.
It is quite common to have binary
counters with over 20 stages — at
100 ns delay in each, the maximum
input pulse counting rate ( if the
outputs are to be used whilst it is in a
counting transition mode) would be
limited to 2ps between incoming
pulses, that is, 500 kHz. Faster logic is
CLOCK CLOCK CLOCK available that provides around 20 ns
delay but this still seriously restricts
the data- transmission rate where the
ripple- through design is used.
COUNT • This disadvantage can be overcome
RESET by increasing the circuit complexity
somewhat to form what is known as a
SYNCHRONOUS counter. Each stage
in the cascaded chain is fed a clock
Fig. 4(b) Synchronous count- down.
25
 pulse simultaneously via control gates.
The control gates for each stage also
receive inputs from all previous stages
such that the bcirticular stage only UP LINE
operates at the correct count.
AND qatrs
By this means all stages operate
synchronously and the propogation
delay is reduced to that for a single
COUNT
stage only. Synchronous counters are
essential where the outputs of all
stages must be decoded in parallel, eg,
where a display of the count is
RESET
required. However for straight
DOWN LI
frequency division applications, where
the output is taken only from the last EITHER UP OR DOWN LINE
ENERGISED AT ONE TIME —
stage, a ripple counter is normally
NEVER TOGETHER
faster than the synchronous type.
The logic-gate inputs of the JK
Fig. 5. Reversible counting uses fast switches to select which output of each stage feeds the
flip-flop allow them to be connected next. (a) Asynchronous — line controlled.
for synchronous counting as shown in
Fig. 3. It is not important to know
how JK flip-flops work internally for
counters are now built by cascading
ICs as per application note B OUTPUTS
instructions — it takes a specific type
of mind to realise digital counting
systems without effort! Fortunately
for those of us without this ability it is
rarely needed except, of course, by IC CLOCK CLOCK CLOCK
designers.

UP, DOWN AND REVERSIBLE

COUNT
COUNTERS
RESET
So far we have only looked at
counters that increase up the binary COUNT DIRECTION 1DO‘NN 0 UP
number scale for each additional
input. To make the same system count Fig. 5(b) Synchronous — line controlled.
down is incredibly simple — we merely
re-connect them so that the
complemented outputs are fed to the
next stage -- see Fig. 4 — instead of
using the normal output, that is, feed
A, B, C, etc. to the count inputs. It
will then count down in binary
INPUT 0 OR 1
sequence. Intuitively we would expect • •
this because of the two- state
complementary nature of binary CLOCK CLOCK CLOCK
numbers. If you are worried about

CLOCK N
numbers passing through zero, try
your digital arithmetic on a
count-down case starting at 000.
In many applications needing
RESET
counters, one-way counting is
Fig. 6. The shift- register is also built from flip-flops but with different connections
satisfactory. Examples that come to to counters. (a) JK type
mind are nucleonic pulse-event
counters, counter- timer units and
counts of objects passing agiven point.
In some requirements, however, the
need is to add or subtract pulses to
provide at any time the instantaneous
sum or difference between two inputs,
or to give the nett value from a single
measurement parameter that alternates
in sign. Examples here are digital-
position indicators where the direction
of movement reverses, integration of
reversible variables such as the flow of
solid or liquid past apoint the number
of vehicles in acar park, and situations
where the difference between two
pulse-train variables is needed.
Fig 6(h) D typp
Several methods may be used to
26
ELECTRONICS - it's easy!
OUTPUTS ( PARALLEL)

SHI FT
1 0 1 = 5 ( decimal)
REGISTER
x 1 0 O. = 4 ( decimal) INPUT
0 0 0 0 0 0 INITIAL
(SERIAL) RESET TO 0
r‘ ,ISD LSD

o o
= 20 ( decimal)
1 0 1 0 0

(a) multiplication —OP- TIME

11 1 (7)
FEED THIS INTO
1111 0 1 0 (21) INPUT TO OBTAIN
(3) — 1 111 THIS AFTER SIX
INCREMENTS
l o o
—1 1
jp_1(L) = 111 in binary o
11
0 11
Fig. 8. Registers can provide serial to parallel word conversion, the
— 11 reverse also applies.
(b) division 21 = 7 in decimal
o 3

also possible to design counters that

Fig. 7. Binary multiplication and division follows the same rules as will accept a new pulse while the
decimal numbers but note how simple multiplication becomes — a system is still in transition. This is
process of shifting and adding 0 or 1. done by ' holding' the pulse ( by
applying a delay) until the counter is
ready to accept it.

SHIFT REGISTERS
Although not a counter in the same
sense as above, the shift register also
consists of a cascaded chain of
flip-flops but with different connect-
ions. The purpose of the register is to
hold a binary number but allow it to
be shifted as a whole to the left
(toward the most significant digit —
called a forward shift register) or to
the right ( toward the least significant
digit — the reverse shift register) one
step for each input pulse.
Fig. 9. Schematics of IC units. (a) Binary stages for counting or other purposes. (b) 8-bit shift
As shown in Fig. 6, the
register.
incrementing signal, which is more
etc. paths. Switching is accomplished usually a free- running clock signal than
construct up-down or reversible
using logic gates as shown in Fig. 5a a one- at- any-time instruction, feeds
counters. The most common method
and 5b — these provide adequately fast the count inputs causing all stages to
is to use a common pulse-count input
switching. In the ripple-through toggle in synchronism. The state of
that accepts both ' up' and 'down'
variety the direction- line commands each following stage, being tied to the
pulses, the decision to add or subtract
must be held stationary until the output of that preceding, goes to that
each individual pulse being decided by
counter stages have settled in order to of the one before with each clock
the simultaneous voltage levels applied
preserve accurate counting. The delay pulse. Thus a number can be fed into
to control lines. The control lines
is less pronounced in synchronous one end in serial fashion and will be
select whether the pulse is routO
designs. Within certain limitations it is caused to pass through the register.
through the A, B, C, etc., or Â, B, C,
The whole is cleared to zeros or reset
to any desired value via the reset input
line. Using D- type counters outputs A
etc., go to D inputs. Whereas when
using J- K counters A, etc. go to J
inputs.

1_1 Li D
CARRY Registers perform
functions in digital systems. Firstly a
three main

COUNT digital number can be delayed in time

CLOCK CLOCK CLOCK CLOCK
by the additive propagation time
(divided by the clock frequency) of
the number of stages it passes through,
or stored indefinitely ( provided the
power is held on).
RESET Secondly, one digital number can
be successively offered up to another
Fig. 10. Four JK flip-flops connected to count as anormal binary coded decimal NBCD stage. for digital summation of the two — a
27
 basic step of multiplication by digital counters three flip-flops are needed
means. One number remains in with three wasted states.
position, the other increments across, Although a decade counter uses
the two being added at each step by only 10 out of the possible 16 states,
the half- adder ( part 22). Binary of its four flip-flops, this is not
multiplication and division follow necessarily as wasteful as it may at
decimal number procedures but are first appear. For example if we wish to
much simpler — see Fig. 7. count up to 9999 with decade
A third use for shift registers is for counters we will need a total of four
the conversion of digital data from counters containing 16 flip-flops. To
serial to parallel form and vice versa. count to the same number with
For example it may require agroup of straight binary counters we need 14
eight pulses to define the position of a Fig. 1. Ready-made IC decade counter
(SN 7490) flip-flops — only two less. Additionally
shaft with sufficient accuracy. The the ease of obtaining an easily
group of eight pulses is called a word interpretable display results in a
and each individual pulse is called a states to describe the individual system cost that is less than if an
bit. To transmit this word we would numbers. Similarly three states are all- binary system were to be used.
normally need eight lines. However a required for a ternary system, five for In computers, information handling
single line may be used if the quinary, eight for octal and so on. Ten capacity is at a premium and the need
information is transmitted one bit at a is not the limit — we use 12s and 60s to display the internal numbers
time, with synchronizing pulses to tell in time and angle subdivision. negligible. In such cases the
the receiving equipment where each From the hardware realisation octal- number system comes into its
word starts and ends. If the data when viewpoint, binary numbers are the own because three flip-flops provide
received is fed into a shift register simplest to hold because of the eight states without any waste of
serially, eight bits at atime, each word existence of the two-stage flip-flop. states, and without need for the extra
will appear in parallel at the output of (Three-state devices exist but have not components required to skip un-
the shift register, providing synch- gained favour). From the user's wanted states. The octal range is 0, 1,
ronism is maintained, where it may be viewpoint, however, we are more 2, 3, 4, 5, 6, 7 and then back to O. A
decoded. familiar with decimal numbers — afew number 312 ( octal) is 3.8 2 + 1.8 1 +
In reverse, a register can be set up people can read and work with binary
2.8° = 3.64 + 1.8 + 1 = 202 decimal.
to the desired number by the displays but more would agree that Note that the decimal number requires
appropriate choice of stage inputs. values like 1025 decimal, which when roughly the same numper of bit
Once set the register is incremented to displayed in binary form required 10 positions as the octal number but to
feed out the number in serial manner bits ( 1000000001), is difficult to read implement decimal in digital hardware
to e single line. and interpret as amagnitude. would need four ( compared with three
Many forms of binary counters and Flip-flops ideally count in powers for octal) flip-flops for each bit
registers are available in integrated. of two -- 2, 4, 8, 16, etc. — so any position. However, where output is
circuit packages. Figure 9a shows the other value such as three or 10 needed for human use — printouts and
schematic of a binary counter having requires modification to the counting readout in numbers — the decimal
two bits on the chip — that is it procedure. system is best.
contains two flip-flops ready to be The flip-flop chain must be made to
connected to count. Figure 9b is an skip the required number of unwanted DECIMAL COUNTING
eight- bit shift register. Both can be states in the truth table in order to To obtain the 10 states 0 to 9 we
cascaded with like units to extend the return to the zero position. It is must begin with enough flip-flops to
bit capacity to virtually any length of probably evident that this implies a
provide them. Decimal or decade-
binary word needed. waste of counting capacity when states counting stages, therefore, need four
are not used. In 10s counting, four flip-flops which count in some form of
COUNTING IN OTHER THAN
flip-flops are needed to create 10 code over just 10 states. The most
BINARY
states — the other six of the 16
In the binary counting system each straight-forward realisation is to let
possibles go unused. Similarly, with
bit position requires acounter element them count through the normal binary
three's ( called modulo- 3) two code and to apply interconnections
that has two stable states. In the flip-flops are needed wasting one of
decimal counter each digit needs 10 between stages which prevent
four states, and with modulo- 5
illegitimate states occurring and, often

CLOCK
CLOCK
CLOCK CLOCK

COUNT
RESET

COUNT DIRECTION

Fig. 12. Schematic of reversible NBCD counters age.

28
ELECTRONICS -it's easy!

.10V

.5V

*10V
R33 R42
R105
6k8
*
' LED 1ON
DISPLAY BC
R63
10k 07 • 15k
R33

818
19k

R21
2M2 C6
____11150pF

R22
27k

(a) (
h)

Fig. 13. In use IC counters are used as black-boxes. (a) How it appears in a circuit schematic. (b) How it appears on a circuit board.

• h

AND
INVERTER
GATE

• • •

D TYPE FLIP FLOP

(:)_:f
CIRCUITRY
ON ALL
PROTECTION
INPUTS
Fig. 14. Schematic of the CMOS 4017
decade counter and decoder IC showing
the circuitry associated with each logic
symbol. This IC contains over 200
transistors for about $ 1.50.
29
 best produced by this subdivision
process.
As four flip-flops will provide up to
16 states they can also divide by 12 by
the use of stage interconnections.
o Figure 15 is the schematic of a divide
by 12 counter. These are used in
timing systems to provide two
seconds, minutes and hours units. A
divide by five plus divide by two is
also available in the same IC.
Just twelve years ago my technician
and Ibuilt one of the fastest up- down
decade counters reported at the time.
Fig. 15. Schematic of adivide by 12 counter the TTL 7492. It used discrete components, it took
several months to build, could reverse
at about 400 kHz, cost about S300 in
equally important, prevent the system are synchronised, not just individual components alone, needed a shoe- box
locking up ( as can happen in some stage flip-flops. The practical catch is, sizé container and a hefty power
counting systems). however, that synchronous designs supply. Today a match- box size unit,
Decade counters using straight require many extra connections as the including a battery, virtually in-
binary are denoted Binary Coded number of stages increases. destructible in normal use, can be re-
Decimal — BCD. As there is no real Reversible decade counters are built versible at ( at least) 10 MHz rates and is
reason why a stage should start with in a similar fashion to the binary types available 'off the shelf'. It costs amere
decimal 0 and binary 0 coinciding, it is — using A or À etc., as needed. The few dollars and is vastly more reliable.
feasible to construct a large number of problem is complicated by the need to We have reached the point where the
different BCD counter stages using lose states but once again few people mechanicals — the knobs, dials, case
different code sequences. If the zeros would be called upon to design one and boards cost more than the
do coincide it is called normal BCD from scratch as single — IC systems are electronics circuitry. At the time of
(NBCD). marketed ready-made. Figure 12 printing, $30 buys a 100 step pro-
As we shall see in the next part, it is shows the schematic of a reversible grammable pocket-size calculator.
important to know that BCD code is synchronous NBCD counter.
used when the system needs a display REFERENCES
in decimal. This is because the USING COUNTERS IN PRACTICAL An excellent inexpensive book worth
circuitry of the decoder depends upon CIRCUITS purchasing is: " Digital Instruments
the logic sequence of the counter as -
Application Notes explain the Course — Part 1, Basic Binary Theory
well as the requirements of the display
connections and any special conditions and Logic Circuits" — A.J. Bouwens,
method — more of that later.
to be observed in using IC counters. N.V. Philips Gloeilampenfabrieken,
At this stage design becomes a
The electronic designer today regards Eindhoven, Netherlands, 1974. ( Avail-
matter for the expert but fortunately
the appropriate IC as ablack-box with able from Philips offices).
ICs are available with the whole
pins which are wired accordingly — This gives a little more depth than
counter and BCD unit constructed
what is inside is of little consequence. this course can allocate and is recom-
ready to act as one decade. To obtain
Figure 13a shows how an IC counter is mended for those people who are
multiple-position decade numbers we
represented in a circuit diagram and involved with building digital systems
merely cascade decade counter ICs in
Figure 13b shows how it is wired onto routinely.
accordance with manufacturers' data
a printed- circuit board. The internal An excellent book for basic
sheets. Figure 10 shows how four JK
complexity is seen in the circuit digital circuits is " Pulse, Digital and
flip-flops can be connected to provide
schematic given in Figure 14. Switching Waveforms" by J. Millman
NBCD counting. Note the need for an
and H. Taub, McGraw-Hill, New York,
additional NAND gate and inverter.
1965.
Comparing costs, it would usually be OTHER COUNTERS
Although now several years old, a
illogical to build acounter this way for
The register becomes a counter by useful work is " Digital Computer and
the cost of ICs to build such asystem
joining the output to the input to Control Engineering" by R.S. Ledley,
is higher than that for a ready-made
form a ring- counter. This system McGraw-Hill, New York, 1960.
decade stage such as that shown in passes a pattern around the loop one Instruments are discussed in " Digital
Figure 11.
step at a time for each input pulse Instruments" by K.J. Dean, Chapman
Many applications require visual and Hall, London, 1965.
Hence the logical state of the elements
display with each decade: it is now Counting circuits are dealt with in
at any time represents the number of
possible to purchase an IC with count "Electronic Counting Circuits" by J.B.
counts accepted. This is also a
and decode functions combined in the Dance, Iliffe Books, London, 1967
convenient way to recirculate adigital
one chip. and in " Introduction to Counting
word which needs constant re- use. A
The question of synchronous versus Circuits and Transistor Circuit Logic"
faster version is the twisted
asynchronous counting in decimal ring-counter. Chapman and Hall, London, 1964.
stages still applies. As there are,
Counters are also used frequently as Logic is the subject of " Binary
however, only four stages in each Arithmetic and Boolean Algebra" by
frequency dividers. For example, a
decade, the ripple- through time is not A.C. Gillie, McGraw-Hill, New York,
1MHz clock source passed through
usually as vital an issue as with binary 1965.
one BCD decade provides a source at
counting. However, this time must be More recent books are " Digital
100 kHz; through two decade counters
compounded with the ripple- through Electronics for Scientists" by Malm-
10 kHz and so on. As it is more
time for all decades — 12 or more are
convenient to provide stable high stadt and Enke, VV.A.Benjamin Inc,
common on calculators. The fastest
frequencies than stable low frequencies, 1969 and " Digital Electronics" by
decade system is one where all decades
precise low- frequency pulse trains are Price Ward, TAB Books Inc., 1972.
30
 BASIC TERMS USED IN DIGITAL COUNTING OPERATIONS
Asynchronous counter Type of counting system in which the count passes through the circuit
stages in sequence.
Batching counter Counter used to provide control signals at preset numbers of counts. They
may count up from zero or down from apreset value to give asignal at zero.
BCD code Abbreviation for binary- coded-decimal; is acode which counts in binary
for the sequence of one decade of ten, each unit operating as adistinct decade.
Binary counting Counting in binary number form.
Binary number Number expressed in the two- state, binary, form.
Bit The name given to the least size number unit of abinary number
Bus Circuit path which conveys signals from one stage to others; acommon
interconnecting link.
Carry The number that results, to be carried over to the next higher number
stage, when the stage reaches the capacity limit of the stage.
Clock input Terminal to which the clock signal is applied.
Control line Line along which control signals are sent to actuate acircuit, such as a
counter stage, when logic desires it.
Crazy digital system Counting system in which the counter fills or empties as pulses are produced
by an external source. No clock is used to pace the counter on.
Decimal code Counting code based on decimal numbers.
Digital counter Counter that totals digital signals. As all numbers are discrete entities, all
counters must really be digital in form.
Direction line Circuit line along which control signals are set in abi-directional counter
to set the counting direction.
Down counter Counter that has its flip-flops connected so that each pulse input causes the
number to count down.
Gray code Digital number sequence in which each step causes only one digit to change
each time. Many alternatives exist. They are used to reduce the risk of error.
Latch Facility often incorporated into counters that enables the number to be
frozen whilst the counter continues to count on.
NBCD code BCD code that starts with its zero at binary zero. Counting on in abinary
sequence from there.
Octal number Number with base of eight.
1.2.4.8. code Alternative name for normal binary number code, the ratios representing
the weights that each stage of afour stage counter will have.
Overflow Carry quantity that has no place to go and is therefore lost. An indicator
of this condition is often provided.
Pulse The basic entity that feeds acounter. Originally they were short duration,
fast rise-time pulses but today they are more likely to be afast level trans-
ition going up or down in polarity.
Pulse event counter Counter used to total physical processes that provide apulse representing
some kind of event. An example is nuclear radiation pulse counting.
Quinary code Code that recycles in steps of five bits.
Register A counter that holds adigital number in short term storage so that it can
be operated upon.
Reset facility Terminal to which the appropriate polarity signal will cause the flip-flops
to change to agiven desired state.
Reversible counter Counter arranged to total pulses, adding or subtracting them as directed
by control line levels or the nature of the pulse.
Ring counter Counter in which the end of acounting chain connects back to its input.
A number set into it can be sent around as needed without losing it out of
the carry output.
Ripple through counter Alternative name for asynchronous counter.
Set facility Similar facility to reset. Two names used to help distinguish between the two
alternatives existing for setting up aflip-flop.
Shift register Register in which anumber has been set up for the main purpose of moving
it digit by digit for presentation to another number usually for binary addition.
Synchronous counter Counting system in which all stages are updated together when an input
appears. This overcomes the delay of ripple-through systems.
Ternary counter Counting in the number three system.
Up counter Counter wired to cause the contents to move up the number sequence for
each input pulse.
Up- down counter Alternative name for areversible counter.

31
 Digital displays

25 -their development
and forms

THE HUMAN operator of a piece of performed using single lamps for each modules. One method used a
digital equipment often needs to know binary bit position in the counter or moving-coil meter movement to rotate
the results of a measurement or to register of interest. A light ON a graticule, containing the 0-9
monitor progress in a process etc. The represented a 1 and OFF a 0. numbers, inside a projection system.
most effective way of conveying the Occasionally this may suffice where The filament lamp illuminated the rear
required information is by means of a binary readout is needed but the of the projection screen to make a
visual display. The display must inconvenience experienced where quite large number appear at the front.
provide the information in a readily decimal display is really needed In this way numbers were generated at
understandable form — that is, in the overwhelms the relatively small the same position giving a visually
decimal numbers, alphabetic additional cost required to convert the well-proportioned multidigit decimal
characters and symbols ( such as pi, binary code to its decimal equivalent. number. A sketch of this method is
decimal point, plus and minus signs Decimal displays using several quite shown in Fig.2. Watching such a
etc) of common visual experience. different means were developed. An display is somewhat disconcerting, for
Everyone will be familiar with obvious starter was the columnar the individual numbers wobble into
common examples of digital display method which used ten individual position with changing values. Another
such as the now ubiquitous calculator, lamps placed in a vertical column be- method used ten individual- lamp
digital clock radios and cash registers hind graticules of the figures 0-9. The projection systems each with its image
etc. ( Fig. 1). As the amount of inform- decimal number being displayed plane set on the same viewing location
ation to be displayed increases we find showed up as a single value on the rear of a screen. Provided only
much more complex displays, such as in each decade digit position. one lamp was energized at a time the
CRT terminals, which can not only The non- illuminated numbers do not number was easily read.
display a full page of alpha- numeric show if a suitable contrast graticule is Projection systems were, by today's
information but can generate graphical used. This technique was certainly standards, delicate, costly and far from
data and line drawings. better than a pure binary- display and ideal. Their continued use and
In this section we will study the many instrument systems of the late acceptance occurred mainly because
various types of device that are used to 50's and early 60's used this method. they did not require high voltage levels
generate the displays of calculators, A considerably reduced peak to operate them. Voltage-wise they
digital multi meters and similar power- supply demand results with were quite compatible with early
instruments. decimal displays as only one lamp in solid-state circuitry although
Decoding techniques used to convert each decade needs power at any time. considerable current gain was needed
those numbers held within the system Disadvantages of the columnar display between the counter circuitry — the
(in the binary or binary- coded-decimal were the sheer mechanics of arranging flip-flops — and the lamp.
form) into decimal numbers and numerous light- globe holders on a
alphabetical characters are discussed in panel ( Miniature Edison Screw MES NEON INDICATOR TUBES
the next section. Displays are dealt lamps were used), the non-aligned In parallel development were
with first because their requirements number appearance which was indicating display tubes based on the
partly dictate the decoding techniques awkward, and the character size that principle of the neon tube. These have
that must be employed. had to be small to keep the panel become universally known as " Nixie"
within reasonable proportions. tubes, a name really applicable only to
HISTORICAL DEVELOPMENT To overcome size and alignment tubes of Burrough's origin. As these
Originally the display of the digital difficulties a number of manufacturers are still designed into new equipment
numbers held within the circuitry was marketed ingenious opto- mechanical today — projection devices are seldom
used now — we will study how they
operate in some detail.
The basic neon lamp has two
electrodes, see Fig.3a, passing into the
interior of a glass envelope filled with
neon gas. With alow voltage difference
applied between the electrodes the gas
acts as an insulator and no current
flows between the electrodes. At
around 70 V the gas conducts
producing a red glow on the wires. If
the voltage rises above this value the
neon continues to glow; if it falls
below, the discharge extinguishes. As
there is no concept of rectification in
this double electrode system the neon
Fig. 1. This typical digital multimeter uses a neon- indicator tube display.
will light with ac or dc excitation of
32
 GLASS ENVELOPE
FILLED WITH NEON GAS

DISPLAY SURFACE

REFLECTOR

ELECTRODES
Fig. 3e. C-mstruction of the basic neon lamp.
b. How amulti-digit neon indicator is
constructed from an anode grid and ten
separate character-shaped cathodes.

Fig. 2. The moving-coil numerical character display is based on

an optical projection system. The numbers are on agraticule
which is rotated behind alens by ameter-line movement.
(Drawing courtesy of Electronics Australia).

sufficiently high voltage magnitude.

Single neon indicators are used
extensively for " mains- on" indication
in instruments, power points and
TEN SEPARATE CATHODES
appliances, in which case a series (INSULATED FROM EACH OTHER)
resistor is added to obtain operation at
240 V.
The neon- indicator tube, developed
from the basic neon lamp,
voltage, a 5 V level change, causes the MULTI-SEGMENT FORMATS
incorporates 10 cathodes ( when discharge to transfer from an Each of the above displays uses
out-of-sight cathode to a visible one. characters generated by the
numbers are to be generated; letters
and other symbols are available) one This single- bit principle has been application of a single signal that
for each 0-9 number, which are applied to a 10 step unit in, which provides the character complete. This
individual separate numbers are is said to be of simple format. An
stacked on top of each other behind a
fine mesh. Each is insulated from the illuminated as needed. This form of alternative method is to produce the
others and has a connection lead neon display has not become popular, character from individual segments or
brought out through the glass envelope probably because the numbers are dots arranged to build-up the shape
as shown in Fig.3b. The mesh acts as a arranged in a circle, giving small needed.
common anode electrode for numerals which do not line up when
After the very active development
whichever cathode is selected. The several displays are used to form a
period of the 60's designers and
tube displays just one of its number multi- digit decimal number. ( One early
suppliers are now settling on the use of
set. Non-energized grids remain dark variety produced a dot glowing at the
seven- segments, hexa-decimal 7 by 4
and are unseen because they do not side of the numbers printed around a
dot and 7 by 5 dot matrix formats.
circle).
glow.
Neon- indicators radiate red light, Seven-segment format — This is the
Numerical neon- indicator tubes are
which ( more by chance than design) simplest and most used composite
made such that the numbers appear
happens to be at a wavelength of matrix method. It consists of seven
either at the side of the glass cylinder
reasonable sensitivity to the eye. Red equal-size bars placed to form the 0
or at the end. Character sizes ranging
is particularly suited to strong ambient through to 9 series of numbers. Several
from 10 to 50 mm are available. This
daylight viewing. distinct alphabetical characters and a
form of display has remained popular
for reasons of the very acceptable
readability, nicely shaped character
format and low-costs. They require a
relatively high voltage supply ( 180
Vdc is typical) and are not as robust as
the solid-state devices described later.
The format and connections of a
typical neon- indicator tube are
illustrated in Fig.4. Note that only one
input drive signal is required to
energize any particular display
character. The majority of all other
displays in use require several inputs to
be energized in order to produce the
desired character. We will see later,
however, that the amount of decoding
circuitry needed for neon- indicator
systems and the solid-state alternatives
is similar.
It is possible to construct
neon- indicators needing lower input —
command voltages. In the Mullard
Digitube, for example, the discharge
remains on continuously. The trigger
 minus sign are also possible. The
appearance of seven-segment numbers
STANDARD NUMERAL I I - I - 1111 - 1
I

and letters is as shown in Fig.5. This

SET
II IC11 - 111111
system is based upon a stylised figure
of eight. Of particular note is the DISTINCT LETTERS I I
requirement that the individual
characters are generated with different
AVAILABLE
II L
combinations of bars being
Fig. 5. Format of seven-segment numeric and alphabetic characteristics.
illuminated.
Methods for illuminating a bar
include separate filaments for each,
separate incandescent bulbs,
luminescent phosphors lit by •
• •
•
•
•
• •
•
• • • • • • • • •
• •

filaments, light- emitting diodes ( LEDs)

• • • • •
así . •• •• •••• •• •••
g • • • •
and liquid crystal indicators — more of • •
••••
•
••
•
•
•
• •
•
•
•

these later.
Hexadecimal format — these rely on
the formation of a character by Fig. 6. Typical format of characters of the hexadecimal system using a 4 by 7 dot matrix.
illumination of the necessary dots ( or
small squares) of a 4 by 7 dot matrix.
••• •••• .••••
Figure 6 gives the appearance of
•••
• ••• • •• .•
...
•

number characters generated this way.

••••
• •
• •
• ••••
• •
• •
•••• • ••
• • •
... •
•
Note again the need to energize •••• ••••• ..• • • ••••: • • •

selected positions to provide the •

••• •
•
•••
• •
•
•••
•
.....
• •
•••
• •
•••
•
•••
•
• •
.• ••••• •...• •...:
required character. ••
••
• • • • • • • •

Alpha- numeric matrix format — the

• • • • • •
•••••
• ••• ••• • ••••• •••• •••••

above 7 by 4 matrix is limited in that

whilst it can generate all numbers, it Fig. 7. Alpha-numeric characters as generated by 5 x 7 dot matrix.

cannot provide all 26 alphabetic

characters. If the matrix size is 20 - 30 lumens per watt. semiconductor junctions ( formed by
increased to 7 by 5 the full 36 Neon- indicators consume less power in the same processes used to make
alpha- numeric characters can be general and deliver a brighter output solid-state signal diodes) which emit
generated. Figure 7 gives the but do require a high voltage that is radiation from the junction when
characters of the American Standard not directly compatible with the now current is passed through it. The basic
Code for Information Interface standard 0-5 Vdc TTL signal levels. materials used are gallium arsenide
ASCII. The life and robustness of both fila- phosphide GaAsP and gallium
SOLID-STATE DISPLAYS ment lamps and neon devices is also phosphide GaP.
Incandescent lamps are very far from ideal. The breakthrough came This form of light source generates
inefficient at converting electrical several years ago when light- emitting relatively narrow wavelength energy
energy into radiant visible energy — diodes ( LEDs) were developed. centred on red yellow or green
conversion is generally only around Light Emitting Diodes — LEDs are colours. ( Typically 635 nm, 583 nm
and 565 nm wavelength respectively)
with high luminous efficacies of 140,
460 and 610 lumens per watt.
Compare these efficacies against that
for atypical tungsten filament lamp of
20 lumens per watt. The term efficacy
should not be confused with
efficiency. Efficiency is the percentage
of radiant power compared to input
power whereas efficacy refers to the
effectiveness of the radiant power
produced in stimulating the eye. For
example an LED producing infra- red
radiation will have an efficiency of say
3% but an efficacy of zero.
The high efficacy of LEDs means
reduced power supply requirements,
and high visibility is obtained even
when LEDs are driven via a resistor
directly from TTL.
Another feature of LED sources is
the high speed of response — 100 ns is
typical. The operating voltage is
nominally 2 V and current require-
ment varies around 20 mA.
Single and multiple format LED
This calculator from Hewlett Packard displays calculations and instructions displays are now available in a wide
from the calculator to the user by means of a full alpha-numeric 5 x 7 dot matrix display. variety of forms and they are the most
34
 35

30

30°

TA = 25 ° C
425
E

40° e, 20
e- 2.0

1.5
I
‘7)
6
e 010 8lu
7o°
80 ° 5
50.5
90 °

o 0.5 1.0 1.5 2.0 2.5 3.0 ° o 5 10 15 20 25 30

VF — FORWARD VOLTAGE — VOLTS IF — FOVVARD CURRENT — mA

Fig.8b Fig.8c Fig.8d

Fig.8a

Fig.8e Fig.8t Fig.8g

1.0
10
RELATIVE INTENSITY
RELATIVE INTENSITY

RELATIVE INTENSITY

o 550 600 650 700 750 600 650 700

500 550 600

WAVELENGTH — non WAVELENGTH — non

WAVELENGTH — nm

Fig. 8 Characteristics of the HP 5082 mini- LED series lamps.

(a) the shape of the lamp; (b) relative luminous intensity versus beam angle; (c) Forward current versus forward voltage. (d) Relative
luminous intensity versus forward current; (e) Intensity versus wavelength of green lamp; (f) Intensity versus wavelength of yellow
lamp; (g) Intensity versus wavelength of red lamp.

1 A typical LED indicator lamp.

An array of 10 light emitting diodes as used for paper tape reading.

35
 ELECTRONICS - it's easy!
used display medium. Figure 8gives the
various data of a typical unit. Figure
9 shows how a single lamp can be
mounted in practice.
Developments arising out of the basic
single LED lamps are units
PANEL incorporating an integrated resistor
MOUNTING. (for direct TTL connection) those
having an integrated voltage sensing
amplifier ( Fig.10) which provides a
lamp that triggers on or off as the
input level passes up or down through
a 2.5 V level and the opto - electronic
CLIP
RETAINING relay or isolator discussed in aprevious
RING
section. Hermetically sealed units and
Fig. 9. LED lamps such as this may be mounted directly military approved units that will
onto a PC board or onto a front panel by means of the operate from -65°C to + 100°C with
clear plastic clip.
very high reliability over a life
measured in years of operation are also
available.
VIN Given a matrix of LED lamps it is
quite practicable to generate numbers
3.0 and characters by what is called an
Iv LUMINOUS INTENSITY — mcd

TA = 25 ° C addressable system in which decoding

2.5 logic decides the diodes to be
LED
illuminated. LED character displays
DRIVER 2.0 are marketed as single unit 7 segment
TEMPERATURE modules and as 4 by 7 and 5 by 7 dot
COMPENSATED
REFERENCE 1.5 matrices. Integration has gone as far as
VOLTAGE 10 mV incorporating a complete decade
OFF TO ON counter stage ( Fig. 11), with the
1.0
LED. necessary decoders, buffer amplifiers
and LED display all integrated on a
.5
single LSI unit. As LED manufacturing
techniques are the same as
oo
1 2 3 4 5 6 conventional integration methods it is
GND possible where large quantity
V IN — INPUT VOLTAGE — V
production is economic, to integrate
Fig. 10. LEDs with integrated voltage sensing amplifiers turn on when the applied voltage
the display with the circuitry —
exceeds a built-in value. examples are to be found in some
LEFT: schematic. RIGHT: luminous intensity versus input voltage. BELOW: ways of styles of IC wristwatch.
increasing threshold voltage.
Seven segment LED displays have the
eight diodes placed on a common
transparent GaP substrate. ( The eighth
diode provides a decimal point). A
Wm typical single unit is shown in Fig.12 —
VTH
EXTERNAL COMPONENT V-
fH they are available in red, yellow and
TC - TA ( m/°C)
green colours. The 7.6 mm letter size
SCHOTTKY DIODE
Is visible at 3 m; alarger 11.0 mm size
0-1111H-0 ( HP 5082 -
VTH + 0.45V
EXTERNAL
Vi-Fi VTH 2835) -2 can be readily seen at 6 m. Another
COMPONENT
series, shown in Fig.13 includes an
P - N DIODE integral optical magnification
Vii.
° 71÷1—O
VTH (1 N914 ) VTH + 0.
75 V - 2.5 technique that provides improved
TH readability for low drive power ( 1mW
LED
per segment). These are available as 3,
i I VTH + 1.6V - 2.9
I 'VOLTAGE NI--1
-÷1.-el-H 4 and 5 character units which are
I ...1, 1SENSING mechanically compatible with
I i LED ZENER DIODE
I I
46_114__Ei standard printed-circuit board hole
L _ V'rH VTH VTH + VZ -1 + ZENER TC spacings.
To meet the demand for portable
R SERIES RESISTOR
VTH + 1THR calculators manufacturers also supply
C
1-eVVV%-0
VTH VTH NOTE 1.
NOTE 2. special units with 8 or 9 digits
mounted on a small plug-in
NOTES: printed- circuit board.
1. ITH IS THE MAXIMUM CURRENT JUST BELOW THE THRESHOLD, VTH. SINCE BOTH ITH AND
VTH ARE VARIABLE, A PRECISE VALUE OF VTH IS OBTAINABLE ONLY BY SELECTING R TO The range of dot generated character
FIT THE MEASURED CHARACTERISTICS OF THE INDIVIDUAL DEVICES ( E.G., WITH CURVE
TRACER). displays is also extensive. A 39 mm
2. THE TEMPERATURE COEFFICIENT ITC) WILL BE A FUNCTION OF THE RESISTOR TC AND THE high character is available that can be
VALUE OF THE RESISTOR.
read from 20 m. This, as can be seen in
Fig.14, is based upon a large size
36
 00IC
5 by 7 dot matrix and includes the
r0 1.04IC CHIP decoder/driver unit for the most
commonly used BCD code — the
{SI RIALL 0 _, 8421 - logi c input ( decoders are
discussed in the next part). Dot matrix
L D-•
•NPUT 'ARAL LEL

CLOCK
displays with characters as large as 45
cm height are produced. These,
however, are not usually solid-state
but use electromagnetic drives to
rotate reflective dots into or out of the
viewing aperture. Such units, given
adequate ambient light, are visible at
300 m. Multi-digit dot matrix
solid-state displays are also made.
Liquid Crystal Displays. Although
LED displays consume little power
compared with earlier filament
displays very little of the power used is
actually transmitted as radiant energy.
CLEAR

Efficiencies of visible diodes are

LATCH
typically only 0.1%! Thus an LED
INPU1
display often consumes considerably
more supply power than the rest of
Fig. 11. The Texas Instruments TIL306 display integrates all the logic of a complete decade the associated digital system.
counter onto the same chip as a 7-segment display. The circuit shown is the schematic of the Indicators of all types, except liquid
device.
crystal, require about 300 to 500 mW
per character ( all segments
illuminated).
The power requirements of the
display could be reduced considerably
if the circuit could switch available
ambient light rather than actually
generate light. Naturally such a
method will only work when
ambient light is available.
In the dark, displays which generate
radiation would still be required.
Displays are available which do switch
ambient light. They are known as
liquid crystal displays and by virtue of
their mode of operation consume very
little power.

Fig. 12. ABOVE: three seven-segment LED Fig. 13. Some seven-segment displays suitable
(DL 704) displays mounted on a common PC for calculators etc are assembled in groups
board. LEFT: internal diode positions for a and have plastic lenses to increase character
right-hand decimal point seven-segment size.
display module.
37
 4— 47.5

Ô0000 PIN FUNCTION

vcc
0 00 2 DP

0 00 3 X1
38.1 4
00000
X2
5 X4

00 58.4
6 X8

00 7 GROUND
8 NC

9 NC Fig. 14. A large size (38 mm

Pis:
character) alpha-numeric
10 VLED display constructed from

,.
7.6 discrete LEDs and a
U v
L decoder/driver IC. (HP5082-
44.1 7500).

MECHANICAL DETAILS FORMAT

Basically liquid crystal displays techniques. At present the life of the In each case,however, design
consist of a minutely thin layer of display is inferior to LED units. information is vital to ensure that the
liquid- crystal material placed within Although manufacturers quote 10 000 displays are used within ratings.
two thin glass covers. The glass covers hours minimum life ( just over a year) Advanced display design has become a
have transparent electrodes deposited experience has shown that units often high-level art and generally
on them in the shape of the characters fail after only a 1000 hours. Application Notes are the essential
or segment needed. This is shown in Seven segment displays are also made guide to their successful use.
Fig. 15. With no excitation the whole using neon lamps, self contained
unit appears transparent, for the liquid filaments and separate incandescent Hewlett-Packard produced an " Opto
crystals remain stationary allowing Electronics Designer's Catalog" in
bulbs. It is to be expected that these
light to pass through virtually 1973 and 1975. The former included
will not be in use in new designs of the
unattenuated, that is, no light is several applications notes, the latter a
future for the price alone of solid-state
reflected. When an alternating voltage list of the range of relevant application
devices will usually undercut the
(40-1000 Hz) is applied to the notes now available from HP: both
available alternatives.
electrodes forming the character Regardless of the display used it is contain awealth of practical data.
shapes, the resultant electric field necessary to decode the binary logic of "Digital display systems", written by
causes the liquid layer to become digital circuits into a code suited to E.G. Breeze and available as Fairchild
turbulent, scattering light between the illuminate the required number and Application note 212/1, 1972 is also
confines of the deposited areas. The combination of characters in the worthwhile having.
display then shows an optically dense system used. The next section will
character because the ambient light is look at the schemes used and at more Many other manufacturers — Texas
reflected. In simple terms application efficient methods of driving multiple Instruments, RCA, National, Hawker
of an input signal causes the liquid character displays. Siddeley Electronics, Monsanto,
crystal in the vicinity of the Mullard, Atron, Litronix, Siemens —
transparent electrodes to act like a REFERENCES also provide service data that gives
mirror. The use of solid-state displays is practical advice on how to use their
The power requirement for the straight- forward in simple applications. display products to best effects. •
circuit driving liquid crystal displays is
around 20/4W per segment ( compare
this with the lowest 100 pW per
segment but more usually 20 000 pW
for LED characters). Response is not Fig. 15. The construction oía
SWITCH CLOSED liquid- crystal display.
as fast as for LEDs — 20 ms rise- time CAUSES CHARACTER
and 100 ms fall-time, but that is not a TO APPEAR.
serious shortcoming in visual
observation applications. In some
instances faster response is needed —
consider, for example, the use of
photographic recording of a character
display. With LED displays the
display, when being photographically LIQUID CRYSTAL
recorded, can be cycled considerably LAYER
faster than the eye can follow.
Liquid- crystals are the most recent
solid-state display to be developed and GLASS COVERS WITH
it is still too early to state with ETCHED CHARACTER
certainty if they will eventually AMBIENT LIGHT DEPOSITED AS ELECTRODE

compete seriously with LED

38
 Code converters

26
and display
systems

IT IS both necessary and convenient

BINARY STORE
to transfer data between sub units of a L_
digital system, by means of some kind
of code. We have seen in the previous
section how counting stages are
r—
sometimes arranged to count in BCD BINARY TO BCD CONVERTER
(Binary Coded Decimal), and how this
form of code must then be ' converted'
into another form that is suitable for
the particular kind of display device
used. Thus codes and code converters BCD BCD BCD
are of great importance in digital
instrumentation.
There are amultitude of digital codes
in use for communication, data DECODE DECODE DECODE
interchange, and for numerical

I_ I
manipulation and display. Although
Fig. 1. Basic arrangement
many of the earlier codes used have
of binary to BCD
now been discarded, there are still converter and display
dozens in use. In this section we will decoders. 3 5
not discuss codes like ASCII, Baudot,
Excess 3 etc, which are computer and
communication codes, but restrict any other system need, the binary
BINARY TO BCD CONVERSION
ourselves to those codes and counter is reloaded and the process
converters which are concerned with
AND VICE VERSA
repeated to convert the next number
counting and display. To convert binary to BCD acommon
from binary to BCD.
The main counting codes used in method is to set a binary counter —
instrumentation are binary and BCD. see Fig. 3 — to the desired number The reverse, BCD to binary, is
Octal and Excess 3 are other counting either by direct counting upward from accomplished in the same manner
codes used in computers but seldom in zero, or by transferring the value except this time the roles are reversed
instrumentation. Converters are across from another stage with a as shown in Fig. 4. The BCD counter is
needed to change from any one of parallel converter. Clock pulses are set to the desired number, the clock,
these codes to any other, and between then fed into both this binary counter, when enabled, clocks the BCD counter
any one code and decimal or vice now set to count back down to zero, down to zero and at the same time the
versa. In addition the counting code in and to an up- counting BCD counter. A binary counter upward. When the BCD
use needs to be converted into aform detector senses when the binary counter reaches zero state its outputs
suitable for driving particular kinds of counter reaches the 0000 state upon logically inhibit the clock input to
display ( seven- segment, dot-matrix and which any further changes in the both counters. The process is repeated
neon tube, etc). count state of both units are inhibited. for each new number after clearing
For example we find converters for The BCD equivalent of the binary and resetting the two counters to the
binary to BCD, binary to decimal, number is now held in the BCD correct starting conditions.
BCD to seven- segment, and BCD to counter. This serial method is fairly slow and
decimal, to mention just a few of the At this point the BCD number is a much faster method is to use logic
possible combinations. Figures 1and 2 cleared into a store or is available for gates in a parallel arrangement. The
show two such variations.

PARALLEL BCD OUTPUTS

B OD

1)11111•

INPUT
Fig.2. An asynchronous CLOCK CLOCK CLOCK —a CLOCK
BCD counter provides SERIAL
aparallel BCD coded BINARY IN
output which
corresponds to the
number of pulses fed to
the input.
39
 BCD OUTPUTS as decimal indication, it is necessary to

II III? MI'
energise the display or character
printer segments appropriately.

UNITS DECADE CARRY

Displays, such as the neon tube or the

DECADE TENS
DECADE HUNDREDS columnar style — see previous part —

BCD
require one output for each of the 0 —

COUNTER 9 numbers. From the truth- table of

Fig. 5 it can be seen that to energise,
say, a decimal '4' output we must set

CLOCK
up a logic gate that provides an output

INPUT
-fp-• NULL
DETECTOR
when the BCD state is 0100. For ' 8'
we need a BCD state of 1000. It is not
possible to totally economise by using

u
only ' 1', as the logical indications for
this leads to ambiguities between

BINARY
COUNTER
numbers. On the other hand there is
no need to arrange for all code

CARRY sequences and bit combinations as that

introduces redundancies using up extra
unnecessary gates.
By careful design, and the use of

L6.11UBINARY INPUTS SETU

inverters to invert ' 0' to ' l's, it is

NUMBER TO BE CODEDTO INTO

BINARYBCD
possible to find aminimum number of
AND gates and interconnections that
will produce the 10 decimal states ( 0
to 9) as distinct outputs from the
four- line ( A B C D) outputs of the
Fig.3. System required to convert abinary number to its BCD equivalent.
BCD stage. One such scheme is given

SET TO BCD NUMBERBCD TO- INPUT

in Fig. 6. Thus, by the use of logic

BE CODED INTO BNARY gates alone we can provide a parallel

code conversion from BCD to decimal.

FM' I I I ?' MI
Getting from BCD to a format
suitable to drive a seven-segment

DECADE CARRY
display requires more gates, see Fig. 7,

UNITS
DECADE TENS
DECADE HUNDREDS
but the technique is basically the
same. A decoder suited for the BCD to
decimal requirement of a neon tube is

BCD
quite unsuited to drive a

COUNTER
j seven- segment display. As both these

CLOCK
and other conversions are in great

INPUT
-1,--4 4 NULL DETECTOR
demand they are available as simple
ICs. Further, in some options the
decoding logic is integrated onto the
same chip as the BCD counter stage.

BINARY
In practice the need to understand

COUNTER
the internal operation of the decoder
arrangement rarely arises, for the ICs

CARRY are clearly marked with

connections to be made — it is just a
case of making correct connections
the

between the counter- stage chip, the

(LSB) 2.2,22
(I
I 22 24.1 25 26 27 29U(29 .1
decoder chip, and the display.

(
) (! 551) THE NEED FOR DISPLAY

BINARY APPEARS HERE

I

DRIVER STAGES
OUTPUT
The power levels available from
decoder stages are rarely adequate for
Fig.4. To convert aBCD number to its binary equivalent the same circuit as Fig.3 is
used but the BCD number is now the one pre-loaded. direct drive of display units. A buffer
stage which raises the power level is
normally required. Again, these are
generally incorporated into the
parallel method works well for BCD to using four bits. The 1-2-4-8 weighting decoder IC stage. Such integrated units
decimal and other cases but requires is the most usual form used but other are known as decoder/drivers.
innumerable connections when set to codes such as the 'excess three' and Different displays, even of the same
convert decimal or BCD back to the ' Aiken' variations, are also used. format, require differing power needs
binary. It is little used in this reverse Thus, each of the digit values 0 to 9 so it is important to select
mode. (in decimal) is represented by four decoder/driver stages suited to the
lines, each having a '0' or a ' 1' state. display being used.
BINARY CODED DECIMAL
This is demonstrated by studying the The buffer stage of a decoder/driver
TO DECIMAL AND OTHERS truth table for equivalent BCD and obtains current ( or voltage gain) by
Each BCD stage stores its decimal decimal states given in Fig. 5. the use of a transistor stage such as a
number as some form of binary code When the output must finally appear Darlington pair or an emitter- follower
40
circuit. A method recommended for
driving seven- segment fluorescent
displays is shown in Fig. 8 — these B.C.D. Decimal
Outputs equivalent
Inputs
displays require high voltage drives.
Sometimes the need arises to drive 0 1 2 3 4 5 6 7 8 9
AB CD
displays from special-purpose one-off
decoding circuits. In such cases a ooo ooooooooo 0
suitable driver stage is obtained by ooo o ooooooo 1
using standard IC inverter chips. 1o o o o oooooooo 2

(Discussed in Part 23). 1o o o o oooooo 3

o o o ooo oooo o 4 Valid state

o oooo oooo 5 of encoding
o o o oo ooo ooo 6
THE ASSEMBLED BASIC
1 o o oooooo oo 7
COUNTER WITH DISPLAY 1 o oooooo o o
o 8
An illustration of the actual Oo 1 o ooooooo 9
construction of a complete counting oo 1
stage is to be found inside the digital
o O 1 O 0 0 0 0 O 0 0 0 o o
clock shown in Fig. 9a. The display — O 0 0 0 o o Illegal states that
1 1O 1 O 0 0 0 0
neon tubes in this case — is arranged to O 0 0 0 0 O 0 0 0 o o should not occur
oo 1 1
o o
appear where required by using O 0 0 0 0 O 0 0 0
o 1 1
O 0 0 0 0 O 0 0 0 o o with BCD code.
end-connected tubes which plug into
o 1 1 1
O 0 0 0 0 O 0 0 0 o o
the main printed-circuit board as 1 1
shown in Fig. 9b. Immediately behind
each neon tube is the decoder/driver
IC which, in turn, is driven by the
counter IC located behind that again.
Fig.5. Truth table relating four- line BCD with its decimal equivalents.
On the circuit diagram these appear as
shown in Fig. 9c and on the
component overlay of the PC board as
>4)
in Fig. 9d.
The Texas Instruments TI L 306 LSI
is an example of an IC unit that
incorporates the counter, decoder,
driver, and display in one package. No
doubt this will be the trend as it
provides reduced assembly costs and
smaller packaging. It should not be
unrealistic to expect complete
single- IC, multi-stage counter units to
be in general use before long.
BCD
CODED 0—a— D__4:4
) INDIVIDUAL
INPUTS DECIMAL
LATCHES FROM OUTPUTS
Direct coupling of the display to the ONE
DECADE
counter stage results in acontinuously
UNIT
changing display value. If the input is INDIVIDUAL
sufficiently dynamic it is awkward, if DECIMAL 0-6—

not impossible, to read values. OUTPUTS

Addition of a latch overcomes this by
sampling and storing the count to be
displayed for fixed intervals whilst the
counter continues to cycle.
This is achieved using a memory
stage between the counter and the
decoder/driver stage as shown in Fig.
10. This system, the digital end of a
digital voltmeter, displays a steady
value for ashort period by transferring
the instantaneous value of the divider
Fig. 6. Logic arrangement for decoding BCD units into one-of- ten outputs.
stage ( the counter) into a buffer-store
stage. The transfer or updating process
is initiated by a common display the correct decimal number. The
counter to be sampled and stored by a
timing line which is actuated at simplest arrangement is when the
single IC.
appropriate intervals. Such latches are range switch of, say, a multi- range
invariably placed between the counter digital voltmeter decides the point to
and decoder stages. DECIMAL POINT
use. In autoranging voltmeters ( etc)
Internally a latch is a bi-stable Facilities are usually provided in all
and in many calculators, solid-state
designed specifically for the purpose displays to enable an input to energise switching is used to select the correct
of storing and transferring the value of a left-hand or right-hand point at each
point position.
a digital bit. Integrated circuit units digit position. Obviously specific
provide four surs, latches in a circuitry is required to energise just
dual- in- line package — see Fig. 11, thus one of the total available in a BLANKING
allowing the four line data from a BCD multi-digit display in order to present In normal writing practice we do not
41
 ELECTRONICS it's easy!

Fig. 7. Logic for converting 8CD to seven-

segment display format.

a
write the O's that appear at either end
of a number, for example, 0001357.0
as could be held in an eight digit
TD
D D—'
display, is better presented as 1357 or
as 1357.0. A facility is sometimes
provided in display-counter systems
that blanks unnecessary zeros. Leading
zero suppression is performed within
the decoder stages of Fairchild
seven- segment decoders by connection

_D—oc
of the ripple blank output RBO ( ripple
because each stage connects to the
AO- •
next) of the decoder stage to the
AO
ripple blank input RBI of the next
BO OUTPUTS TO lower decoder stage. Blanking of
INPUTS BO DISPLAYS least- significant zeros is not usually
FROM SEVEN included. The actual arrangement for
BCD SEGMENTS blanking control varies from maker to
CO
STAGE maker. Fig. 12 shows a method using
é
-
ripple blanking.
DO The blanking facility can also serve
A
other purposes. It can, in certain
applications, be used to blank- out
illegal display values resulting from
incorrect codes. The RBO output also
provides a detection output indicating
when the decoder stage is at the BCD
zero state.
Blanking is also valuable as a way to
save display power for it can be used
to hold all displays off until there is
something to display.
When no blanking input is provided
it is also possible to blank the system
by applying the spare BCD code states
that result in no drive to the display.
Yet another procedure is to disconnect
the supply from the display itself.
680
+ 5V INTENSITY CONTROL
Displays are usually manufactured to
FD 111 supply one value of output brightness.
When brightness is to be tailored to

+5V

Fig. 8. How buffer transistors are used to obtain

increased drive for fluorescent displays (seven-
9 segment format).
—0 RB18 f

X. LT e
A3 d
a A.
— A2 z c
b , OTHER I
A. E 'DECADESI
"'RBO D—

100k

—60V 0 —4

DECADE IM.S. DECADE

42
Fig.9. This digital clock illustrates use of counter stage subsystems. (a)
(a) The frontal appearance of the Nixie display.
(b) Top view of component assembly.
(c) Circuit diagram of counter to display section.
(d) Component overlay. The V1 to V4 positions are the sockets for
the Nixie tubes, behind are the decoder/drivers which have the
dividers behind those again.

(3)
150V ac.

DISPLAY TUBES

DECODER/DRIVERS

DIVIDERS

(c)

(d) CIRCUIT SHOWN IN ( C)

The schematic of Fig. 12 includes an energises the correct dots or segments.

particular ambient light conditions an
intensity modulation facility. If each characte - has its own decoder
appropriate kind of display can be
selected that provides the desired With LED displays, intensity we would need 7 lines for each digit cf
modulation can actually increase tne a seven- segment display. And 35 iines
luminance level. This however, does
apparent brightness. The human eye for each digit of a 7 x 5 dot-matrix
not always leaa to a satisfactory
choice when other considerations are has a characteristic response to display!
radiation that has greater sensitivity to Obviously a method is needed to
taken into account.
the peak value of modulated light, reduce the number of lines and
Intensity of any display, however,
can be controlled in a digital manner rather the average or rms power. LEDs circuitry required for murti-digit
(that most desirable in digital systems) can be pulsed at high frequency with displays.
high peak currents because of their One such method is called strobing
by turning the display on and off with
nanosecond response time. The net where lines of dots or segments are
an appropriate duty cycle ( ratio on to
result is apparently higher brightness illuminated sequentially. The 7 x 5
off period). Tnis is called
pulse-duration irtensity modulation. for agiven amount of power. array can be either strobed as lines
Provided the repetition rate exceeds horizontaily or as rows vertically as
100 Hz the eye cannot detect that the STROBING OR SCANNING illustrated in Fig. 13. Each row is
Displays which generate characters in selected one by one in sequence and
radiation source is being modulated.
Modulation may be achieved with any the 7 x 5 dot matrix or seven- segment the appropriate diodes in the row
of the blanking methods given above. formats require decoding logic which energised. Provided each row is
43
 ELECTRONICS it's easy!

•MRLIFIER
AND
TO OTHER LATCH
SHAPER

RESET

TIME CONTROL.
BASE
OSC

(al
CLOCK DATA

COUNT STORE
V
> CLEAR

BCD
CODE
Si C1 C2 C2
1,2.4.8 A'
ID 10 D 10 B.
BUFFER BUFFER BUFFER BUFFER 20 20 D 20 C. TO
STORE STORE STORE STORE MONSANTO
30 30 3D 30 cy
MAN 1 OR
D 40 D d E' MDA 100
CODE
1,2,4,8 MDR 100 F' DISPLAYS
DECADE COUNTER MOL 105
TORAGE ELEMENT MS0 047 G
DECODER D DECODER/DRIVER
DECODER DECODER DECODER
R

DRIVER DRIVER DRIVER DRIVER

A (b)
vcc

NUM, NUMI NUMI- NUM' Fig. 11. Schematic (eland oin connections (b) of counter, latch
TAON TRON TRON TRON
decoder system using the MonsantoMQL 105 four-bit bistable
latch.

Fig.10. Buffer stores are used to latch the display causing it to

remain steady for selected periods. (circuit shown is acounter/timer
using RCA incandescent displays).

A2 Ag
B2 B8 C2 Cg
D2 Dg
Al A4 B1 B4 C1 C4
Dl D4

IN
MODULATION
INPUT

LAMP
oTEST

o 00 0
LT RB1 LT RB1 LT RB1
9317 LT RB1
9317 9317 9317
7 SEGMENT DECODER 7 SEGMENT DECODER 7 SEGMENT DECODER 7SEGMENT DECODER
abcdefg abcdefg abcdefg
RBo?O Q 0 CI O C1 RB 0? 0 0 O o o o
abcdefg
RBoyomogo R go?ovacbceçece

II

LEAST SIGNIFICANT DECADE

TO SEVEN SEGMENT DISPLAY LAMPS MOST SIGNIFICANT DECADE

Fig. 12. Connections for ripple blanking in a four-decade display system. ( RBO ripple blank output, RBI ripple blank input).
44
O 0 0 0 0 0 0 0 0 0
O 0 0 0 0 0 0 0 0 0 COLUMN

1
O 0 0 0 0 0 0 0 0 0 1qP Pht i SCAN
VERTICAL
O 0 0 0 0 0 0 0 0 0 STROBE
IA IBCD IE 2A 2B 2C 71)2E
O 0 0 0 0 0 0 0 0 0
O 0 0 0 0 0 0 0 0 0
O 0 0 0 0 0 0 0 0 0

HORIZONTAL
III
STROBE

Fig. 13. Horizontal and vertical strobing of a7x 5

dot matrix display.

ROW
SCAN
Fig. 14. Schematic of switches needed to address a 7x 5 dot
matrix.

•
•
•
6 LINE ASCII () • •
O
O •

ARRAY SELECTO 2 5

•
MASTER • • • •
CLOCK
•
returned to at no greater than 10 ms
TIMING
intervals the characters will be flicker CIRCUITRY 6 BIT INPUT STORAGE BUFFERS
READ ONLY
free. MEMORY
A diagrammatic illustration of how
specific diodes are selected in a row is 5 BIT OUTPUT STORAGE BUFFERS

given in Fig. 14. The row switches are

scanned in turn to cause a vertical
•
scan. Simultaneous excitation of the
other switch sets decides which diodes
in the row are to be illuminated.
A strobing system requires a
procedure to sequence the scanning
action and a method of setting the
selection switches that corresponds to
the characters needed. The whole is
controlled by a clock and timing ROW
generator. Storage buffers are also DRIVERS

required to store the sequentially

generated information. The task of
creating the appropriate character
Fig. 15. Basic block diagram of a vertically
timing codes is performed by a
strobed display using 7 x 5 dot-matrix
read- only- memory ROM. Clearly this devices.
method adds up to a complex
system ... really beyond this course's
purpose. A schematic block diagram of MULTIPLEXING display to achieve the same brightness
(LEDs are not so critical as other
a vertically-strobed five- digit LED When the input data to be displayed
display is given in Fig. 15. Although of appears in serial form or when large forms of display); the scan frequency
apparently great complication this numbers of displays ( over four digits) must be at least 100 Hz to prevent
flicker; transients must be carefully
method is less expensive to employ are involved, multiplexing ( selection of
complete digits sequentially) becomes decoupled; and a clock failure ( which
than direct actuation through fixed
gates. ( Considerably more detail is to advantageous for driving seven stops the scan) may produce partial
be found in the suggested reading list). segment and one- of-ten displays. The display failure because of excessive
Another scanning method scans the basic multiplexing system requires the dissipation brought about by the
main system units shown in Fig. 16. increased voltage applied. ( It is usual
matrix as a raster — across a row, one
by one, and then to the next row. An upper limit to the number of digits to include a failsafe protection
is around 12 and higher for LEDs. circuit).
Strobing obtains its advantages by
time-sharing common elements in a There are disadvantages; namely, a Again, the complexity appears great
higher voltage is required in the but in practice the multiplexed system
time- multiplexed manner.
45
 I ASTABLE
MULTIVIBRATOR
CLOCK specified for applications where their
analogue equivalents would have been
COUNTER more suitable. Now though, common
sense is beginning to prevail. Analogue

ISCAN DECODER I
displays are gradually regaining ground
as it becomes clear that they are more
BCD INPUTS suitable for trend and other dynamic
observations. Nevertheless, many of
1111 t
-
-/1111
- today's analogue displays use digital
INPUT techniques internally.
MULTIPLEXER

Rotating pointers, bar-graphs and

_ similar analogue visual effects are now
—
1 1 = Il = 1 1=1 1= being developed for use in the auto-
o _ 1
- 1
Il II — Fl
w
o
fr
o LI =11= =1_1=11= Il motive industry. Large-scale pro-
duction prototype systems are already
undergoing trials in cars. From this
Fig. 16 Schematic of amultiplexing system for use with amultidigit, 7-segment display.
area of development it is logical to
is simpler to build. For example, a However specific ICs are marketed expect these new forms of analogue
multiplexed, seven-segment display, to suit various interfacing require- display to find their way into other
with storage for eight digits, involves ments. Other converters are needed applications. Now that most of the
around 10 dual- in- line packages and a for sending digital signals through development has been completed the
few discretes which compares with standard transmission lines in comm- costs should be low. The consequence
about 16 ICs for a non-time shared unication links, for receiving signals is that they will be introduced very
system. from lines, for increasing the logic rapidly into general use.
To further reduce the connections level differences to increase noise Manufacturers are seeking and find-
that must be made upon assembly, immunity ( again for transmission), ing ways to produce displays that can
manufacturers offer multi-digit dis- and units that drive peripheral devices be made cheaply in volume. This is
plays in which the anodes and such as relays and indicators. Signal possible using integrated circuit and
cathodes of the LEDs are internally inversion may also be necessary — thick film techniques. Figure 1 shows
connected ready for multiplexed op- we have already dealt with the inverter a conventional instrument panel com-
eration. block earlier in the course. pared with a solid-state electro -
Another class of converter is needed luminescent unit developed by Smiths
OTHER CONVERTERS for converting digital signals to ana- Industries. Note the drastic reduction
Apart from digital-code converters, logue voltages ( and currents) into in the number of individual parts and
other converters are required in digital form ( A to D). Such converters the great reduction in thickness.
instrumentation; for example, when will be dealt with in the next sect- It is also possible to make meter-
interfacing different systems of logic, ion. style indicators in which the needle
eg TTL to CMOS, it is necessary to rotates — but that is more costly. We
alter the dc levels of signals so that THE CHOICE OF DISPLAY can expect to see changes in car
the output of one system provides the Choosing a display can be quite a instrument panels because of cost
logic levels required by that following. task because many options exist. advantages of bar-graphs.
This may require amplification or During the early 1970s it was not Electromechanical devices will still
attenuation or shifting of a level. uncommon for digital displays to be have their place for some time yet.
Small-sized meter movements will be
400 moo
phased out but as yet solid state
devices are not commercially viable for

_ Fig. 1 (left). The

conventional mechanical
the large displays such as railway
platform indicators and expressway
instrument panel (upper)
contains 430 parts and is
signs.
much bigger than the Factors of key importance relate to
35-part electroluminescent the appearance of the display as seen
solid-state display (lower).
by the user, reliability, ease of servic-
Fig. 2. Comparison of the ing, and power consumption.
display technoligies now in Of particular importance is the
vogue. 'price to use'. This can greatly exceed

Optimum

Tungsten Filaments 1 - 20
Light Emitting Diodes LED 1 - 30
Cathode Ray Tubes CRT 10K - 250K
Gas Discharge ( Plasma Panels) 30 - 5K
AC or DC Electroluminescence DCEL 30 - 3K
Liquid Crystal Display LCD 5 - 200
Electrochromic ( liquid or solid) 5 - 200
Electrophoretic 5 - 200
Vacuum Flourescent VAC.FL. 10 - 100
SOO

46
 F I
1K

plays. The LED does not compare well

for applications requiring over 100
bits. On this basis the CRT is way
ahead. As yet it is not even remotely
100
matched by any solid-state technique.
The CRT's main drawback is that it is
%4
1 bulky and fragile compared with most
other types of display.
°D 10 I
cr • 4:: :)
LED GAS DISCHARGE TUBES
O Gas discharge tubes were the first dis-
e moat • us .
io NntAG/:
fA
plays that could reasonably be regarded
ogíli w••,4 ,-• eni•isiallx as versatile. Many older readers will
F .
IISI recollect the Dekatron counter tubes
of the 1950's in which 'dots' moved
GASLC
DDisic
circumferentially in a scale of ten. A
later development incorporated grids
10'
AUTOMOTIVE INSTRUMENT ANGE placed behind one another in a single
I
, ale

. J ,, , [ I •
glass envelope in what was generally
100 1K —K
. called the Nixie tube. Their main dis-
NUMBER OF BITS advantage for use with solid-state
Fig. 3. Comparison of the cost per bit of the various display options. circuits was their need for a 170 volt
day technology can provide 50 urn supply.
the cost price because of the costs of
square elements of which 300-600 With the introduction of solid-
power supply, mounting, wiring, and
may be integrated into a matrix. Such state displays it might be thought that
possible connectors. It is also impor-
LEDs are available with light output gas discharge tubes would have been
tant that at least two sources of supply
sufficient for aircraft instrumentation supplanted. This has not happened so
are available.
Another factor to consider is the (10 5 Lux) and can be made to full far. Indeed indications are that they
special characteristics of a display. MIL specifications including operation will be used for a considerable time
over a temperature range of — 55°C to yet. Their brightness and large size are
Each has some .good and some bad
characteristics. For example a liquid 125°C. still strong advantages.
crystal display is fine where ambient They are available in colours Gas discharge tubes are made in
light exists but needs auxilliary illum- ranging from red (the most common many forms. These include low
ination in low light conditions. LED's and cheapest) through yellow, green, profile, alpha- numeric, bar-graph,
on the other hand are best seen in the orange and violet. Blue LEDs have special purpose graphical displays, and
dark — they need to be very bright to been made but appear to lack a large still in the research and development
be seen in full sunlight. enough market to enable them to be stage, are phase addressed matrix
produced at a commerically attractive co- planar units which use thick film
It is also important to assess if the
device is really fully developed. Many price. manufacturing techniques.
new products reach the marketplace The reliability of LED sis variously It is possible to construct gas-
before they have been fully tested. claimed to be from 10 4 to 10 9 hours. discharge cells so that a particular cell
There is asuggestion ( based on evidence is set to strike and erase at different
Today a new solid-state product can
be realised and marketed in a matter from large-scale users) that price wars discrete voltage levels. Thus, increasing
have tempted makers to reduce relia- the voltage level to a line of adjacent
of ayear but it is not possible to test it
bility. Reject rates as high as 20% are cells will produce a bar-graph effect.
for the whole of that time. The tens of
said to be experienced by some Once struck, the cells latch on exhibit-
thousands of hours life that may be
ing abistable storage characteristic.
postulated by the manufacturer is buyers.
LED s are fast operating: typical Cross-bar arrangements of grids ( as
often merely conjecture. Liquid crystal
rise times are 10 - 50 ns. They are shown in Fig 4) enable 'dot' discharges
displays were one example. No user
offered in pcb packages, in larger to be established at the junction of
wants to be part of a test programme
metal packages suitable for sealing and any two selected bars. Thick-film
. . . especially if he's paying for the
in more-expensive-still ceramic pack- replication methods are used to
privilege. New is not necessarily best!
manufacture the units.
The main display contenders are ages.
The ready ease with which they can The colour of discharge tubes can
currently LED's, gas discharge tubes,
be assembled into lines, circles, matrices be finely tuned to just about any
cathode ray tubes, liquid crystals, and
and other graphical forms enables wavelength. This is done by adding an
the fast-emerging electroluminescent
them to be used in analogue displays. appropriate phosphor to the cell
panels.
LED s are not necessarily the best during manufacture. White and blue
The time-honoured filament lamps
choice for all displays. Figure 2 com- remain difficult to produce.
continue and need no further comment
except to say that they are being re- pares various displays on the basis of
placed in small power displays by the the optimum number of bits for each ANODE
BARS
SPACER PLATE
WITH GAS CELL
more up-to-date devices. alternative. It can be seen that LEDs APERTURES

are restricted to applications where the

LEDS type of display requires only a small
CATHODE BARS
LED's emerged first as single element number of bits. Electrolunninescent WITH INTEGRAL
RESISTORS -
light sources of rather low brightness panels ( discussed below) are more suit- 2000 A PLATE

and in red only. Today they are avail- able where the application calls for the
able in brighter forms and of many use of many bits. Exploded View
different optical styles providing diffus- Another factor is the cost per bit to
ing effects, wider angles of viewing manufacture. Figure 3 compares this Fig. 4. Construction of dot-matrix gas-discharge
variable for the various types of dis- display unit.
and generally greater utility. Present
47
 A phase-addressed technique has
been developed to reduce the number
of leads otherwise needed to connect
all matrix positions to external circuits.
Gas discharge tubes are still, and
will remain, a strong contender in the
choice of adisplay. Figure 2 illustrates
this well.

CATHODE RAY TUBES

A CRT screen of good quality and hav-
ing a good linear scanning system can
accommodate a display of 1000 by
1000 elements. The full range of
colours is available as well as an
intensity scale having perhaps 200
levels. Cost per element is very low
but size and fragility go against the
CRT in many applications. Eventually,
as matrix manufacturing methods be-
come more developed, the CRT's
thin flat digital equivalent will become
Fig. 6. On- vehicle electroluminescent display dash panel (Smiths Industries).
a serious rival. At present though,
the CRT has no rivals for displays pose manufacturing problems for the example is considering 1250 character
requiring large numbers of bits. glass enclosing the LCD material must displays for phone call costing. Smiths
be flat to within amere 10 urn. Industries have vehicle instrument
LIQUID CRYSTAL DISPLAYS Manufacturing methods are con- panels in pilot scale production.
In many ways LCD's got off to astart stantly being improved. The glass front (Figure 6 shows a recent panel of this
less worthy than they deserved. Relia- seal has been greatly changed . .. that type).
bility was variable: many failed rapidly was a cause of many premature As always, addressing the display is
whilst others did very well. Failure of failures. Purer LCD material and a problem. Multiplexing methods have
an individual display within a batch improved stability with temperature been used to reduce lead counts from
could vary from almost immediate and humidity have also improved. 257 ( for a 256 unit) down to only 32.
through to years. Matrix units are being investigated The displays can be used in a con-
The second generation of LCD s but, as with all such units, connecting tinuous mode or they can be pulsed.
has shown itself to be very much problems remain. Some LCD scurrent- Pulse durations of around 0.5% duty
better if made by more controlled ly being released have shift registers in- cycle are typical using 5 — 15 micro-
procedures and with better materials. tegrated onto the display. This trend second pulses.
Figures such as 90 000 hours to reach a may become common practice, for the Around 120 volts is needed to drive
2% cumulative failure have been user does not wish to connect any the display, present day units require
claimed for twisted-nematic LCD more than the minimum of leads from 50 mW per character. A prototype
displays. the drive circuits to the display. unit using CMOS circuitry consumes
A key factor has been the realis- In general LCD manufacturers only two watts for a 480 character
ation that a non- zero dc cell level suggest that their products are best display.
rapidly degraded the cell. That restric- suited for applications requiring a This information was compiled
tion was originally controlled by the portable display. The current LCD s from lectures delivered at an Institution
use of ac bias but now zero level dc are certainly much better than the first of Electrical Engineers day meeting
working has been devised. generation and their low power con- held in London in January 1978. No
Initial commercial incentive came sumption gives them a firm place in full Proceedings were published but
from watch manufacturers, but now the display range. the five speakers would be able to pro-
researchers are seeking ways of build- vide further information if contacted.
ing much larger panels — 150 mm ELECTROLUMINESCENT DISPLAY Details can be obtained from the Con-
square for example. Such large sizes PANELS ference Secretary, IEE, Savoy Place,
Electroluminescent devices are London. Smiths Industries kindly
basically just a layer of special paint provided most of the illustrations used
between two pieces of glass. here.
Two basic groups exist . . . ac
working ( called acel) and dc working REFERENCES
(called dcel). Each uses zinc sulphide, "Digital Display Systems" referred to
manganese- doped phosphors which in the previous part deals with blank-
radiate ayellow- orange light at 585 nm ing and multiplexing.
wavelength. The ac cells operate in a "Solid State Alphanumeric Display
capacitive mode, the dc units in aresis- Decoder/Driver Circuitry" — Hewlett
D.C. ELECTROLUMINESCENCE ASSEMBLY tive mode. Figure 5 shows the Packard application note 931 gives
1. FRONT GLASS 4. ALUMINIUM REAR ELECTRODE
schematic of adcel unit. details of scanning methods used with
2 INSULATING MASK 5. CONDUCTIVE STRIP — mining 4 to 8 Manufacturing processes are mainly 7x 5 displays.
3. PHOSPHOR 6. HERMETICALLY SEALED COVER
— in contact with 7 vacuum deposition using photolitho- "Mullard TTL Integrated Circuits —
7. TRANSPARENT CONDUCTIVE
through aperture, in 2 PATTERN graphic procedures for masking. This Applications" includes a chapter
Fig. 5. Schematic of a dcel electroluminescent method offers great prospects for the devoted to various kinds of code
display panel. future. The British Post Office for converter gating layouts.

48
 D-A and A- D

27 conversion

IN ORDER to control or modify the

physical world around us we must first
measure what is happening. The
measurement data is almost always in
analogue form, as is the actuation
required for control. Between
measurement and control some kind
of electronic system is needed to
amplify and shape the data.
We have seen that electronic systems
may be of either analogue or digital
form and it would seem best to use an
analogue system between inputs and
outputs that are both of analogue
form. But not necessarily so —
analogue systems are plagued with
problems such as noise, dynamic range
limitations, accuracy and linearity.
Digital systems, as well as offering
improved performance in the above
areas, offer more economical
processing of data, the ability to store
data as long as needed, and more Fig. 1. Typical A/D converters.
readable displays of data held within
the system.
By use of certain input combinations the Voltage- to- Frequency or VF
Thus there is much to be said for
converting primary analogue signals DAC's can also be used as converter.
into equivalent digital forms that are multiplier/dividers of two signals and The uses for A/D and D/A converters
as summing/subtraction units. It is also are limitless. Their application is
pro cessed and stored etc until
conversion back to analogue form relevant in this part to deal with ever-increasing as the unit cost falls to
becomes a necessity. Electronic sub- multiplexers in a little more detail and undreamed-of prices. Extreme
systems that perform these conversions with the so-called sample- and- hold complication using digital techniques
are called Digital- to- Analogue circuit. These are often used in often costs far less than simpler but
Converters ( DAC's or D/A converters) conjunction with DAC's. less accurate analogue alternatives.
and Analogue- to- Digital Converters Figure 2 illustrates the basic Hence D/A and A/D converters will be
(A/D converters). ( See Fig. 1) requirements of a typical found in digital panel meters, digital
We will see that these are quite data- acquisition system in which a multimeters and data acquisition
complicated systems in themselves — number of physical variables are systems. They are also found in
their design a skilled task. measured and processed to provide industrial plant; in process control of
Nevertheless, many such sub systems digital signals for storage. It uses chemical and other manufacturing
are now marketed as single, largish multiplexer, sample- and-hold, and A/D plant; in telemetry systems and other
circuit blocks that are wired into the converter sub-systems to form the data transmission applications; in the
whole. interfaces ( units matching the output
total system in the same way as other
complicated system building-blocks we Also pertinent, because similar signal requirements of one system with
techniques are involved, is the method the input requirements of another)
have already encountered. It is,
for converting an analogue voltage to a found between sensors and computing
however, important to understand the
signal of proportional frequency units; between stages of hybrid
basic techniques used if not so much
(which is a form of digital signal) — computers; and the like. Although
the refinement of actual practice.

PHYSICAL
SAMPLE
VARIABLE AMPLI- ACTIVE
TRANS- AND
^-11111.— Fl LTER A/D '""-- DIGITAL
DUCER FIER HOLD
CON- OUTPUTS
VERTER
ANALOG
MULTI-
PLEXER

OTHER CONTROL
ANALOG INPUTS
CHANNELS PROGRAMMER
Fig.2. Typical layout of an analogue data SEQUENCE
acquisition system interfacing to adigital
output.
49
 ELECTRONICS it's easy! needed the currents feed an op- amp.
The detail of a precision ladder
network is shown in Fig.4 — it is set to
convert the input digital number
Fig.3. Schematic of adigital to analogue 11010001 by appropriate positioning
converter which provides a current or voltage of the digit position switches. When at
output. To obtain a voltage output the
0 input the inputs to the summing
current output is fed to an op-amp.
CURRENT
op- amp are held to ground; when at 1
OUTPUT to an appropriate stabilized voltage.
The simplest form of ladder is used
in the circuit of Fig.4. It arises from
the use of a binary weighted resistor
sequence shown in Fig.5. The actual
VOLTAGE
DIGITAL DIGITAL ELECTRONIC PRECISION
LADDER
OUTPUT values of resistors are selected to
INPUTS INTERFACE SWITCHES
NETWORK obtain adequately sized lowest and
largest values, for at either end the
PRECISION op- amp Ioses accuracy due to
REFERENCE
imperfections of resistance ratios. It
can be seen from the circuit of Fig.4
that resistors, even in a smallish
capacity 8 bit converter, can extend to
extreme values. The least significant
INPUT DIGITAL WORD = 11010001
bit must be clearly resolved when its
REFERENCE E = . 5.00V switch operates, implying that all
other resistors must have precision of
absolute value and constancy with
time that rises very rapidly with the
number of bits required.
In practice this simple form of ladder
is not used beyond about 8 bits
conversion due to the cost of the
precision resistors required.
The disadvantages of the simple
ladder method are mostly overcome
by the use of the R- 2R ladder network
shown in Fig.6. The through leg of the
chain is permanently grounded, each
spur is switched as needed to a
reference stabilised voltage level. The
features of this method are that only
Fig.4. Switched resistor two values of resistors are needed ( an
network of 8-bit, binary easier practical problem) and that the
weighted DIA converter. absolute range seen by the op- amp
11010001 is being
converted.
varies much less than the above
method for a similar bit capacity — it
presents a virtually constant
impedance regardless of the binary
code sequence switched in. With the
R- 2R ladder it is routine to provide 12
bit conversion.
Vout
It is probably obvious that other
forms of digital-coding conversion can
be handled by the use of appropriate
resistor weightings. For example it is
often necessary to chart- plot the
Fig.5. Basic binary- output of a digital instrument. Thus a
weighted resistance BCD to analogue converter is required
ANALOGUE INPUTS FROM SWITCHES
sequence.
for such applications. Figure 7 shows
the weighting sequences for the simple
highly accurate DAC's may cost as and the R- 2R ladder DAC's needed to
You will remember that each digit
convert BCD inputs to an analogue
much as $ 2000 the great universal position of a binary number ( held in a output.
demand has resulted in lower register, counter or other form of
Resolution and accuracy — DAC's
performance units being available for storage) has a weighting factor, eg
as little as $ 10. rarely go beyond 12 binary bits ( or 3
1:2:4:8 etc. An analogue signal
digit BCD) because the output
equivalent to the binary number can
DIGITAL-TO- ANALOGUE analogue signal for greater bit- ranges
be obtained by using each digital digit
CONVERSION must be of high stability. A 16 binary
position to switch an amount of
bit ( or a 4 digit BCD) unit could
We begin with Digital- to- Analogue current ( proportional to the position
provide ± 0.005% full scale linearity
converters because they are the weighting) to a common summing
and accuracy, a performance
simplest in concept and use only one junction. This system concept is
requirement that is best avoided where
basic technique. shown in Fig.3. When voltage output is possible because of the high cost of
50
 R
the DAC. By contrast 8 bit DAC's can
be obtained with accuracies ranging
from ± 0.2% full scale to ± 0.01% full
28
scale. It is important, however, to
realise that whereas analogue
resolution, see Fig.8, is a function of
the number of bits that are equivalent
to full scale, the accuracy and linearity
o
ERe
of DAC's depend upon the tolerances
DIGITAL SWITCH INPUTS
and stability of resistors used in the
conversion networks, for these decide Fig. 6. The R-2R ladder network requires only two values of resistance and presents constant
the value of the slope and straightness impedance to the output.
of the slope — Fig.s 9a, 9b. It is,
therefore, possible to have an highly
accurate converter that has quite
coarse resolution — in which case the
resultant analogue output signal will
consist of very large step changes. This
step form of signal defect is called
quantum or quantization noise. In
practice resolution and accuracy are
tied together keeping quantum noise
to an acceptable level.
R - 2R
Conversion and settling times — As
both DIA and A/D conversion are
dynamic processes, a finite amount of
time is required for each conversion
point to reach its final value. In DAC's
the switching and settling times of the
op- amp largely dictate the time for a
bit change to finally appear as a
steady-state analogue signal level.
Early DAC systems using mechanical
switches were slow indeed — today
output settling times range from a DIGITAL SWITCH INPUTS

slow 25ps for very- low power

consumption units to ultra- fast 25ns
units.
Fig. 7. Resistance layout for simple BCD-
Temperature coefficient — Each
weighted DAC.
subsystem of aDAC has atemperature
coefficient; resistances alter with
temperature and the op- amp
characteristics deviate. Both the
overall conversion gain and the dc zero TO DIGITAL INPUT
SWITCHES
will be affected.
Gain will be affected due to the
temperature coefficient of resistors 28

which is typically from 50 to 100 are shown in Fig.10. If the output is

PPM/ 0 C ( 100 parts per million, PPM, only to be observed after the system MSD 48

is equivalent to 0.001% change per has settled, these matter little. In

SR
degree Celsius). The main op- amp dynamic use of DAC's however, they
characteristic which affects may well excite unwanted behaviour
performance is offset- voltage drift — in the system they are driving. 10R

typically 30 microvolts per degree Deglitching in DAC's is not feasible

with a low-pass filter on the output, SIMPLE
Celsius.
For each particular type of DAC it is for glitches vary widely in nature. The
necessary to consult the makers' best solution is to use adequately fast i 408
2OR

specification sheets, for no general and matched switching coupled with

808
special deglitching, ( sample- and- hold)
rules apply for these parameters.
DAC's are available as either current circuits, that hold the output fixed
or voltage output systems. Typical during unwanted switching — transient 100R

outputs deliver around 3 mA and conditions. Glitching states are,

however, known states and are quite 2008
10-20 V swings.
Further explanation of terms used is unlike random noise which defies .4--/VW- •
LSD 4008
to be found in the articles listed in the prediction.
.1"
--/VW- •
further reading section. Integrated circuit current sources —
8008

Glitches — Certain digital input As the DAC principle finds avariety of

states, whilst in a transient state, can uses, manufacturers offer an integrated
cause the output to produce circuit which provides an output
noticeable transients to the smooth, current, the magnitude of which is
controlled by a four bit, binary- code Fig. 7. Resistance layout of R2R. BCD
stepwise analogue — signal progression.
ladder D.A.C.
These are known as glitches; examples input. The IC, as shown in Fig. 11, has
51
 ELECTRONICS it's easy! method are the
speed available.
high precision and

ANALOGUE Addition — If the difference or sum

V OR mA of two digital signals is needed as an
10 2.00
IDEAL STAIRCASE GENERATED AS analogue output, two DAC's may be
DIGITS PROGRESS combined, as sum or difference, into
8.76 1.75 the output op- amp, as shown in
STEP TOLERANCE Fig.14.
VARIES
7.5 1.50
ACCORDING
TO ACCURACY ANALOGUE TO DIGITAL
OF DAC.
6.25 1.25 CONVERSION
Conversion from analogue to digital
5.00 1.00 code can be obtained by many
alternative techniques, each alternative
3.75 0.75 having many variations. Basically,
methods group into open- loop and
2.50 0.50 feedback- loop systems. In each group
some four to eight ways are in
1.25 0.25 common usage. Here we look at afew
Fig.8. Output/ of the most popular tech niques,
input characteristics
beginning with open-loop methods.
0 0 of typical 3-bit DAC
000 001 010 011 100 101 110 111 having full-scale Analogue- to- frequency — The
current of 2 mA or analogue voltage is converted, on a
BINARY DIGITAL CODE INPUT STATES 10 V output. continuous basis, into a signal of
proportional frequency by the use of
an appropriately accurate V-to- F
ANALOGUE
OUTPUT converter. ( Voltage controlled
oscillators are used — they must be
10000000
LINEAR BUT if adequately linear). This signal, see
INACCURATE Fig.15, is gated into a digital counter
01111111
using fixed times of gate aperture. The
—
— counter accumulates a digital number
MINOR
MAJOR
equal to the average analogue level
GLITCH
IDEAL GLITCH over the gate period. The counter
ACCURACY output is released upon demand when
gate periods expire. Th is method
WITHOUT I WITH suffices for low accuracy analogue
DEGLITCHING I DEGLITCHING
inputs but becomes expensive when
precision, wide dynamic range is
required.
Fig. 10. Glitches occur in basic DAC's due to
certain transient dynamic error conditions. A variation of this is to reverse the
000 110 They are eliminated by deg/itching circuitry. philosophy and vary the pulse width
DIGITAL INPUT STATES
of accurately generated pulses of
(a)
opposed to analogue power control. If constant frequency. The variable pulse
ANALOGUE
larger output powers are needed width is then converted to a digital
OUTPUT digitally- controlled power supplies can count proportional to the pulse width.
NON-LINEAR be used — also referred to as digitally Simultaneous, parallel or flash
ACTUAL programmed supplies. They are conversion — The input analogue
available with digital control of signal is presented to a stack of
LINEAR current or voltage outputs. comparators ( Schmidtt trigger action)
IDEAL Instruments in the Hewlett-Packard each set to trigger at increasing
range, for example, can provide up to binary-weighted signal levels. The set
125 watt maximum demand whilst the of comparator outputs are then
output is controlled by binary or BCD decoded to provide the required
inputs with a programming time of binary output form. Decoding is
around 350µs. needed because at any instant all
comparators set to below the signal
000 110 level are in the one state, all above in
DIGITAL INPUT STATES
ARITHMETIC OPERATIONS the other. The method is given in
WITH DAC's
(b) Fig.16. Although extremely fast —
Multiplication — The resistor quantization time is the speed of a
Fig.9. Output/input definitions for DAC's network of a DAC has two inputs — single comparator — the method has
(a) deviation of accuracy.
the reference supply (which is fixed in the serious disadvantage that large bit
(13) deviation of linearity.
normal DIA conversion) and the ranges require many comparators and
switch inputs representing digital numerous decoding gates.
enough precision to be used in 12- bit numbers. If the reference is allowed to Closed-loop methods — are more
D/A and A/D conversion. Figure 12 vary as an input variable, see Fig.13, popular and there are about six main
illustrates how two of these ICs are the output of the DAC is the alternatives The methods known as
combined to produce an 8-bit DAC. multiplicand of the two signal inputs. integrating, successive approximation
Digitally controlled sources — The The reference may also be an ac signal
and servo-DAC are most generally
DAC's described above are usually and division and attenuation are also appl ied.
concerned with signal processing as possible. The advantages of this
Dual- ramp integration — The
52
Fig. 11. Schematic of dual- 0 E• 04 0 7 ONSÉ

v•
in- line 4-bit current source
(Fairchild p.A9650).
o

0 10

,•I

04 0,, 0

4•00.40
o"•"

•S4
Out
0 .50
0o'

analogue voltage is first converted to a

time period which in turn is converted DIGITAL INPUTS

into a binary number by a

?Uri)
timer/counting system. Referring to T î
2
î
3
î
I 2 3 4

Fig.17a, conversion begins when the 12 yAf1650 .A3,11

17 rA8650
312
switch connects the analogue-signal 313 313

i
••
input to the integrator which 11 13 14 IS e
13 14 IS

commences to ramp up. At the same

lo 0k
110111:104 201, 80 40 It eo.
time the counter begins, from zero, to
II V
count the clock pulses. When a
auno 2 DIVIDER
predetermined number of pulses ( 1000
is convenient) appear in the counter
the integrator is electronically
;ULL SCALE AM,
switched over to the reference. At this
point the capacitor has then charged
1 4.
In
••••&
SA/41

linearly from the input, rising as a 312

ramp to a voltage level decided by the

average input-signal value as shown in
Fig. 12. Applying 4-bit current source IC's to form an 8-bit DA C.
Fig.17b.
As the switch changes to the
VARIABLE BINARY INPUT (SIGNAL INPUT AI
reference position the counter is reset
to zero and begins counting again. The
reference, chosen to be of opposite
polarity to the input signal, now
causes the charged capacitor of the
integrator to ramp back downward at
a constant slope. When the integrator
output reaches the zero threshold the
VARIABLE REFERENCE ANALOGUE
counter is stopped and its contents INPUT (SIGNAL BI OUTPUT
displayed. The count displayed is the AXB
ratio of downward ramp counts to
Fig. 13. Using aDAC to multiply an analogue signal by adigital signal.

DIGITAL INPUT A

î
ANALOGUE VOLTAGE TO
INPUT FREQUENCY
CONVERTER AND
SERIAL
—•
DIGITAL
SUM i
DIFF FIXED OUTPUTS
TIME COUNTER
; +0
BASE
R
EFERENCE INPUT

OUTPUT A+B RESET

OR A - B

PARALLEL
o DIGITAL
OUTPUTS

11DIGITAL INPUT B
Fig. 15. Voltage- to- frequency method
R
EFERENCE Fig. 14. Addition or of A /D conversion.
subtraction of two signals
with DA C's.
53
 ELECTRONICS it's easy! upward ramp counts which, when a
1000 upward limit is used, gives a
direct reading of input voltage if the
reference voltage is appropriately
chosen.
A simpler form, using only one ramp,
is also used but it lacks the features of
INPUT COMPARATORS SET AT INCREASING the dual- ramp method in which the
3R/2 s LEVELS BY RESISTOR NETWORK absolute value of the capacitor and the
clock frequency are of no significance
provided they are stable for the
duration of the conversion period. The
dual ramp method does, however,
require a relatively long conversion
period but this is an advantage in one
respect — the value measured is more
accurate. This is due to the fact that
BINARY
OUTPUT when noise is integrated over an
extended time period it tends to zero.
A more sophisticated triple- ramp
method provides increased speed and
accuracy for a moderate increase in
cost and complexity. In essence two
reference signals are provided, one
acting as a ' coarse', the other as a ' fine'
ramping control. The ' coarse' ramp
Fig. 16. Flash type converters rapidly converts the bulk of the input
give high speed A/O conversion.
signal level leaving the ' fine' ramp to
add the extra resolution. Figure 18
shows the schematic and timing
diagram of a triple- ramp AID
ANALOGUE
INPUT SWITCH converter.
o Successive approximation — Due to
its high resolution and fast conversion
speed successive approximation is the
most widely used method. A
schematic diagram is given in Fig.19.
Conversion progresses step- wise with
the precisely generated DAC output
being compared against the unknown
analogue input. The first comparison is
made with the most significant digit of
the DAC, which gives 1/2 full scale,
being compared against the unknown.
If it is smaller, the bit is retained as a
'0', if larger it is set to ' 1', thus the
(a) MSD value of the programmer is
fo und. The next digit, working
towards the least significant end one
DIGITAL OUTPUT digit at a time, is then tested for the
same criteria being set accordingly.
The process is repeated until all
Fig. 17. A/O conversion using dual- ramp integration. (al schematic (b) timing diagram.
programmer digits are set to ' 0' or ' 1'.
SLOPE ' UP DECIDED The value in the programmer is then
BY VII,
transferred to the register for
vo SLOPE ' DOWN 'DECIDED
BY Vre f outputing in parallel or serial form.
Conversion time is not decided by the
, value of input as in ramp methods,
Win = t Vrefi
duration of conversion being the
INTEGRATOR
LOWER VALUE OF Vi n
number of bits times a fixed digit test
interval, which can be as fast as 100ns.
By comparison, from one maker's
options, successive approximation
T te.i
—0- TIME instruments offer conversion times
which range from 1-60,us compared
with 2.5-6.0ms for integrating
converters. Accuracy clearly depends
COUNTER
upon that of the DAC which forms
1000 PULSES PULSES PROPORTIONAL TO t
part of the comparison system.
Servo-DAC method — Fig 20 shows
(b)
this system. When conversion begins, a
counter is gated and commences to
54
 Fig. 18. Triple-ramp A/D converters offer
faster speed for slightly more complexity
of integrating method.
COARSE
INPUT RAMP
RAMP

FINE
RAMP

SWITCHES o TIME

-V,..
ICOARSEI

T
=S
I
a

-=
1
1
-
INTEGRATOR
COMPARATORS

.-HR«
SI 52 53
tr" CONTROL
CONTROL
LOGIC
h
EXTERNAL

CLOCK

¡File
t
COUNTE f /UNTER
III 2

OUTPUT OF DAC
count upward. Its digital output is OD—AAM— .
converted back to analogue form by a ANALOGUE
INPUT ANALOGUE
DAC. The output of the DAC is INPUT
compared against the unknown input
voltage. When the two analogue
voltages are equal, the comparator I- COMPARATOR,
OUTPUT RESISTOR
1 BIT TIME
inhibits the counter. At that time the
value in the counter is a digital TIME
representation of the input — with 1:1 DIGITAL
correspondence; or other ratios
OUTPUT 1 01101111

depending upon the summing Fig. 19. In the successive approximation A/D converters aDAC is used to convert the
resistances used. It is asimple low-cost incremented digital stage output back to analogue form for comparision with the input.
method providing reasonable accuracy
but operates at a slower speed than
ANALOGUE
INPUT COMPARATOR
offered by successive approximation
designs.
Non-linear conversion — Each bit of
the above methods represents an equal
ANALOGUE
quantum error. Thus one quantum INPUT
error in full-scale is considerably less
inaccuracy than in say a tenth or
hundredth of full-scale. The smaller
the reading, the greater the relative
error of quantisation. When
range-changing is not practicable a
non-linear digital method can be used
to compress the large scale in order to TIME
reduce the percentage of reading error OUTPUT FROM DAC
The method is explained in Motorola
Fig.20. Schematic of servo A/D converter system.
Application Note AN-471.

signal and holding it in a simple value in a way that can be rapidly and
SAMPLE- AND- HOLD UNITS analogue store long enough for the effectively isolated from the source. In
A digital signal provided by an A/D converter to act with a much wider practice low leakage FET switches are
converter represents some measure of aperture time. Figure 2 shows such a used in conjunction with IC op- amp
the analogue level seen in a certain system. To preserve the highest signal integrators. It is also important to
gating period — the so-called aperture frequency of acomplex signal we only buffer the output of the
time. Aperture time, bit resolution and need to sample at twice ( or higher) the sample- and-hold to reduce the loading
maximum signal frequency are strictly signal frequency ( Shannon's sampling which would otherwise decay the
interrelated. Figure 21 is a chart theorem). This is aconsiderably slower stored level.
enabling this characteristic to be rate than needed for direct A/D Many circuit variations exist, the one
found. For example, we may need to conversion. If sampled too slowly, not shown in Fig.22 — a closed loop
digitize a 10 kHz sinusoid ( as the only will higher frequency information configuration — gives good linearity
highest frequency to be preserved in a be lost but an effect, called aliasing, and accuracy. When extremely long
complex waveform) to a resolution of will occur by which alower frequency storage times, or negligible decay with
12 bits. The chart shows we must have is generated that may not exist in the time is needed, the voltage on the
an aperture time of no greater than capacitor can be transferred via an
original signal.
42 ns. Thus we see extremely fast A/D converter into a digital storage
As mentioned above sample- and- hold register and back again into analogue
converters are needed for direct
circuits are also used in DAC's to form via a DAC as shown
conversion of moderately
remove glitches. schematically in Fig.23. This naturally
high- frequency signals at high
resolution. Basically a sample- and- hold increases the cost considerably. More
A sample- and- hold circuit comprises a capacitor with which to detailed information is available in
circumvents this difficulty by taking a store an analogue voltage level, and a "Analog-digital conversion handbook"
rapid narrow- aperture sample of a switch to charge the capacitor to that TH6, by Analog Devices.
55
 ELECTRONICS-it's easy! MULTIPLEXERS
The task of the multiplexer, shown
in Fig.2, is to sequentially connect a
WAVEFORM BEING
multiplicity of compatible inputs to a
DIGITIZED
AMPLITUDE
single output line. In the case given in
UNCERTAINTY Fig.2 it feeds asample- and- hold, which
stores the signal for A/D conversion.
A multiplexer consists, therefore, of
APERTURE TIME as many switches as there are input
channels to be combined. In practice
these must possess adequate speed and
I
mS
very low on- to-off switch resistance
ratio. Solid-state multiplexers mostly
UIIlIIIUHIIIUUIHIIIuuuuIIIUuuuuflI use MOSF ET switching devices feeding
10 0p S NOIVIIIIENIIIC1111111111111.111111111111111111 a buffer stage ( a voltage follower
configuration which has extremely
mcànmizracerau
1W1111111.”•11\11111111CII muumuu' high, 10 9 ohms, input impedance).
010111111.1111;!Illa, I11111•111111111 Fig.24 shows such an arrangement.
Fig.25 is the internal schematic of an 8
Iheillb • •
11111011i ' 101111111; channel unit which has an ON
resistance of 300 ohms and OFF
1µS NIVIIii ll eS111 1111
aim'IZi• ate resistance of 100 megohms with
Immminm. •••
MU» .. wesne "1.:::(;1e"'Cree=6,48,11, 0.03nA leakage. The settling time to
IMUKIIIMIË1111111C111»:14111
0.01% of value is 1p sec and it can be
100nS 11111111d
œ=. sequenced to 500 kHz. Channel
selection is made via the appropriate
iIIIIIIMIIIIIIIIMMIIIIIelloillells.'11118311:11111MUI hi
digital- code input.
Fig.21. Graph showing
relationship between
aperture time required,
1nS 1111111111111111111111..11111111M 10 11 0 110111111 resolution of signal
10 100 lk 10k 100k conversion and frequency
of sinusoidal signal
SINUSOIDAL FREQUENCY ( Hz)
undergoing conversion.

o
o

FAST GATED c INTEGRATOR

SAMPLING SWITCH
ANALOGUE .41(
INPUT

ISOLATION
ANALOGUE
SWITCH
OUTPUT •

0—MM—r
1 e
ANALOGUE
INPUT

Fig.22. One form of stable and accurate sample -

and -
hold circuit.

Fig.23. Infinite hold is obtained by transferring the

integrated value into adigital store using A/O and
D/A conversion.
MULTIPLEXED
ANALOGUE OUTPUTS
INPUTS Z = 10 9

CHANNEL 1 OUTPUT
BUFFER
50 pF

SWITCH INPUT

Fig.24. Solid state multiplexers use

MOSFET switches to feed the inputs
sequentially to the output via abuffer
stage.
56
 REFERENCES
A comprehensive discussion of the
topics, and along bibliography, of this
+5VDC GRD + 15VDC — 15V (- 20VDC) part is to be found in " Analog-digital

1
conversion handbook", D.H.

I
14
) 1
)12 1)
13 Sheingold, Analog Devices, U.S.A.
ANALOGUE 1972. Less extensive but nevertheless
INPUTS MOS FET SWITCH WITH DRIVER
ce :)_.__________<î)_______ very useful articles are — " Engineering
CH1
ANALOGUE product handbook", Date! Systems,
9 OUTPUT CAT-T99405, 1974, U.S.A.
CH 2 CH1THROUGH14
"Analog- to- digital conversion
0
8 6 techniques", E. Renschler, Motorola
CH 3
Semiconductor Products Inc., AN-471,
1969, U.S.A. " Product Guide",
CH 4 Analog Devices, 1975, U.S.A. •

CH 5
ANALOGUE
3 OUTPUT
CH 60 CH5THROUGH8
0
5

CH 80 1

DECODER SECTION

16 17 18
ANALOGUE OUTPUTS
MAY BE SEPARATE
OR CONNECTED
2 4 TOGETHER
8 /
Fig.25. Internal schematic of Datel 8 channel,
V
CHANNEL SELECTION MM-8 multiplexer.

Measurements on digital circuits... the logic probe

In analogue circuits, performance
is checked or faults found, using, at
the very least, amulti-meter. It is used
to check the voltage levels at various
places as they can provide much in-
formation about the operation. An
oscilloscope is very much more useful
but is not always available. The simple
analogue reading meter can provide a
good start.
In digital circuits, voltage levels con-
tain far less information about circuit
condition as information is contained
as one of just two voltage levels. For
digital work, another simple tool IC is any 7400 series.
exists. It is called the ' logic probe'.
These are usually self-contained units
— they may require an external voltage
SPECI FICATIONS
supply — and are used to check the
Supply voltage 5 Volts dc ± 5%
logic state of a circuit at a chosen
Input voltage ( high input) > 2.4 Volts
point. As an example consider the
circuit given here ( from ETI project Input voltage ( low input) < 0.8 Volts
Indication — steady positive lamp on
113).
Connection of the probe input to the Indication — steady ground lamp off
circuit point will cause one of five Indication — fast positive pulse lamp flashes on
circuit conditions to operate the in- Indication — fast negative pulse lamp flashes off
built lamp in a different mode. It is Indication — pulse train lamp glows less brightly
suited for use with logic of RTL, DTL Minimum detectable pulse width 50 nanoseconds
and TTL systems. It is not necessarily Extended indication of lamp 100 milliseconds
useable on CMOS logic as that can use
as wide a variety of levels as the
designer selects. Specifications of supply and operation for logic probe.
57
 Digital instruments

28 and test equipment

exception is counting — a digital

procedure). Hence the digital
instruments that are used to measure
real-world variables usually have an
analogue-to-digital converter at the
input. However there are many
modern instruments designed to 'see'
into logic circuitry — such instruments
(having direct digital inputs) are truly
digital.
Many test procedures could be
implemented using solely analogue
circuit techniques. However, there
now exists a definite trend to replace
analogue techniques with much more
complex, but cheaper, digital
equivalents.
Digital measuring equipments range
in size and complexity from the simple
panel meter for measuring voltage,
current or resistance, Fig. 1, through
medium complexity, portable and
highly- flexible digital instrument
systems, Fig. 2, to large automatic
testing plants, Fig. 3, which operate on
the commands of in-built computers.
Today, many testing instruments
(even quite small units) possess
self-testing facilities, in-built diagnostic
ability and other advanced capabilities
such as the automatic readjustment of
the circuit under test to bring it within
quoted specifications.

SOME HISTORY OF DIGITAL

INSTRUMENT DEVELOPMENT
We have seen in Part 21 that most
analogue signals can be converted to a
digital equivalent. This concept first
found serious economic application in
computational systems, as was briefly
explained in Part 22.
In the early days of digital systems
even simple equipment had to be built
using large numbers of thermionic
valves and electro- mechanical relays.
In the early 1940's the power
requirements and the sheer bulk of
such systems severely restricted the
application of digital techniques to
This compact hand-held digital multimeter is produced by Hewlett-Packard.
computers. Hence the early uses of
digital systems in testing and
evaluation first appeared in
IN THE previous two sections we Digital instruments may be defined applications where portability was not
discussed the basic building blocks of as those in which the major proportion required.
digital systems. We are now in a of the circuitry is digital rather than The need for an adequately- fast
position to study how these blocks are analogue. The circuitry of such training simulator for aircraft pilots
assembled into specific types of instruments is seldom all digital, as few led to the development of the
general-purpose instruments and test of the natural processes we require to 'Whirlwind 1' computer by MIT in
equipment. measure are in digital form ( anotable 1946. This evolved from work
58
 scan controlfer
(El

CALte , Ar,

273 • digitai printer

Fig. 1. Digital panel meters are extensively

used in equipment, often replacing the
moving needle indicator meter.

Fig. 2. The H.P. 3470 series measuring system

consists of modules that clip together to
form the digital multimeter instrument
system required.

previously done on the use of digital consists of analogue input, function traditional multi- meter measurements.
arid range selection stages which feed Frequency measurement, timing,
techniques in the fire control of guns.
an AID converter; this drives the counting, totalizing, and ratio
Digital voltmeters first appeared in
the mid fifties and at least two digital readout form of display. They measurement may all be offered.
are marketed in awide range of forms Readout of any physical variable is
companies, Non- Linear Systems and
— from the neat compact fixed possible provided a sensor exists that
Schlumberger, lay claim to being first
purpose units like that of Fig. 4 to may be correctly interfaced with the
in the field. In the period 1955 : o
1960 new skills, developed in the systems such as shown in Fig. 2, in digital end of the measurement
manufacture of computers, were used which modules are interchangeable to system.
suit the very wide range of possibilities The simplified block diagram of the
to build smaller cheaper and more
effective instruments thus. that the common power supply and very compact H.P. 970A a hand-held
commencing the swing from aralogue cigi taI readout allow. Functions digital multimeter, is given in Fig. 5.
to digital instrumention. The offered ( by choice of appropriate The circuits are of analogue form until
introduction of transistors enabled modules) may include more than the the comparator stage after which
large-scale digital process control to be
realised — Texaco Refinery at Port
Arthur, 1959, was the first. They also
allowed portable digital
instrumentation to be made.
In the early 1960's integrated-circuits
were conceived. The cost and space
savings gained with their use removed
any doubt that digital techniques were
not competitive with traditional
analogue methods.

THE DIGITAL MULTI- FUNCTION

MET ER
We have already dealt with the basic
sub-systems used to form most digital
instruments. They usually comprise
various assemblages of digital displays,
A/O and D/A converters, counters,
registers, gates and sources of precise
frequency signals. Fig. 3. Automatic Wiring Tester VD30, by Siemens — it handles 12,000 connections.
A digital multi- meter, for instance,
59
ELECTRONICS-it's easy!

Fig. 6. Digital instru-

ment circuits are often
manufactured as
custom-built chips
(al Digital chip of
Fig. 4. This digital multimeter, by Non- HP 970A is just 3.9
Linear Systems, fits in the palm of the hand. by 4.3 mm in size.

digital signals are used. This unit uses multimeter and CRO functions within COUNTER- BASED DIGITAL
the dual-slope integration method to the one case. Such an instrument is the INSTRUMENTS
convert from analogue to digital. Most Tektronix type 213 unit as illustrated
of the circuitry is manufactured on in Fig. 7. This instrument displays Combining a display with a suitably
just two custom-made monolithic IC either the waveform or the gated counter and timing system
chips - Fig. 6a shows an assembly scan- generated multimeter provides the ability to count events;
with 40 flip-flops, 19 MOSFET measurement as required. totalize and indicate elapsed time;
switches and some 3500 bits of ROM determine frequency and period of
(read only memory); Fig 6b shows the A modular system approach is more periodic waveforms or provide a
chip which carries the bulk of the expensive but gives versatility and time-clock. In these options little
linear circuits used. continued flexibility to adapt to analogue circuitry is involved, the unit
Multimeter measurements of changing needs by not- so- expensive either generating its own digital signals
alternating current or voltage are only additions. Figure 8 shows one maker's (a clock) or operating on input signals
meaningful if the waveform is modular approach - it retains the that are already in digital form.
sinusoidal. If the waveform is complex aesthetic shape of acomplete portable We have dealt with the internal
it is necessary to use an oscilloscope to set regardless of the number of operation of counters in Parts 24 and
gain knowledge of the ratio of peak to modules needed at one time. The 25: here we expand their use by
average or rms as measured by the wasted space and extra weight penalty studying the various modes of
multimeter. There are instruments of the slide- in modular package is operation possible with a basic
available which incorporate the thereby avoided. counter.
1V REF

k (.

C3

o
Cl
o
10M
H1 •
(TIP) V- COMPARATOR
o

r
—"AAA-0
+ REF
10M
FUNCTION INTEGRATOR

- REF -

o
L0
(CLIP)
AUTOZERO
1
DISPLAY COUNTER

-T-
Fig. 5. Schematic of HP 970 A digital multimeter. ICONTROL
,LOGIC

60
 TOTALIZING AND BATCHING
Power Supply NOS Chap
AC to DC
Converter
Amplifiers (Logic end SwItches) This is the simplest use of acounter.
Comparator
Events to be counted ( for example
1. t L packages on a conveyor belt may
• intercept an optical link, thus causing
é an electrical pulse to be generated each
I
time the beam is broken). These pulses
enter the counter, ( Fig. 9) and are
I I.
shaped into clearly recognisable
I P.
counting pulses. Whether or not an
event is counted is decided by the
I condition of the input gate which can
be opened or closed on electronic
command. In many applications the
.1 Ft'
Fig. 6(b). gate is quite simple, but its design can
hybrid of HP 970A be a major problem when very fast
carries much of the
integrator Input J FOI
signals are to be totalized to
Amplifier Laser Trimmed analogue circuit on
Amplifier Resistors
its 28 x 38 mm high-accuracy.
substrate. A batching counter goes a little
further in that it counts to a
predetermined value. When this is
reached it provides an output
command to the process being batched
(for example, tins being counted into
cartons) which causes some change in
the process. At the same time, if the
process is repetitive, the counter is
reset to the starting value ready to
count the next batch. It is sometimes
more convenient to count downward
from the number required, operating
the batch command at the zero value.
More complicated batching systems
may have a stored program that sets
each batch sequence to varying count
values.

TIME INTERVAL MEASUREMENT

It is possible to measure the

time- interval between two events by
feeding pulses of known time
separation into the input of the
counter, as shown in Fig. 10, where a
Fig. 7. The digital multimeter function is combined with aC.R.O. facility in the Tektronix
213 unit.
clock drives the counter. An example
is the timing of a race. The gate is
Fig. 8. Compact HP5300 series of modules allows many counting/timing measurements to be opened at the starting signal and
made. stopped at the end. Counts
accumulated represent the time
interval. Obviously there is an
advantage in choosing a pulse
repetition rate that suits the units of
time being used. The choice of clock
frequency therefore depends on the
resolution needed. For example, to
measure one second with a resolution
of 1 in 10 6 a pulse frequency of
1MHz is needed to gain 1
discrimination.
In some applications acommon gate
— control input is suitable — where the
on and off event reproduces the same
situation such as in period
measurement of a sine-wave signal.
Often, however, two separate input
channels are needed so that each can
have the specific pulse conditioning
needed by different signal sources.
Race timing, for example, might
initiate the count on the sound of the
starter gun and stop it on the signal
from apressure- pad sensor.
61
 ELECTRONICS it's easy! DISPLAY

SCHEMATIC
DIAGRAM HOLD
DISPLAY

COUNTER
HOLD

CONTROL_. SCHEMATIC
COUNTER CLOCK DIAGRAM
OUTPUT '« I COUNTER
BATCH

RESET

TRIGGERS
SHAPING
0 e TRIGGER
EVENTS CONTROL GATE
IN

SET LEVEL START STOP

INPUT INPUT COMM°—1
-

START STOP

,Jt
TIMING
DIAGRAM

PULSE
AMPLITUDE EVENTS
,_....} TRIGGER LEVEL

A INPUT

fun HR SHAPED AFTER

TRIGGER STAGE
SIGNAL TO CATE ON

GATE OPEN B INPUT

TIMING
DIAGRAM

TIME
EVENTS COUNTED
OR BATCHED ri SIGNAI
TO GATE OFF

Fig. 9. Schematic diagram of totalizing and batching use of counter

with display. GATE OPEN

Because many triggering signals are

ill-defined in time, most
counter/timers have input stages that
IIIIIIII111111111111r1r1 KNOWN TIME INTERVAL

11111111111IIIIIIr"
CLOCK
trigger at preset adjustable levels —
Fig. 10. Addition of a
Schmidtt triggers or comparators are clock to the totalizer
used. This enables the operator to yields time- interval SES COUNTED
ATE INTERVAL
discriminate the events to be counted measurement with
from relatively noisy backgrounds that this configuration.

have a lower peak value — see later.

Another reason for selective- level
triggering is to allow the counter to
basic procedure is to gate the clock PULSE WIDTH
operate at different points on a
into the counter for the interval
waveform — a sinewave input can be
between the same trigger level of A special case of period measurement
used to produce pulses of varying
successive waveforms. The precision occurs when the width of apulse is to
widths in this way. The counter
can be greatly improved by extending be determined. If the pulse had a
output may also be used to trigger an
the gate- open time to 10, 100, 1000, perfect square response profile the on
event so as to provide automatic
or 10 000 periods, dividing the count and off gating points would always
timing sequences, this being similar in
by the appropriate divisor ( which give an accurate answer because the
priniciple to batch counting.
means a mere shift of the decimal triggering transitions would occur
point if decirral multiples are used). precisely on the rise and fall of the
PERIOD OF REPETITIVE SIGNAL
This is referred to as multiple- period pulse. Trigger- level would not affect
The time interval measurement measurement, it yives greater precision the interval measured. Practical pulses,
arrangement also enables the period of but at the expense of greatly increased however, will not be perfect, the edges
a wave- form to be measured. The most time for each measurement. having definite rise and fall times. In
62
this case the trigger- level becomes POSITIVE SLOPE TRIGGERING NEGATIVE SLOPE TRIGGERING
critical in width determination, as is
TRIGGER LEVEL -
depicted in Fig. 11. Counters usually
PULSE \it HIGH ( b)
provide a slope selection control. This AMPLITUDE (a)
(b)
(a)
decides whether the trigger operates LOW
on the positive or negative slope, that
is, a or b slopes respectively ( Fig. 11).

TIME

PHASE DIFFERENCE
Two identical waveforms can be
regarded as two separate inputs for the
start-stop inputs. If both trigger at the
same point on each waveform
(preferably at the zero- crossing to gain MEASURED PULSE WIDTHS.
maximum precision) the ratio of the
Fig. 11. Trigger level must be considered in
time- interval between the two pulse width measurement to obtain the DISPLAY
crossings to the period of the parameter required.
waveform is the phase shift in terms of
afraction of one cycle.
HOLD

FREQUENCY MEASUREMENT
Fig. 12. When the clock operates the gate
The frequency of arepetitive signal is length the counter system can measure the
COUNTER
defined as the number of cycles frequency of a signal.
(events) per unit time. A digital
frequency meter, therefore, can be FREQUENCY
INPUT
made with controlled gate- on period
o TRIGGER
of precisely known time interval. In
other words, it is the same
START STOP
combination as the interval timer ( i.e.
oscillator counter with display, gate
control and precision) but with the
difference that the clock now controls
the gate ( not the counter) through a
precision divider that scales the basic CLOCK DIVIDER/SCALER
clock frequency down to obtain the
gate duration needed. Gate periods
range from 1jis to 1s. A simplified
schematic is given in Fig. 12. The
input signal is often sinusoidal; the
input stage shapes this into a square
wave to enhance individual cycle
detection by the counter. INPUT TO
BE MEASURED

PULSE
AMPLITUDE

NOISE ERROR REDUCTION

PJL
SHAPED AFTER
TRIGGER
The above descriptions give the basic
operating modes of the various
(NOTE ERROR OF DEFINITION)
counter/time/frequency-meter
combinations. In practice anumber of GATE PERIOD
refinements may be incorporated to (DECIDED BY SCALER)

obtain better practical performance.

Noise can be reduced by

I I
POSITIVE SLOPES
incorporating a fixed amount of COUNTED IN KNOWN
backlash in the trigger circuit; this PERIOD

produces what is called the trigger

TIME
window. On the way up the trigger
level is at a higher level than on the
way down, as shown in Fig. 13.
Provided the noise added to the signal Low frequency measurements can automatically selected by the counter
has an amplitude smaller than the often involve interference sources that itself after it has made adetermination
window width, the counter will only produce rapid spike transients. One of the frequency of the signal.
trigger once on the way up and once simple method of reducing this is to A recent approach to the noise
on the way down. This method works use filters. Advanced designs contain problem is to set up atime- window ( as
well for high- frequency measurements filter systems that reject all opposed to the trigger height window)
where the noise is usually a small frequencies higher than that being that, once the counter gate is on,
percent of the signal- plus- noise signal tested, the appropriate filter being inhibits the off-state chance until after
amplitude.
63
 ELECTRONICS it's easy!
GATE GOES OFF
JUST ONCE
ON
O
ITRIGGER HEIGHT
WINDOW WIDE
OFF
OFF
GATE GOES ON
JUST ONCE

TRIGGER t
LEVELS

(a) WINDOW SET TO EXCLUDE FALSE TRIGGERING TIME

Fig. 13. Noise reduction
in counting by use of
a trigger height window.

ON ON
ON ON ON
ON
OFF TRIGGER WINDOW NARROW
OFF
OFF OFF OFF
OFF

TRIGGER!
LEVELS

Inn
(b) WINDOW TOO NARROW — FALSE TRIGGERING OCCURS TIME

a time just shorter than the expected Before using a counter/timer on an to decide the best strategy for
interval. This is known as trigger unknown waveform it is, where trigger- level and height- window width
masking. It is very useful in feasible, good practice to study the settings. Figure 14 illustrates the
eliminating contact- bounce retriggers. waveshape on an oscilloscope in order differences between window level
settings on various waveshapes.
As the readout is in digital form it is
TRIGGER WINDOW
necessary to hold the display at the
determined value for a period long
enough to allow the value to be read.
Some units incorporate a control that
PULSE WITH RINGING gives the operator a choice of hold
time.
The following part of this series will
continue with this general discussion
of digital instruments, covering
physical variable transducer systems,
SINE WAVE WITH STRONG 3RD HARMONIC DISTORTION
the various kinds of analysers and
correlators, waveform generators and
computer-controlled testing systems.
Fig. 14. Trigger height
for the window must REFERENCES
be chosen to suit the Varies issues of Hewlett Packard
SIGNAL WITH NOISE SPIKES wave shape.
Journal contain many detailed articles
on a wide range of digital instruments.
7 "Digital Instrument Course — Pt. 2",
- A. J. Bouwens, Philips Industries
Holdings, Sydney, is useful to have.
PULSE WITH DC OFFSET
This booklet describes the practical
PRODUCES FALSE OR ZERO BETTER CHOICE OF TRIGGER use of counter/timer instruments in
TRIGGERING TO OBTAIN DATA REQUIRED many varied applications.
64
 More about digital

29 instrumentation

VERY FEW VARIABLES to be

measured by electronic means provide
a digital signal directly: this is because
the real- world is predominantly
analogue by nature. Consequently,
most so-called digital measurement
systems involve anumber of stages to
make the signal compatible with the
digital circuits of asystem.
The most straight-forward 'digital'
measurement method ( at least at
present), is to employ a suitable
analogue sensor that provides avoltage
(or current) output, related to the
variable being measured. This signal
then feeds an A/D converter to obtain
adigital equivalent.
Fig. 1. This Fluke
The low cost of digital calculation 2100 series digital
circuitry now enables linearization of thermometer incor-
sensor processes at moderate cost. porates digital
linearization. It is
Figure 1 gives an example of the suitable for use
digital linearization used in a with any one of six
thermocouple thermometer unit. common thermo-
couples.
By referring to Fig. 2we see that the
linearization process, in the dual-slope
digital voltmeter section of the system, Fig. 2. Block diagram of the digital linearization tech-
nique used in adigital thermometer.
is achieved by changing the ramp slope DISPLAY
at a number of points. To do this a
rate multiplier is used to multiply the II •
clock frequency by avariable number,
N/256, where N may be any number THERMO
COLD
BUFFER
JUNCTION A/D COUNTER
between 1and 256. By this means 256 COUPLE 'COMPENSATION AMP CONVERTER

different ramp slopes may be

generated. The slope in use is tracked
by a segment counter which, in turn,
causes a read-only- memory ( ROM) to CLOCK H
RATE
ROM
SEGMENT
COUN rER
MULTIPLIER
set up the correct digital- readout code.
Some sensing principles lend
themselves to a more direct digital
signal approach. For example, in the
Moire-fringe displacement sensor, a
grid of fine lines ( called a grating)

r?.
formed on glass is attached to the
moving ( or fixed) member of the
machine whose movement is to be
monitored. This is shown in Fig. 3.
The other member carries a small
index grating set to produce
Moire- fringes which move as the two
grids pass relative to each other.
Movement of these fringes is
monitored by photocells which Fig. 3. Accurate
measurement of move-
provide a number of electronic pulses ment is performed by CONDENSER
proportional to the magnitude of the using optical gratings
displacement. These pulses can be to produce "moire- •••.
F XED INDEX
fringes" which can
counted directly with a reversible GRATING
be counted as the LIGHT
direction counter, thus allowing both gratings move with SOURCE
directions of movement to be respect to each
followed. Such a unit is known in the other.
35
 ELECTRONICS—it's easy! to pulse loss or gain, or to power
failure errors which occur in the
previous system if not fitted with a
special non-volatile memory. The discs
of such a system are read optically as
though they were registers or other
forms of digital store, each position
having a different digital code as read
across aradial line.
In some forms of digital pulse
transducer it is the rate of pulse
+ I50325 4
production that represents the
,
variable, not the absolute number of
+05"c'715 pulses. An example of this sensor is
the turbine flow meter used to
measure liquid or gas flow. Figure 6
shows such a flowmeter where asmall
turbine rotates inside apipe at aspeed
related to the flow rate. Rotation may
be converted into pulses using optical,
magnetic, capacitive or, in earlier
designs, mechanical sensing. This form
of sensor provides avariable frequency
output which can be converted by a
counter/timer system into a direct
readout of flow rate.
Digital transducers are somewhat
similar. They provide a signal which
varies in frequency as the variable
being measured changes. The sensor of
such a transducer is made such that it
alters a parameter of a frequency
generating circuit. For example the
quartz- crystal thermometer shown in
Fig. 7 operates in this manner. In this
unit temperature causes the resonant
frequency of acrystal, mounted in the
Fig. 4. This lathe is fitted with adigital readout of displacement. The self-contained end of the probe, to change in a
Heidenhain linear sensor is to be seen high at the back of the cross slide. predictable fashion.
It is interesting to note that many
natural physiological sensors operate
metal working industry as digital on the pulse- rate system — neurons
instrument must not lose or gain stray
readout ( DRO) — a typical system is (the digital nerve sensors on the end of
counts or else the wrong value is
shown in Fig. 4. A somewhat similar the nervous system) trigger with pulse
indicated. Such systems are called
length measuring system is the laser repetition rates that rise in accordance
incrementals.
interferometer — this also provides with the intensity of the actuating
There is an alternative method which
fringes that can be counted to provide signal ( heat, cold etc).
uses digitally encoded discs similar to
a measure of absolute displacement. Considerable effort has been
those shown in Fig. 5. This is an
Clearly in such cases the digital expended — especially in the Eastern
absolute method which is not subject
European countries in the late 60's —
to produce reliable low-cost industrial
sensors that provide a digital form of
output. These have not, however, been
accepted to the extent hoped. The
current low-cost of extremely
powerful digital circuits, however, is
likely soon to produce a trend toward
sensing devices having digital output.

AN
Analysis is the general process used
to break down an unknown by
methods which separate and
distinguish basic elements of seemingly
complex arrangements, the elements
so derived being satisfactorily
understood basic quantities.
Synthesis is the alternative approach
Fig. 5. The absolute digital code disc acts
wherein a system is built up from
as astore of the angular value of dis- Fig. 6. Turbine flow meters like this one
placement. It is interrogated as though it known elements to produce the
provide an output in the form of apulse
were a register whose value is read optically train with apulse repetition rate pro- complex case.
across aradial line. portional to flow rate. Analysis may be regarded as being
66
 required when the behaviour of an
existing system needs to be studied.
Synthesis is used when asystem is to
be devised. There are of course many
instances when both approaches are
used to yield asolution.
Various types of electronic analysers
are used in electronics. We will look
here at spectrum analysers, logic state
analysers and pulse-height analysers as
Fig 7. Quartz- crystal these types are commonly met in
thermometer from modern circuit work. Each of these
Hewlett Packard.
operates on an existing electrical signal
breaking it down into frequency
f content, logic-state content and height
of pulses, respectively.

AMPLITUDE
SPECTRUM ANALYSERS
FREQUENCY Signals in the time- domain, that is
those displayed as amplitude versus
time graphs, can also be displayed in
terms of their amplitude- versus--
frequency and phase- versus-frequency
characteristics. ( This was discussed in
Part 4 where an example wave- form —
asquare wave — was broken up into its
harmonics). The relationship between
time, amplitude and frequency are
AMPLITUDE- TIME AMPLITUDEFREQUENCY
seen by studying the three forms
(shown in Fig. 8) of a fundamental
sinewave having a large degree of
AMPLITUDE second harmonic added in. Signals
displayed as amplitude ( or phase)
versus frequency are said to be in the
frequency domain. This kind of plot
2f
FREQUENCY shows the frequency spectra of the
signal, hence the name spectrum
Fig. 8. Second harmonic distortion is not always easily seen on an amplitude analysers.
versus time display. In the amplitude versus frequency display the second
harmonic distortion and its amplitude are clearly seen.
The role of spectrum analysers is to
display the signal content in its
frequency domain form. There exists
many instances where this form of
display is better than a time-domain
representation. Typical examples are
where a fundamental has distortion
(Fig. 9a) or where low levels of
modulation or noise exist ( Fig. 9b).
Neither of these conditions could be
satisfactorily detected, let alone
measured, by atime- domain test.
Basic spectrum analysers use
analogue circuitry and therefore do
2p.sec/div o 1MHz not qualify properly for inclusion in a
discussion on digital instruments.
Fig. 9. Frequency domain displays are often better than the time- domain method as these HP However, as we will see later, the
displays illustrate: (a) In the time domain (left) the signal looks pure but the spectrum current trend is to include digital
analyser shows that it has significant distortion.
techniques in such instruments.
Advanced analysis equipments, for
example, often use a built-in digital
computer.
There are two alternative forms of
spectrum display. First, the repetitive
signal can be studied over an extended
time period by scanning across the
expected frequency- range with narrow
band-pass filters. A speedier, but more
expensive method, works in areal-time
mode thus preserving the
ti me- dependency between signals.
20 msec/div 15 MHz 200 kHz/div These are known as swept-tuned and
Fig 9b. A 2% amplitude modulation is barely discernible on time domain plot (left). The
real-time spectrum analysers
frequency domain plot clearly shows the frequencies present and their amplitude. respectively.
67
 ELECTRONICS it's easy! plates of an oscilloscope. The sweep
signal drives the X plates. Figure 10
depicts this arrangement.
CONCEPT Real-time systems — These use astack
of band-pass filters and detectors each
INPUT FREQUENCY
connected to the signal simultaneously
ELECTRICAL
SYSTEM and with each having staggered centre
frequencies. This is shown sche-
INPUT matically in Fig. 11. The scan gener-
ator multiplexes the individual
SWEPT
channels in order to produce a
TUNING
continuous spectrum on the oscillo-
scope screen.
MIXER
It is clear that this method is much
more expensive because many filters
IF are needed. It does, however, enable a
detailed analysis of once- only transient
FILl'ER<>A4 signals which could not be analysed
LO
with the swept- tuned arrangement of a
CRT
DETECTOR spectrum analyser.
DISPLAY
SWEEP A range of spectrum analysers is
available for the study of signals from
5 Hz to 50 GHz. Different instruments
(or the use of different plug- ins with
Fig. 10. Schematic of swept- tuned form of the same display unit) are needed
spectrum analyser. because units typically cover only 4 to
5 decades, that is, say 5 Hz- 50 kHz,
10 kHz-300 MHz and so on. The range
Swept-tuned systems — Basically the bandwidth and roll- offs. The is, however, ever widening. Wide range,
task is to establish the amplitude ( and bandwidth of the filter may also need however, is not always the virtue
sometimes phase) of the signal at each to change if the requirement is for a needed for spectral resolution is
frequency in turn. Many practical filter bandwidth that is always agiven related to width of display screen.
difficulties exist because the proportion of the signal frequency as Fourier Analysers — A third method
absolutely narrow band filter does not it sweeps the range.
of providing a frequency analysis is
exist and even if it did, it would take Most difficulties are overcome by based on direct mathematical
an enormous time to sweep it across mixing the signal with a swept local calculation using the Fourier
the full bandwidth of the signal. oscillator and th - n-i detecting the
transform technique to convert a
Practical filters also have finite output and using it to drive the Y
time- domain signal into its
frequency-domain equivalent. Such
systems are extremely expensive
FILTER DETECTOR compared with the above analysers,
1 1 but provide a vastly greater capability.
They can also handle signals at the
very- low-frequency end — dc to
FILTER DETECTOR
2 2 100 kHz is typical. Their operation is
quite different from the above in that
ELECTRONIC the signal is fed as data values into the
FILTER DETECTOR SCAN
3 SWITCH
analyser unit via keyboard or paper
3
fs
tape from another computer or
•
mass-storage system. It can also be fed
in as an analogue signal from, for
example, magnetic tape. The heart of
the Fourier analyser is a micro-
programmable computer system which
SCAN
GENERATOR can be set to compute using various
FILTER I DETECTOR
programmes such as the so-called Fast-
Fourier method of analysis. The same
(a) CRT
unit may also be able to carry out
correlations between signals, plus
many other processing techniques.

Digital circuitry in spectrum analysers

AMPLITUDE FREQUENCY — Digital circuits are being added by
RESOLUTION
manufacturers to enhance the
BAND WIDTH
performance of analysers. Advantages
claimed include operating ease and
FREQUENCY
better placement of controls. Digital
(b) FREQUENCY RANGE
storage of the display signals can be
used to enhance the display brightness
and to allow a spectrum to be ' held'
Fig. 11. Block diagram of real-time spectrum analyser based on stacked filters:
(al schematic for comparison against a second
(b) frequency response showing individual filter windows. spectrum obtained later. Digital
68
included; this reduces the noise
thereby enhancing the signal/noise
ratio on the display — as is illustrated tinitot«.1

, •
in Fig. 12. Character generation ( using
Fig. 12. The Tektronix 9e3
digital methods) has been incorporated
spectrum analyser in-
to display the relevant graph-axes corporates various
factors — as shown in Fig. 12. The digital techniques that
same unit also uses a photo-optical provide character :et
.. .61111.

generation on the dis-

absolute-digital code disk to replace play and reduce the
the mechanical switch usually used in noise level of asignal.
arange control- knob. The adjacent photo
Spectrum analysers are invaluable shows the original
unfiltered signal
and are finding increasing use. containing the two
Successful use is, however, amatter of small signals and noise
experience and frequency-domain (recovered in the CRO
display).
techniques are not dealt with as
extensively as time-domain ones in
training programmes. More details are
available in the reading list — we can
only provide the most elementary
introduction here.
Logic-State Analysers — We check the
operation of analogue circuits by
measuring signal levels and frequency
spectra at various points in the circuit.
Digital circuits are different in that
they contain the signal information in
the form of multi-digit 'words' made `.2' mums
atwa MO" VIER

up of two-state bits. To check

operation, therefore, we must
ascertain simultaneous logic-states at
various points in the circuitry. The
simplest analyser for this work is a not otherwise be seen in the lamp moving florn point to point in turn.
probe which indicates logic hi or to display. It must also have connections To speed-up the analysing process a
state at a selected point; coloured suitable for PC board digital circu;try more extensive facility to use would
lights are used as indicators. A store — see Fig. 13. be one that simultaneously shows the
function can be built-in to the probe The single probe can be used to logic states of multiple points in the
to catch ashort transition that would analyse the state of a circuit by so-called Data- domain. The

Fig. 13. Simple logic

probe in action.
69
 ELECTRONICS it's easy!

Fig. 14. Hewlett Packard logic-state analyser as set up to test aprinted-circuit board card.

DISCRIMINATION
t PULSE LEVELS
those pulses (for counting) that arise
COUNT
HEIGHT
ZERO from the particular source of interest.
Pulses above the trigger window, or
COUNT below are rejected (not counted) as
ONE
}WINDOW
demonstrated in Fig. 15.

CORRELATORS
COUNT Fig. 15. Pulse-height
ZERO analysers count only Correlation as aphysical process can
those pulses which be visualized by considering two
have amplitudes within identical optical patterns formed on
the specified window.
ZERO film transparencies. When the two are
exactly overlaid, the maximum
transmission of light occurs. At any
Hewlett-Packard system, for example other position the light transmission
SIGNAL m SIGNAL ti (shown in Fig.14), displays over 500 falls off to a minimum at greatest
points as a matrix of O's or 1's on a misalignment. The mathematical
CRO screen. These instruments are process modelling this is that whereby
ml nftl used to debug, test or trouble-shoot the two patterns ( represented as
complex digital circuits. Only large formulae) are multiplied together and
ADJUSTABLE
TIME DELAY DELAY laboratories, however, would be able the multiplicand signal is averaged as
to justify the cost of such advanced depicted in Fig. 16. This process is
logic analysers. repeated using a slightly different
Pulse- height analysers ( Discriminators) spatial (or timewise) phase-shift
— Measurement processes involving between signals each time the sums are
ionising radiation and sometimes performed. The same idea of
light-intensity levels rely on pulse correlation applies for time domain as
MULTIPLIER
counting, the pulses appear as rapid well as space-domain signals. When
mft-T;nill
electrical currents produced from a two identical ' patterns' are compared
photo- multiplier or ionisation in this way it is called auto-correlation.
INTEGRATOR
detector. The relative amplitude of a When two patterns that are not
pulse often distinguishes it from pulses identical are compared it is
_ from other sources. For example, cross-correlation. Correlation brings
different radio-active isotopes produce out any similarity as a peak in the
pulses of different energy, enabling an correlator output when the greatest
INDICATING RECORDING assay of radio-active mineral to be degree of correlation is achieved.
METER OUTPUT
made by a study of occurence of Early co rrelators used entirely
pulses of different height. Photons analogue methods to carry out the
arising from the various noise sources signal processing. Today, correlators
OUTPUT = ft m (t—T) n(t)dt
in photo-multipliers have different mainly use digital techniques in the
= CROSS CORRELATION energy from those generated at the manner shown schematically in Fig.
OF m,n SIGNALS
photo-cathode. This is atrue detection 17. The two inputs are converted from
process: noise can be reduced by analogue to digital form in the
Fig. 15 Block diagram of the mathematical
function performed by acorrelator unit. discrimination of pulse heights. quanti zers. One channel is
Pulse height analysers use carefully progressively delayed by storage in a
selected trigger levels to accept only shift register. Multiplication and
70
 CIUANTIZER

ANALOGUE
SIGNALS AMPLIFIERS MULTIPLY AVE RAGE DISPLAY
A/D CONVERTERS
IN

MULTI STAGE
11 OUANTIZER — 110— SUIF T
HE GISTER
Fig. 17. The simplified block
diagram of an auto correlator.

SHIFT

averaging are also achieved by digital formed by addition of a number of the output via a programmable BCD
methods. sources or by manipulation of asingle, digital input. Programming enables an
Correlation finds use in the detection stable- reference frequency. A music enormous range of signals to be
of periodic signals buried in noise; in synthesizer provides a whole range of synthesized, a typical requirement
establishing the degree of coherence musical sounds by combining many being as part of an automatic test
between random signals; in different tones into a single output. procedure. Figure 18 shows the
establishing transmission time and Although synthesizers work upon philosophy of the HP 3330 series of
source of transmission of a signal; in basic analogue signals the trend is to automatic synthesizers with a typical
identifying the characteristics of a combine or modify the signals using programme card marked up for a
complex system ( the input is digital control. frequency sweep routine.
perturbed with a noise- like signal Digitally-controlled power sources
The advantages offered are ( in the
which is cross-correlated with the may be used to synthesize varying
variable frequency generator kind of
output to give the transfer functions voltage ( or current) levels over a test
synthesizer) that a very stable
of the system). It is a particularly period at the commands of a mini
reference oscillator has its frequency
powerful tool and we should see more computer in the same way as the
translated to ( literally) billions of
of its use if the cost of digital above unit synthesizes frequencies.
other values ( the HP 8660 gives
correlators continues to fall. Frequency and voltage synthesizers
10 kHz-2600 MHz) whilst retaining
are often combined in the
SYNTHESIZERS high stability. By pressing digital- key
hybrid-computer ( digital and analogue
These are a special kind of signal inputs, any chosen frequency value is
combined) in order to generate
generator in that the signal output is generated. It is also possible to control
synthesized signals which are needed

a.

! OUTPUT
FREQUENCY ATTENUATION! --10. CARD NO .
SYNTHESIZER AND LEVELLING

t CONTROL CONTROL
ADDRESSING
NO.

2
STE > CODE 200

elf

•
,00:

;O
40 20

o
10

o
fin

o
4

o
2

51011 130 ▪ V I ;0 Mani O O

ei
• 'RIO 114 O D : LT) EJ M ain O Ci

DIGITAL CTR. 065 ▪ 0:•• OME2ilat

DIGITAL FREQUENCY
CONTROLLER DISPLAY 41117 077 O 0;ffl.IMMMIIMMIMMil.

`;1?
2̀ 0 a ;0 O Waal 0 113

FREQ 065 • O ; Manna 0 ; la O VIP

STEP
1007 1.76 O:

116 6=1 M I; 0 0 M 1.1111M3

11 1 06 , 0 0: 'MOB 0 ; 0 0 ira
Fig. 18. Automatic frequency 1M
AMPL 12 J >s, 0 O ; MM. 0 , 0
synthesizer. (a) Block diagram.
,
-MI
KEYBOARD OR 13 .. A ^' ,, I
= 1; SIMMEEMMI : IIIMP
(14 H-P 3330A unit. (c)
0 0
--IMI
REMOTE
Programme card marked up
14 VI'. 1. 0 MI : 1= EM IEM OI
RMINIM 0
eM
PROGRAMMING
for a frequency- sweep NO. OF STEPS 15 ee. 12 , 0 M11; O D 0:

routine. TIME STEP t6 O III : 0 IMP ;

5Ye D go; D D 0, MI.

START SINGLE '

FREQ. SWEEP IN 18 D ID; D O = 1

1 D'e a
UP DIRECTION 133 0 ; 11».. ; D M OM

20 D o:o o o:o D o
b. 21 O 0; 0 0 D :D D o
22 O 00 0 0;0 0

23 O 0 ; O D O :O 0 0 a

24 ED; O 0 0; 0 00
eal
25 O 0; 0 D O

26 O DD 0 0 D 0 0

2, O D :D 0

28 ;D 0 0 o D O

29 1=1 D D D D C .
J 1=1 C _I

30 O 0;0 0 o; D 0 0

00;0 O 0,0 O EJ
ffl
277 M k : (
S1 Mr a: all Q M OW
UNADDRESS 32
a
u ss soFT pseen.
2. DO NOT MARE IN SHADED AREA 1TORI
3. ERASE COMPLETELY
4. INSERT TN 'S S.DEUD

0,7 PANT No 9,20-211181,

71
ELECTRONICS it's easy!
FROM
COMPUTER
UTILITY FLUKE 1100A INTERFACE
PROCESSOR TEST I

1100A-7016 IN

OR
t .

1100A-7026
INSTRUMENT
CONTROL
INTERFACE INTERFACE CARD SLOTS
(15 TOTAL)
1200A

j TEST I
1300A
OPERATOR _I INSTRUMENT
INTERFACE I

-_40(
PANEL
DISPLAY
KEYBOARD SYSTEM
INSTRUMENT) - ANALOG
SWITCHING
r- SUBSYSTEM

SYSTEM
e ( INSTRUMENT.> - 1

1600A

I
<
Fig. 19. Block diagram of Fluke
AUX. KEYBOARD SYSTEM
INSTRUMENT) calibration standard.

to derive a simulation of a complex console which usually incorporates a on sensors but rarely as digital
system, such as amissile in flight. wide range of facilities that are chosen concepts, specifically. Recent releases
with flexibility of operation in mind. are:
COMPUTER CONTROLLED TEST Figure 19 shows an automatic system "Transducers in measurement and
SYSTEMS used to calibrate a test instrument. control" — P.H. Sydenham, UNE
With the enormous increase in The test programme must be devised Publishing Unit, Armidale, 1975
complexity of routine complex by a highly-trained professional (Available ETU.
processes ( such as aircraft designer, but once developed and
"Transducers in Industrial Measure-
instrumentation and controls, programmed the testing can be ment" — P.N. Mansfield, Butterworths,
refineries, automatic and large- volume performed by aless trained person. London, 1973.
manufacture of electronic systems) It is not possible here to deal in
"Analysers — detail is available in the
came the need to improve and depth with automatic testing as the
data sheets and journals of companies
speed-up the testing procedures range of requirements and equipment
such as Marconi Instruments,
needed to check out the thousands of available are both great. The overall
Hewlett-Packard, Tektronix,
different parameters involved. concept and scope of an automatic
Honeywell, Spectral- Dynamics, Bruel
Computer controlled testing is far test system is shown in Fig. 20. Suffice
and Kjaer and others.
more reliable than human operator to say very complicated automatic
"Counting Photons", ETI,
testing and is extremely fast. It can be testing systems are in routine use in a
November, 1974 and " Electronics in
economic even for the testing of small wide variety of manufacturing and
Medicine — Pt. 2" ETI, August, 1975,
volume electronic equipment, maintenance situation.
dealt with pulse- height analysis.
especially where a large range of tests
is involved. REFERENCES Hewlett-Packard offer a series of
The instrument or process to be Digital Sensors — these are variously varied booklets on Spectrum
tested is interfaced to the main test described in the many general books Analysers, as well as video tapes.
Application Note 150 provides basic
understandi ng.

CONTROLLER
Synthesizers of the music kind are
COMPUTER HP 2100AI HP 95000 discussed in " Electronics in Music", F.
CONTROL PANEL
TELEPRINTER
DISC CONTROLLER
SYSTEM
C. Judd, Neville Spearman, London,
PUNCHED TAPE READER
DISC MEMORY
1972. A complete project of a music
synthesizer was covered in articles in
ETI beginning October, 1973. Part 1
contains reference to digital control of
MEASUREMENT the synthesis process.
STIMULUS
SWITCHING SUBSYSTEM
SUBSYSTEM
SUBSYSTEM Manufacturers' data sheets are a
Idc. ac. resistance. frequency.
Id,. ac pulses ill
onlerval pull., lme, source of reference to both laboratory
powet tits- 10.°n Ousel
synthesizers and programmable dc
power supplies ( which are also referred
to as programmable power — DACs).
INTERFACE PANEL
Auto Testing Systems have been
described in a number of ETI articles.
"Computer- interfaced
UNIT UNDER TEST instrumentation in the Development
Fig. 20. Overall concept
of HP9500 series of Laboratory", April 1972 discussed the
automatic test systems. potential. •
72
Notes

73
 Digital

30
computers

THROUGHOUT THIS COURSE history of the development of logical mathematical operations. The
we have been steadily building up operation by electrical switches and analogue machine cannot perform
sufficient information to enable how logic gates operate to perform logic operations: ( unless cojoined with
discussion of computing machine simple arithmetic was covered in Part a digital computer, in which case it is
operations. What follows is necessarily 22 — other basic digital functions known as a hybrid computer — as
an introduction only — computers are being dealt with in Part 23. Storage of shown in Fig. 1) its use is generally
now extremely sophisticated in design digital numbers in solid-state counters restricted to what are called linear
and the manufacturing methods very was the subject of Part 24 and the mathematical problems in which
specialised. It is, however, quite conversion of signals from signals vary continuously and
important that the operation of code-to-code and from analogue- to- information is transferred as levels not
computers be understood by digital, and vice versa, were discussed as digital codes. Analogue computers
electronic craftsmen at a general in Parts 26 and 27, respectively. How can be very good at such operations,
systems level. This, and the next part, computers become involved in testing often better than adigital computer of
will introduce the philosophies, the was briefly mentioned in Part 29. similar cost. The digital machine, on
hardware and the operation of digital These are each important concepts, the other hand, ( a general purpose
computers from a technical rather worth revising at this stage. installation is shown in Fig. 2) can
than user- only viewpoint. ( Analogue perform almost any kind of
computers were briefly mentioned in With this much learned there is little mathematical manipulation, however
part 12 — they are still valuable in else that need be studied about special techniques are often needed to
some applications but in general, components ( stores of various kinds solve analogue problems. Analogue
machine computing is now mainly and IC developments in computers are type signals must be sampled and each
done digitally). discussed later) in order to understand sample converted into a digital
Already we have introduced the how digital computers operate. Our equivalent before they can be
systems approach of understanding emphasis must now be on the design processed in digital machines: this is
analogue and digital electronic systems arrangement of the computing where the digital machine in certain
(Parts 1, 21); how information can be machine as a whole and how the user applications may be less efficient than
conveyed in binary code form and can make it work. the analogue alternative.
how different channels can be handled As well as performing arithmetical
simultaneously on a common WHAT IS A COMPUTER? operations ( called scientific
transmission line ( Part 5); how Regardless of whether a computer is computing) the digital machine can be
square-wave clock signals are generated digital or analogue in operation its role instructed to process or sort discrete
was covered in parts 17 and 18. The is to perform various kinds of data in digitally encoded form ( called

Fig. 1. Here adigital computer and an analogue computer are combined — the result is known as ahybrid computer.
74
 performing calculator like tasks and
some calculators are now so flexible
that they can be programmed to
perform avariety of tasks.
In the 1950s, when powerful
electronic computers were emerging,
the popular concept was of amachine
that would soon have thinking powers
of its own — and its own will and
imagination — as depicted in Fig. 3.
Although we must concede such is
probably possible one day — no one
has yet gained an inkling into how this
extra facility could be realised.
Computers are merely machine slaves
that, if working internally as the
designer thinks and intends, will
perform as commanded. The operator
informs the machine of its job via the
programme presented to it. Where the
computer has valuable merit is in its
ability to perform calculations and
process numerical data at rates vastly
Fig_ 2. General purpose digital computer, this centre is used by Lloyds Bank in Britain. greater than a human mind, with
rarely an error, and for hours on end if
need be. It is atool and no more. To
data processing or DP, for short). of all angles between 10° and 90° at
say the computer accidently sent the
Typical computer data processing 1° intervals.
$1,000,000 bill to Bill Blogs is entirely
operations are the sorting of numerical We pause now to note that we cal'
incorrect. The programmer or the
data — for example to see how many such machines computers not
machine did not perform as hoped
people have heights of various chosen calcJlators. The term calculator has
through one or the other being
values, or the booking of airline seats. traditionally been used to describe
defective in the instructions given or
Mixed working, where scientific machines which perform afixed set of
the way they were obeyed.
calculation and data-processing are mathematical calculations. The term
computer on the other hand, is As well as computers that operate
both involved, occurs for example, in
reserved for those machines which only when the operator gives
costing out abuilding estimate, raising
may be reconfigured by a set of instructions there is also the dedicated
a stock value for a business, or
programme instructions to perform machine that, once set internally to
producing pay-slips.
any particular task. However such compute or process in apredetermined
Digital computers may also calculate
way, becomes part of a process. It
tables by automatically incrementing distinction between the roles of
helps control by working at the same
the input data between preset limits. calculator and computer is becoming
rate as signals are generated in the
For example the computer could be increasingly difficult to make. Some
process — real time working.
asked to generate and print the sines computers are now dedicated to
Process-control computers, as these are
called, operate on data and perform
calculations as part of many feedback
loops in, say, achemical plant. Figure
4 shows this use in a diagrammatic
form. Other names variously used to
describe this use are in-line, on-line,
direct-digital-control ( DDC) or just
plain computer control. Wherever
automation of extensive complex
process is necessary a computer will
usually be found — waste-water
treatment plants, paper manufacture,
natural gas and electricity distribution
networks, satellite control and
power-station plant operation are but
a few of thousands of in-line
applications. Computers are far more
useful in this task than human
operators — see Fig. 5.
On-line operation ( although not
generally agreed upon) is a term
probably best reserved for cases where
each of many input terminals
connected to a central computer can
gain access to the unit when it
becomes available. This is also known
Fig. 3. As yet, computers can only
do what they are programmed to as time-sharing and is used where the
do. signal processing rate need not match
the process. The computers used in
75
 ELECTRONICS—it's easy! in connection with the minicomputer
because that term is used in computer
jargon in two distinct ways. It may
describe the hardware used — first
INPUT X
generation computers use thermionic
OUTPUT Z valves and ordinary cable wiring, such

s)
PROCESS
as shown in Fig. 8, second generation
INPUT machines use discrete transistor
circuits on printed-circuit boards, third
SENSING generation machines use
CONTROLLERS DEVICE
integrated-circuitry and the most
recent, about to emerge, fourth
generation computers use large-scale-
integration LSI manufacturing
L _ methods — see Fig. 9. A fifth
SET POINT STORAGE
DETERMINATION
LOGIC AND <- generation computer is yet to emerge
COMPARISON
as an accepted concept. The other use
A
of "generation" is in describing the
system interconnections — the
CHECK INGE-- -I TIMER H philosophy of system hardware
interconnection and style, and
PROCESS COMPUTER
capacity of the store involved.
Fig.4. Process contro computers are electronic data
processing machines, dedicated to aspecific task.
A HISTORY OF COMPUTING
MACHINES
banking in Britain operate in a mid-sixties manufacturing techniques Intertwined with the development of
time-sharing mode — bank branches, as and designs were such that anew style machine operated logic ( studied at the
shown in Fig. 6, can gain access to the of less versatile but compact computer beginning of Part 22) was the gradual
central-account records — a short wait was marketed — the so-called increase in sophistication of
may be necessary. When the computer minicomputer. Figure 7 shows but one computing machine systems.
works on diverse problems at the will kind of mini- computer system Earliest devices were simple
of the operator and is not used for any employed to control a process by calculators based on mechanical
dedicated purpose it is said to be concepts. They performed simple
providing instructions as needed. ( It is
off-line. addition, subtraction and sometimes
not used in closed- loop as this process
Originally electronic computers were does not feed data back to the multiplication and division, doing this
huge — several rooms filled with racks computer). without the ability to store or hold
of valve electronic circuits. In the We do not use the word "generation" values other than inputs and computed
output.
Space does not permit extensive
description of this history — see the
Machine Man reading list for that. Figure 10 shows
Speed Much superior Lag 1sec. the style of the first calculating
Power Consistent at any level 1500W for about 10 sec, 350 W machine of the " modern" kind. This
for afew minutes, 150 W for performed arithmetic addition and
continuous work over aday. subtraction only, by mechanically
Consistency Ideal for routine, repetition, Not reliable — should be manipulating interconnected counting
precision monitored by machine. wheels and was probably made by
Complex activities Multi- channel Single channel. Pascal in 1642. In 1671 Leibniz
Memory Best for literal reproduction Large store multiple access. modified the same mechanism ( see
and short-term storage Better for principles and Fig. 10) to obtain multiplier action,
strategies. producing his own design calculator
Reasoning Good deductive Good inductive. much later — in 1694. Because
Computation Fast, accurate — poor at error Slow, subject to error mechanism manufacture at that time
correction Good at error correction. was crude indeed — all parts were
Input sensitivity Some outside human senses, Wide range ( 10 12 )and variety individually hand-crafted — the
e.g. radioactivity of stimuli dealt with by one Leibniz machine was not reliable even
unit, e.g. eye deals with though the concepts involved were
relative location, movement sound. Improvements in mechanical
and colour. manufacture had to occur before a
Insensitive to extraneous Affected by heat, cold, noise routinely useful gear and crank
and vibration. calculator could be built ( by de
Poor for pattern detection Good at pattern detection. Colmar in 1820). Thus, through these
Can detect signals in high and many other gradual improvements
noise levels. to method and manufacture, the scene
Overload reliability Sudden breakdown Graceful degradation. was set for grander ideas.
Intelligence None Can deal with unpredicted and A major advance was made by
unpredictable. Babbage. Charles Babbage was born in
Can anticipate. Devon, England. In 1792, he became a
Manipulative Specif ic Great versatility. Professor of Mathematics at
abilities Cambridge University and had a
Fig. 5. Fitt's list summarizes the relative advantages of man versus machine control. consuming passion for mechanical
machines that could perform far more
76
advanced mathematical onarations
than any previous apparatus. His first
machine, shown in Fig. 11, was
devised to solve differential equations
by calculating differences. This was his
"Difference Engine" of about 1812. In
1833 he conceived a second, quite
different general-purpose engine — the
so-called " Great Calculating Engine".
In principle, it could do any
mathematical operation by following
instructions programmed into it by the
operators. It could also make
decisions, on what to do next, that
were based on its just calculated
results.
Babbage used punched-cards for
input information ( areasonably logical
choice in view of the many repetitive
industrial processes using this control
medium at that time), a memory
(which he called "the store"), a
number processing section ( called the
mill), ameans of transferring results to
and from the store, and automatic
output ( as cast type ready to print). It
was agrand machine having ability to
store 1000 fifty-digit numbers in its
store. It even had overflow indication.
Fig. 6. In time-shared operation acentral com-
puter is made available to terminals. This map
shows the links of bank branches to two
central computer centres via concentrators.
Fig. 7. Mini-computers come in all shapes
and sizes. On the left, in the console, is the
H.P. 2000 that controls the pattern being
knitted on the Kirkland knitting machine.

77
ELECTRONICS-it's easy!

Fig. 8. Compact electronic

computer systems become
reality when valves were
replaced by solid-state
components. This single
plug-in unit, from a
Pegasus computer of the
50s, would today have
its entire function made
on apinhead in LSI
technology.

The intended power supply was

steam. Sadly, Babbage's engines were
gummy:, not proven in practice in his time;
those built were either not completed
or proved too unreliable. Manufacturing

111111fflirMathemeScian PR
methods were still incapable of
maintaining the tolerances needed — it
was a classical example of a concept
waiting for the requisite technology.
However from that time on
calculators rapidly became more
sophisticated. Keyboard entry ( Fig.12)
of data, instead of the need to turn
wheels, was introduced, ( but the
mechanism was still handcranked)
number length was limited and speed
was very slow ( by today's standards).
wtop Around 1910 electric- motor drives
were incorporated to perform the
numerous mechanical rotations needed
0e.
to transfer the carry-over value
through all decades.
Complicated mathematical equation
solving in the 19th and very early 20th
century was performed on other kinds
of special purpose mechanical
calculating devices. The planimeter,
Fig. 9. Today's computers use LSI techniques in which thousands of transistors and diodes
which determines area under a curve,
are contained within asingle chip.
was devised in 1814, the mechanical
ball- and- disk integrator was devised in
1876 ( by Lord Kelvin's brother). With
these and other basic
mechanical- function solving ideas,
Lord Kelvin and others put systems
together that carried out specialised
calculations. Kelvin produced a tidal
78
 PASCALL STRIPPED
GEAR CARRY

Fig. 10. Pascal's calculator Fig. 12. Keyboard data entry was introduced
of 1642 used stripped- around turn of the century — but speed of
LEIBNIZ STEPPED gear toothed- wheels to entry was still very slow.
WHEEL CARRY
produce acarry to the
next decade: The Leibniz
machine made use of the
stepped wheel.

111 ITELRELLT.T.LL [
1111

21 0 OS,
,,, I , ....4481d
'
m ,,,
WIIIIII .J.
..=. Iiiini*A
cmtaart, I; r" 0" "VS
0 cell t.,
C ° eel. ET,"
1 . ' - Vre' i 1 0 94
Ili
II J.Ire
¡Mil 111'1 . 1 OVEIVI I

II ,, 411111111118•1111111!‘1.1111

.1 • : Cla
niA.,dmiumn.
r -'
'faun i nummuummamp i
-±-
• ... ,,, . migiimiLmia. log we emmi
g

- '""r-
u es e I ',
el§— MIL
' it,:i•ui
id! 1
"' " . . ..,_ ,..
i
,

du IiinÉM-- -. J inni

Aram rm.,
___ 11111111 I;
1111
71111UMMHII!
7111-31111111e
MULL -1•
1111111111111111g;
I '
11111

18111111111. It
•'
ilaià
, 1111111fflptittme
itt
1
10E. ii,-;
..
1 1111Mre-=''L Nleinlil li 1elm» 1 gm el L- 1;-• -:,1; L
. -
r IffINFV,11 ••, 1 -' • re
,- ' m".',,,
— "le,;.•`'•-..
•••••e 1 1111
,, ,____ a ,• , 1• 'Nunn
Iiimuiy—
„ MY. ineirm • Inimmillimeniri.:Uk w
re lni
,
i
;,‘, mu 1
1
âpienuomernese ft., ' ' ¡
idol.11 . - .11

.',. i'!,.., ,.,,,•

.
.

mminr— .1111M , ,I
ilia u

lil .... e " -i

., r ,
oak. MI Es

..
"............, 1 ,u.....4...P.!ie=4...
1._ il Ent11149471l'i"
Fig. 11. Babbage's difference
engine of circa 1812.
1 ,..
. H. II. Bmhbobre

79
 can be purchased for less than an
hour's wages. The cost of modern
computers is now governed by the
cost of the peripheral bits and pieces
rather than the processing unit itself —
the cost of the electronic components
is now just a minor part of the whole.

BASIC ORGANISATION
The complete electronic data
processing ( EDP) system comprises
hardware and software. The former
pertains to the physical machinery of
the computing system — that which
can be seen to exist in containers and
cabinets. Software is the jargon term
used to cover the multitude of
different programmes devised to
instruct the hardware about the tasks
it has to perform — these may come in
punched card, punched tape, magnetic
tape and disks or in written format.
The hardware of electronic data
Fig. 13. This relay-switched digital calculator was built by Zuse in Germany in 1936. processing systems comprises the
((This photograph has been included because of its historical interest — unfortunately several basic functional blocks
the original print is of border-line quality).
depicted diagrammatically in Fig. 14.
Peripherals enable the electronic
circuitry of computation to com-
amplitude and phase predictor for Germany, built the relayswitched
municate with external information
sea- tide forecasting around 1874. digital calculator ( shown in Fig. 13).
flows via the input and output units.
Later in 1898 Michelson ( of speed of This machine featured automatic The heart of the system is the central
light fame) worked with Stratton to computing, binary arithmetic, floating processing unit ( CPU) comprising a
produce a mechanical harmonic decimal point and punched-tape very fast- access store ( also called the
analyser. programming. In 1937 the USA's IBM high-speed memory) of digital
Special-purpose mechanical Corporation began development of a numbers; a unit that performs
calculators were still in use in the machine called the Automatic simple arithmetic operations at
1940s. During World War II, for Sequence-Controlled Calculator, or, high-speed ( called the arithmetic unit),
instance, gun crews fed data locally, just Mark I. and a control unit that coordinates all
concerning range, direction and wind The trend toward to:al electronic units by stepping ( clocking) the
strength into computers by which the working continued. ENIAC, generally system on bit- by-bit by means of a
correct aiming information for the gun recognised as the first all- electronic clock pulse source.
was computed. computer, had 18 000 valves and could A CPU can serve many different
Today a few equipments still operate at 500 additions per second. functions and all CPUs are not
perform 5.mple operations by This was followed, after many other identical by any means. Typical tasks
mechanical means for in applications developments, by the first production are to control the peripherals and the
where electrical power is not available computer — the Remington Rand input/output information flow,
and the inputs not in electrical form it UNIVAC I. It has been estimated that perform the arithmetic in scientific
may be more economic to use all computers installed in the U.S. in work or compare data in data-
mechanical methods. 1955 could do just 250 000 additions processing uses where the logical
With the advent of electronic per second. Just one low-cost mini can capability is exploited more directly.
amplification at the turn of this do that today. Data is shunted back and forth
century electronic circuitry gradually In 1959 a U.S. refinery installed the between units on the bus lines using
replaced mechanical mathematical first process-control computer system parallel, and serial forms of binary
functions. This was feasible because of and in 1960 a large steel corporation number transfer. ( A number of binary
the super'or speed of calculation, in U.S. was the first to use acomputer bits, when combined into a number,
reduced manufacturing tolerances and to carry out inventories, handle orders are described as words). Different
greater reliability of electronics. The and control production. Airline manufacturers use different word
swing to electronics was intensified by booking by computer began in 1964. lengths — 24 in ICL 1900, 32 in IBM
the need to process an ever increasing Integrated circuits ( in the third 360; 36, 48 and 60 are also used. The
amount of data that arises in, for generation machines) carre into use in term ' byte' will also be met and this is
example, more complex equation 1964 via : he IBM 360 system and by the designation for a short segment of
solving, census taking, or warfare. 1970, in the U.S. alone, roughly the full word. For example an 8 bit
Hollerith devised the punched-card 1000 000 people were employed segment of a24 bit word.
sorting machine to help handle the in making and using digital computers. Words are held in the store when not
U.S. census data. This device won an Single chip, fourth generation being operated upon. As well as being
1890 competition organised oy the machines came to reality around 1972 a binary number that is directly
U.S. Government. with the use of LSI. Today ( or at least equatable to a decimal number, words
Electric computers using the same when this was printed early in 1981) can also represent instructions for the
basic system that we use today became pocket scientific calculators containing control unit to use, a piece of a
reality around 1936 when Zuse, in over 30 000 transistors in LSI form number, a sequence of letters or any
80
 OUTPUT
PERIPHERALS
language; this being achieved by
building more and more automatic
programming functions into the CPU.
The closer the language to everyday
expression the higher the level of the
computer language. Many aids have
POSSIBLE
EXTERNAL
been established to ease the skill
OUTPUT UNIT STORE needed by the programmer in
compiling a workable programme. We
are however a long way from
programming by merely talking to the
machine. Computers must still have
their instructions in a strict written
format. In compiling the exacting
programme sheets the user must first
establish what he wishes to do
FAST CENTRAL
UNIT PROCESSING
STORE mathematically, and detail the steps
required on a flow chart. Figure 15
shows a typical flow- chart for a data
ARITHMETIC
/
UNIT processing problem.
It is not possible to state, in general,
how long a computer may take to
• provide a solution: tasks can take
many hours to mere fractions of
\\ seconds, depending on the size of
INPUT UNIT problem and computer power. Off-line
computations may not necessarily
have to be performed at high speed
(except when other jobs await their
turn) but in process-operations it will
often be vital that a calculation is

INPUT
PERIPHERRALS
made in sufficient time to gain stable
control. Remembering that all

- -)› - - CONTROL LINES calculation required must be reduced

DATA (BUS) LINES to the basic four functions of add,

subtract, divide and multiply it does
not take much of a calculation to
Fig. 14. Basic functional blocks of electronic data processing system.
consume many milliseconds, especially
when the decimal accuracy needed is
high. A particularly fast computer may
other symbols ( eg graphics) as desired. and original language is machine
be essential to obtain millisecond
Words usually include one extra bit language wherein the programmer
time- constant control in computer-
called the parity bit. The parity bit is must specify exactly which bit must
control work.
used as a continuous check that the go to exactly what store location and
words transferred between locations so on. This is seldom used today at
have arrived as sent without any
binary position of asingle word having
operator level. With time, languages
have come closer to normal spoken
X-RAYED
PAST YEAR?IN
its state altered on the way.
In order to make use of this versatile
arrangement, a CPU must have aset of
instructions to tell it when and where
the data has been placed for optimum X-RAYED
PAST 3 IN
YEARS?
use of time. The programme, that is
the software, performs this task at YES
NO
speeds much greater than human Fig. 15. Typical flow chart
f .
for data processing system.
operator or the input units could. A
programme is loaded into the CPU at
the speed allowable by the input
OV ER' 35?
mechanism, during which time the NO IVES
computer is often employed on other
tasks. Once loaded the system is
started on the problem, and then runs SEX READ NEXT
RECORD.
at the maximum speed of which it is
capable. To give some idea of speeds MALE FEMALE
involved, aCPU internal operation will
take around 1ps or less whereas a fast PRINT
ON NAME
peripheral barely gets down to VVOM ENS LIST
100 000 bis per operation. The design
of EDP systems is very much one of
careful systems organisation to avoid
wasted operating time. ADDELIGIBLE
TOTAL 1 TO
TOTALADDINELIGIBLE
ITO I

We say software programmes operate

with various languages. The most basic
81
 ELECTRONICS it's easy! systems shown in Fig. 17. In Fig. 17a,
the data path can be via A or B which
is clearly more reliable than via just A
DEVICE
COUPLERS if both A and B have equal reliabilities.
If an additional path C is added, as
INPUT/OUTPUT INPUT/OUTPUT UNIT 2
MEMORY 1 MEMORY 2
shown in Fig. 17b, we have improved
CPU 1 CPU 2 the chance of data being processed by
DISCS
•--1 1 an M and P unit sequence.
The reliability of systems is measured
in terms of the mean- time between
failures ( MTBF) and the mean- time to
repair ( MTTR). As a guide only, Fig.
18 shows typical values for the various
MAG TAPES kinds of units involved. ( The following
part gives more detail of peripherals
mentioned in the figure).
Taking the idea of interconnected
redundancy to the limit we have a
system schematic like that given in
Fig. 19. At each nodal point any one
DATA
of a multiplicity of units can be
LINES brought to bear on the nodal task. If a
single unit fails, the effect is not atotal
shutdown of the task but a slight
degradation in speed and capability of
the whole system. This has been called
the fail-soft design and such systems
CRTS
exhibit the graceful degradation that
occurs in physiological brains. The
concept of total interconnection is
loosely analogous with the way in
which physiological brain cells are
connected.
In the next part of this series we look
INPUT MULTIPLEXERS /N\
at the peripherals used, various kinds
of stores, microprocessors and the
Fig. 16. Redundant circuits are incorporated so that one unit will continue to perform a vita, latest manufacturing techniques.
task even if though its complementary unit fails.

REFERENCES
Two books, already referred to in Part
ADVANCED ORGANISATION when designing computf:rs for manned 22, are relevant, these are:
Even the best designed digital circuits space shots, the equipment may be "A Computer Perspective", C. and R.
occasionally go wrong or pick up stray duplicated or triplicated — this extra Eames, Harvard University Press,
noise thus causing errors. A single equipment is called ' redundant'. An Massachusetts, 1973. ( This is a
parity check greatly enhances the obvious way to incorporate definitive work on the development
chances of detecting errors, but with redundancy is shown in Fig. 16 where of data processing equipment from
the development of faster machines all units are simply doubled- up and 1800 to 1940).
that conduct vastly greater numbers of connected so that one can perform the "Electronic Computers - - Made
operations in a given time, the task if the other fails. There are Simple", H. Jacobowitz and L.
reliability of the systems to perform preferred ways to connect extra Basford, W.H. Allen, London, 1967.
correctly without error comes into equipment, the general rule being that (Although out of date with respect
question. When reliability is a vital as many cross- connections are made as to certain aspects of hardware this
consideration, as for instance it is possible as demonstrated by the two provides avaluable basis for technical
understanding of both analogue and
digital computers. It also explains the
arithmetrical operations).
A PATH 1 M
"Introducing Computers", M. Laver,
HMSO, London, 1973. ( A version
compiled for users with a little
B PATH
technical knowledge. It discusses
programming procedures).
"Computers at work" J.O.E. Clark,
Bantam Books, London 1973 ( A
most useful book on where
A PATH computers are used).
"Electronic Computers", S.H.
6-1111.— lb)
Hollingdale and G.C. Tootil, Penguin
B PATH
C PATH
Book A524, Harmondsworth, 1965.
(A fine layman's summary of
Fig. 17a. The data path may
analoçiue and digital computers
DATA BUS CARRYING DATA be via A or B. Fig. 17b, if an
FROM INPUT TO OUTPUT
additional path ' C' is added, including a lengthy chapter on what
VIA UNITS M. P
reliability is clearly improved. sort of jobs computers do).
82
 Fig. 19. The fail-safe MEMORY C.P.U.'s INPUT/OUTPUT PERIPHERAL
Computer programming is covered BANKS PROCESSORS CONTROLLERS DEVICES
design — see main
in many texts and booklets. One
text.
example is:
"Elements of Computer
Programming", K.P. Swallow and
W.T. Price; Holt, Rinehart and
Wilson. New York, 1965.
When the need to learn how to
programme acomputer arises it is best
to seek specialised advice about
reading material pertairiing to the
computer to be used. There are
numerous models available each having
its own peculiarities and each requires
considerable operator training-time.
Fortran, by IBM, and its dialects are
commonly used programmes; an

I
DATA BUS
COUPLERS
inexpensive programming primer is:
"A First Course in Fortran", E.J. Burr,
Department of Continuing

_
ig.18. Typical
Education, University of New

Input/output unit
7ean time between
England, N.S.W. 1974 ( Third ?ilures — and

Mag-tape store

C.R.T. display
Device coupler
edition). vpical mean time s...

Line printer
9 repair.
Q)
ALGOL language began to emerge in x

Disk store
a)
1958 as a step toward a universal a
computer language for scientific o •....-
D E
working. COBOL is the commercial 0_ a) 2
(-) 2
counterpart. Relevant books are:

..
"Basic ALGOL", W.R. Broderick and
M.T.B.F. 15.0 6.0 12.0 15.0 3.0 1.5 4.0 10.0
J.P. Barker, IPC Electrical and 6.0
hrs x 10 3 ,
Electronic Press, 1970.
"A Guide to COBOL Programming",
M.T.T.R. 3.0 3.0 3.0 2.0 15.0 2.
0 1.5 2.0 2.0
D. McCracken, Wiley, New York,
hrs' .
1970. •

83
 Computer

31 peripherals,
stores and
microprocessors
ALL COMPUTING SYSTEMS HAVE systems communicate by the same
a Central Processing Unit, ( discussed each row. Figure 2 is a section of
signal formats.
previously) and a number of pieces of punched tape: these are available with
Card and Tape Punches and Readers —
external equipment associated with 5,6,7 and 8 hole positions across the
In order to make good use of the high
them. Such additional units, known as tape width. ( The smaller hole is for the
speed of electronic computing circuits,
peripherals are necessary to handle the timing drive sprocket). Tape readers are
the input and output functions should
flow of information between the built to read code from aspecific width
ideally be capable of transferring the
outside world and the Central tape: that is, a 5- hole tape could not be
data at acomparable speed. Rarely has
Processing Unit ( CPU). used on an 8- hole system. Tapes and
this ideal been realised. The throughput
The range of peripherals available cards which are to be used extensively
rate of peripherals has been speeded up
tcday is extensive. Basically the design can be made in more durable materials
enormously since the first EDP system
aims are to provide interfaces between such as oiled paper, Mylar and
but, similarly, the rate of computation alum inium-Mylar.
the human or automatic plant user and has been increased.
the computing system which are the The holes in cards are produced by
Because of this shortcoming, data ( in mechanical punches. These comprise a
easiest to use, the cheapest to human operator use) is first prepared by
punching head by which the appropriate
implement and which have the means to hand onto a medium that can feed into holes are made for each character in
transfer data as fast as is desired. the EDP system at rates far exceeding response to a typewriter
At present — though this will the operator's ability. It is then stored
undoubtedly change in the future — we keyboard- input. Keyboard layouts are
in the machine ready for access when based on the familiar office typewriter.
are unable to communicate with the the CPU needs it.
Extra keys are added for computer
computer by the same means that we The earliest form of input/output applications to enable agreeter range of
communicate with each other — that is medium used punched holes made in a control by the operator. Such additions
by direct speech and vision. Peripherals, pile of paper cards or acontinuous tape. vary widely.
are by necessity of our technological We inherited these from a 17th century Tape can be punched automatically
and economic I imitations still very weaving machine via the Hollerith whilst the teleprinter type of terminal,
much compromises to the ideal, except census sorter. Figure 1 shows the
such as shown in Fig.3, is used as a
in applications where the computer commonly used Hollerith coded typewriter. Where the tape is generated
interfaces to hardware plant, such as in punched card. The holes are punched as part of an automatic process — as in a
process control, when interface out in a code that represents the data logger, a smaller punch unit is used
problems are easier to solve as such alphanumeric symbols shown above which incorporates punch drivers
activated by control signals — no
keyboard is needed. Such a unit is
UPTO•10“1
COMPCUT
0123456,199 AlqCDEFGHIJKLMNOPQRSTUVWXYZ illustrated in Fig.4.
881111111 1 1 1
1.o.xfu.c. Card and tape readers consist of a
M.C.” 111111111 81 I
1 , 1 11 00000 80 0000000 110081 108011111111 1018101
transport mechanism that passes the
000000 retellIqUunn114.4APIIDNBUI,M,1•POOMMIMP 01,110

1
UOUO•. , U40MPUUM».VUU181

/
fe

1/11/11/11t111/1111/1t1/11111
medium across reading heads.
Recognition of a code represented by
1/111,1111//t11111/11/111111118t/111,11 1111/1111 1

2222227122222222122222/27722222172212212 1722222 /1 2 2M222222222222/2222272222211

holes is accomplished by mechanical
13123333 831231111313341113133311333,31
31333111231133333133333113313333133333333
fingers making direct electrical contact
1 I 1 8 1 8111 (in the slower readers) or by solid-state
ISSM5855$55,1555$155555$55153555155551555 1 5555555 1$ 5 $5151335553555555555153555 optical sensing using LED lamps and
I 11 OiSliii$18 1 photo-diode arrays set to sense the
117111111111171111117177711771777111111111171 1 1117 117117117177111717771111777711
passage of light through ahole position.
Some method of synchronising the code
11 1 1 1 1 1111,11181
position with the data values is essential.
1
000000 rt.e.oumnno..»neue'lls
1
Cards can be punched by an operator
COWMN
at rates between 250-500 per hour.
)
MIAMI

They are often checked on a verifier

Fig. 1. Standard Hollerith Code used for punched cards.
machine that determines if the card is
punched in the same way as the check
operator keys the code a second time.
••
•••• •••• •••• • •••
They can, by contrast, be machine read
e • ••• ••• •••
• or sorted, at 200-1000 cards per minute
•• •• • • • • • depending upon the complexity of the
• • • • • • • • •••
•••• ••• •••• • ••• ••• 1111•• task.
Tape punching is confined to similarly
Fig. 2. Section of 8- hole punched tape. Two rows on the wider side of the central sprocket
holes are not used in this data. slow rates of production at the operator
84 stage of preparation. When the punch is
 Fig. 4. This Philips P803 unit punches tape at 75 c.p.s.

Fig. 3. Keyboard teletype terminals provide hard copy and pet forate a
paper tape (seen lower left) at the same time. The same facility
usually can also print hard copy from a ready made tape.
typically 132 characters and the faster
The compact unit shown in Fig.6 can
machine operated, punching rates can
transfer data at 6000 bits per second at models Can print lines at rates exceeding
rise to 150 characters per second. The 1000 lines per minute.
a density of 30 bits per millimetre of
speed at which punched tapes can be
tape. ( Total capacity on a cassette —
read varies from very slow, using Printing mechanism vary considerably,
mechanical sensing up to 600 characters five miliion bits). These can also be used
ranging from development of the
per second or more with as additional memory in the system.
fundamental typewriter method, to
Printers — Teletype units are able to
high-performance optical equipment. devices that print each character from a
A considerable amount of electronic provide hard copy printout but due to
5 x 7 matrix of dots. Line printers were
logic and drive circuitry is needed to the slow printout resulting from letter
originally bulky units. Today desktop,
operate a punch unit. Figure 5 is the by letter operation they are not used as
typewriter size units, are in common use
block diagram of a reader using brushes the main alpha- numeric output of an
(Fig.7).
to sense the presence of holes. Input extensive EDP system. They can
Printers can be programmed via the
commands to the punch would emanate printout at only 10 characters per
EDP system to provide any format
from the control unit of the EDP second or sc.
required — periodic reports, invoices,
The line printer was evolved to speed
system. records, data lists, software record. A
Magnetic Tape Input/Output Units — up this form of output. It prints all the
crude form of graphical display can also
Cards and paper tape store information characters of a complete line
be produced using the position in aline
about commands to the EDP system simultaneously. Line lengths are
(the programme) and hod the
numerical data to be manipulated. They
are, therefore, aform of permanent data
Fig. 6. Cassette form
storage. They suffice ( ir the form of magnetic tape is
described above) as a data store when finding greater
the data quantity is not great. A recent applicarion as a standard
EDP and computing
trend, which has speeded up data
caculator peripheral.
transfer and reduced the oLlk needed to
store the programme and data, makes
use of magnetic tape in cassette form.

1
Fig. 5. Block diagram of early model Data Dynamics low-spEed tape
reader (30 c.p.s.).
85
 ELECTRONICS it's easy!
demand and to generate alpha- numeric
legends on the plot. It is an easy matter
to replicate the plot — the programme is
run again.
Plotters may be of the analogue drive
Fig. 7. This Hewlett Packard kind ( a later part discusses plotters in
2607A desk- top line printer detail) but due to the nature of digital
provides hard copy output processing the result may still have a
(with 6copies) at 200 lines
per minute. quantized appearance if the resolution is
not sufficiently small. Alternatively the
axes may be driven with stepping
motors — such machines are called
as one ordinate and the lines as the incremental plotters.
The y- t plotter has a roll of graph
other. Flat-bed style of x- y plotters are
paper which is driven at aconstant ( and
When computers are used for
usually adjustable— speed; the pen can available which can handle paper of all
automatic pagination the printer can be be driven in one axis only ( y axis). sizes — from afew centimetres square to
one that produced print-type direct.
Hence the y- t plotter basically plots a size of a wall. A medium-size computer
Graphic Display — Plotters — Many single variable against time. Plotters controlled flat-bed plotter is shown in
computational tasks ideally require a made particularly for computer Fig.8.
graphical display of output information operation will be provided with the Line drawing rates are limited by
not a long list of numbers. Plotters may interface facility that enables direct inherent electro-
mechanical response to
be of x- ytype or y-ttype. connection to the EDP system. ( Normal around 0.4 m/s in small plotters. The
The x- y type of plotter is arranged so plotters require an extensive amount of very large machines, when under tight
that the graph paper is held stationary extra equipment to make them control, are usually capable of around
and the pen is capable of being driven compatible). 0.1 m/s translation rates when working
both vertically ( y axis) and horizontally Computer controlled plotting of x- y to precisions of 25 pm. A desk top x- y
(x axis). plotter is shown in Fig.9.
format has the ability to be scaled on
Some y-t plotters incorporate
bi-directional drive for the t axis ( the
paper drive) enabling very long lengths
of paper to be driven back and forth
along the roll in order to produce an x- y
form of plot from ay- tformat machine.
Graphic Display — Visual Monitors —
Many applications require rapid call-up
of data that is presented in away that
can be easily read by the operator. It
may be quite unimportant to receive it
as hardcopy. The cathode ray tube
(television) type of display was an
obvious choice. Such displays are
known as visual display units, VDU for
short.
Originally, visual display units were
very limited because of the need for a

Fig. 9. The 9862A H-P

plotter has aresolution
of 0.01% full scale
within the 25 x 38 cm
Fig. 8. Series 500 automatic drafting system by Gerber Scientific. area. It will plot vectors
86 at 30 cm per second.
considerable amount of storage with Fig. 10. In this Litton
digital simulator air
which to generate written and graphical traffic controllers learn
display forms. However solid-state mass their skills with make-
data storage is now relatively believe aircraft move-
ments that appear on
inexpensive and VDUs in one form or
their display monitors.
another are now standard peripherals.
The simplest use of VDUs is to display
alpha- numeric information — a section AIRCRAFT
of the software programme, a readout CONTROL '

of process plant variables, airline arrivals

and departures. This is achieved using
digital control and data storage to cause
the beam of the CRT to deflect,
blanking appropriately, to form the 2 " PILOTS"
appearance of astatic written page.
When the operator becomes involved
with the data on the screen and is given
4 STUDENTS
the ability to manipulate it toward a
desired task the terminal is said to be an
interactive graphic terminal. An early
example of this is given in Fig.10 which
depicts a system whereby air traffic INSTRUCTOR'S POSITION

controllers are trained using display

terminals.
Once it had been realised how the
VDU could be used to produce line Instrumentation Interfaces — When the
drawings designers sought ways to developed acolour display terminal that digital computer has to manipulate
can call-up the data recorded by the
'draw' on the screen. The result was the measurement and control data from
ERTS satellite. The computing system
'light-pen'. The operator holds aspecial analogue processes, the system must be
has in its memory file copies of the
stylus on the screen of the CRT. Closed provided with the appropriate A to D
original ERTS data. Using the graphic
loop controls cause the spot to lock and D to A converters, and the
terminal the operator can select which
onto movements of the stylus. If the multiplexing arrangement which forms
form of photograph — I R, false colour,
trace path is to be retained, the x, y and the data logger These interface
intensity coordinates values are fed into etc., to study. He can then rapidly zoom
peripherals were mentioned in the 29th
the digital memory. Once aline is drawn into a particular area using a joystick
part of this series.
it can be retained and regenerated in control expanding the spatial scale as
MODEMS and other links — When
the search becomes concentrated. Other
this way. Other operations enable the computer data has to be transmitted
control includes enabling the colours to
operator to automatically erase sections over considerable distances it becomes
be digitized into level zones and to be
of line, straighten lines and smooth expedient to use telephone lines or
curves by computer processing. The complimented.
complete drawing can then be
permanently recorded as hard copy on a
plotter or as a data set. Interactive
methods have saved an enormous
amount of time in tasks such as deciding
the extremes of a motor- car wheel
movement during the many
combinations of springing and steering
positions within the wheel arch.
Today's graphic terminals are
extremely versatile. Completely
self-contained units which incorporate a
built-in processor are in common use. A
recent release is shown in Fig.11.
Improvements in the storage-tubes
used to hold the displays of a CRT
system have been coupled with the
power of modern computing to provide
display terminals that have half- tone
photographic quality presentation.
Figure 12 shows the quality ( after our
recopying) obtainable. The images
shown are entirely reconstructed on the
VDU from digital, not analogue data.
Colour displays are also coming into use
Fig. 11. Interactive
adding yet more dimensions to the graphic units often now
interaction available to the operator. incorporate their own
A recent project of the Australian processing and memory
to form an off-line
National University gives some idea of
self-contained unit —
the use of the interactive VDU. In the 4051 Tektronix BASIC
Department of Engineering Physics a graphic computing
team of research workers have system.
87
 ELECTRONICS-it's easy! microwave links. Units interfacing
computers over telephone lines have
become known as MODEMS ( a word
built by combining Modulator and
Demodulator). Links are dealt with in
î 1, Part 32 — in which signals and
transmission are covered.
--- 7 Miscellaneous Peripherals — New
methods for communicating with the
o power of an EDP system continue to be
devised in an endeavour to overcome
the interface difficulty humans have
with electronic machines. We are still a
long way from the stage where we need
only casually to talk to the machine.
Steps are, however, in progress toward
this aim with research into spoken word
and written word recognition. Neural
research into brain waves may one day
be coupled with electronic hardware to
provide direct thought- links.
Special cases demand special solutions.
Work at Warwick University in England
has resulted in computer controlled
production of braille maps for the blind.
Automatic mapping and language
translation are other areas where
positive progress is being made into very
complex human communication
processes.

STORAGE
Inside a CPU and external to it will be
found a memory of some kind. This is
Fig. 12. This multiple image presentation is photographed from the screen of DICOMED
used to store the vast quantities of
digital image display unit. coded data needed to perform the
various tasks.
Memory within the CPU is
Fig. 13. (a) When a large enough write current passes in one direction through the ferrite core characterised by the need for high speed
the core becomes magnetised in one polarity. It thus records a bit.
(b) A second wire is added to act as an inhibitor or enhance line.
access to any data bit needed. The
(c) Finishing touches being added to a Philips 3-D core store. (20 planes of 64 x 64 requirement on capacity is less stringent.
cores, one X wire, one Y wire, read and inhibit wires). Memory external to the CPU will, by

WRITE ( OR READ)
WIRE CARRYING
CURRENT PULSE

lal

FERRITE LOOP
MAGNETIC CORE

SELECTED
CORE
(b) SWITCHES
FOR % + Y, IN PHASE
CURRENTS

% UMT
CURRENT

UMT
CURRENT

88
the necessity of machine organisation,
be a little slower to access but it will
usually need much greater storage
1151
capacity.
CPU Memory — Core — storage is
needed in the CPU to hold important G°77

1024 ,81T MEMORY CELL

programme instructions and to act as a 1OF 32
DECODER 32 32
MEMORY MATRIX
temporary home for data generated in
the course of amanipulation.
There are many options open to the BINARY E0
SELECI
designer but the storage method that
has emerged as the optimum for CPU DO
141
storage is magnetic core storage —
known simply as the core store. ( This
situation will, however, soon change,
the preference going to solid-state
methods). Magnetic core storage makes C0
IOF 8 1OF 8 10 8 10 8

use of the fact that magnetically hard HO DECODER DECODER

DECODER DECODER
A10
materials, such as ferrite, will swing
,

remanent magnetism polarity from one MEMORY t 40

ENABLE Gu o
1131

state to the other with the passage of a 1141

quite widely toleranced current through

PIN 116/ = Vc,
a wire passed through the core — see PIN 181= CND.
21
Fig.13a. To make apractical core store INTERNAL

it is necessary that any chosen core can SCHEMATIC

be switched on demand. If a second OUTPUTS

wire is passed through the loop this can

be used to prevent or enhance the G, G, y, y, v,

ho 15 13 112 1 10 19
magnetic switching action by the
passage of the current. Fig. 14. Internal schematic
A core store comprises a plane of and dual-in- line connections MORT
ENABLE
OUTPUT
2à'f.eec'i'iges
ferrites arranged in a grid as shown in of National Semiconductor BINARY LECT
1024 bit, read only memory.
Fig. 13b. Two half-current units 8 lines are supplied to
appearing in the same direction in acore operate readout from the r j f2i la 15 16 17 18

will switch that core but no other. Thus memory using 8 bit code. G D A BC ONO.

two lines will select aunique core in the

plane as the place to store or readout
one bit.
To read out the values it is necessary of propagation. Early computers used access but all are expensive. Many
to interrogate the selected core using mercury delay lines in which the others and cheaper forms of storage can
input signals in the write wires that will, acoustic equivalent of a binary word be used if short access time
if switching takes place, induce currents was sent down a tube of mercury to requirements are relaxed.
in an additional readout wire. As this emerge at a later time at the other end.
process can destroy the data on the core Whilst in transit the word was in Magnetic Tape — This is basically the
atest means may be provided to rewrite same as reel-to-reel domestic tape
storage. The method ( if used at all in a
it again ready for reuse. Figure 13c computer today) would now be recording, magnetic tape storage used in
shows a stacked core-plane. Ferrite computing however records digital
implemented using solid wires or
cores are typically 0.1 mm overall. rather than analogue data on the
clocked- on registers. It has the severe
Planes are either stacked one on the magnetic coating of the tape. Reels are
shortcoming of low storage capacity.
other or mounted flat on a printed Solid-state — Although core storage still generally 10.5 inch in diameter with
circuit board to provide amemory unit. forms part of many computer multiple track use. They are run at
The capacity of core storage varies from much greater speeds than domestic
installations the current trend is clearly
thousands to millions of bits. Core- store units. They can store around 30 bits per
toward the use of asolid-state circuitry
is more usually quoted in word which stores bits in register style millimetre and maybe run as fast as 25
capacity, words being of 32-60 bit flip-flop systems. Read only memories metres/second. Speeds used are not
length. The terminology is to refer to (ROM), content addressable memories standardised to any degree. Each track
capacity as, for example, 32 K of 16 bit (CAM), random access memories on the tape can only be accessed
words. Note the use of the upper case (RAM), and Programmable ROM serially: to obtain a specific data word
'K' — this denotes the multiplier 1024. devices ( PROM) are available as IC chips may involve the whole tape being run
Don't confuse this with ' k' which with typical arrays downward from 512 through with subsequently long access
denotes 1000. Core storage can be eight bit words — that is 4096 bits on a time. Figure 15 shows a typical
cycled with 100 ns ( typically) with single IC chip. Figure 14 shows just one reel-to-reel unit.
some systems taking only 10 ns. The of a huge range of alternatives —
disadvantages of core are relatively high Magnetic Disks — These are thin disks
1024- bit read-only memory. Memories
cost resulting from the labour intensive coated with magnetic recording
such as this exhibit atypical delay from
production method and the com- material. Their advantage is that they
address to output of 36 ns. Chips such
paratively large space needed. can be accessed at any point on the
as these are also available ready
Delay Lines — Another reasonably fast surface by moving the read in read out
mounted as memory cards with as much
storage system makes use of the head to the appropriate part of the disk,
as 65 536, 16-bit word capacity.
delay- line concept. It is the property of as the disk rotates, ( at speeds of 3000
materials, such as mercury, to pass only Peripheral Memory — The storage media r.p.m.). In an alternative procedure the
waves of acoustic energy at agiven rate listed above gives high-speed rapid reading is done by afixed head for each
89
 ELECTRONICS it's easy!

Fig. 16. Honeywell

Fig. 15. The Hewlett- model 4720 moving-
Packard 7970 head disk storage
magnetic tape unit device.
designed as a
peripheral to EDP
systems.

track. Each track may store 36 000 bits. used. These are known as disk packs. short strips of tape that are individually
The moving head disk storage unit Floppy disks are a variation of the disk selected to be drawn through a reading
shown in Fig.16 can store up to 7.5 memory. head, and magnetic cards which are held
million words.
Magnetic Drums — Where better access in magazines ready for automatic
Even greater storage is obtained by
times than disks are needed, but not at sorting in a special console. Card
permanently stacking as many as 72 the cost of magnetic core, the magnetic systems are not as slow as might be
disks on top of each other on acommon
drum may be suitable. A large drum thought — any one of, say, 500 million
drive spindle. Each surface has its own (0.3-0.6 m in diameter) coated with characters can be accessed in 100 ms by
head giving access to any part of any
magnetic material rotates continuously asuitable design arrangement.
surface. Such a unit could store 600
at high speed. Reading heads are stacked
million words. Access time is, however,
up the drum. Access time with these is MICROPROCESSORS
limited by mechanical response times — as low as 5 ms. Storage is upward of
typically 100-300 ms. Small We saw in the previous part that
2000 million characters.
interchangeable disk stacks are also computers are based upon the
Other magnetic arrangements include
availabil ity of a CPU, stores,
input/output units and other
peripherals. Integrated circuit
manufacturing methods become
economical only when very large
volume sales result and it was to the
computing systems market that the IC
makers looked around 1970. The main
problem, however, was the need to
devise a basic general-purpose
integrated- circuit that would satisfy a
large enough group of users.
At first the trend was to manufacture
special-purpose computing systems that
were hardwired ( connections made
permanently) to cause the system to
perform a stated computing function —
such as a pocket calculator for
commercial or scientific computation.
The trend then moved toward another
philosophy — the microprocessor. These
single card integrated- circuit systems
(one is illustrated in Fig.17) possess the
ability to be programmed to perform
the task needed by the customer.
Although the overall system is usually
more complex than hardwired specials,
Fig. 17. This National
the much greater increase in demand has
Semiconductor IMP-8C
reduced the price to quite unbelievable
general purpose
processor uses MOS/LSI levels — a few hundred dollars buys a
devices. complete basic micro-processor system
90
 e

ABA Add Accumulators INS Increment Stack Pointer

INX Increment Index Register
ADC Add with Carry
ADD Add
ANO Logical And JMP Jump
JSR Jump to Subroutine
ASL Arithmetic Shift Left
ASR Arithmetic Shift Right
LOA Load Accumulator
BCC Branch if Carry Clear LOS Load Stack Pointer
BCS Branch if Carry Set
LD X Load Index Register
REQ Branch if Equal to Zero LSR Logical Shift Right
BGE Branch if Greater or Equal Zero
BGT Branch if Greater than Zero NEC Negate
Branch if Higher NOP No Operation
BHl
BIT Bit Test
BLE Branch if Less or Equal ORA Inclusive OR Accumulator
BLS Branch if Lower or Same
Branch if Less than Zero PSH Push Data
BLT
Branch if Minus PUL Pull Data
BMI
BNE Branch if Not Equal to Zero
BPL Branch if Plus ROL Rotate Left
Branch Always ROR Rotate Right
BRA
BSR Branch to Subroutine RTI Return from Interrupt
Branch if Overflow Clear RTS Return from Subroutine
BVC
Branch if Overflow Set SBA Subtract Accumulators
BVS
SBC Subtract with Carry
CBA Compare Accumulators
SEC Set Carry
CLC Clear Carry
Clear Interrupt Mask SEI Set Interrupt Mask
CLI
SEV Set Overflow
CLR Clear
STA Store Accumulator
CLV Clear Overflow
CMP Compare Index Register
STS Store Stack Register
Write COde
ST X Store Index Register Flowchart
COM Complement F unchont to be Using M6800
CPX Compare Index Register SUB Subtract Performed Ins« uchon Set
SVVI Software Interrupt
DAA Decimal Adjust
TAB Transfer Accumulators
DEC Decrement
TAP Transfer Accumulators to
DES Decrement Stack Pointer
Condition Code Reg.
IBA Transfer Accumulators
DE X Decrement Index Register
TPA Transfer Condition Code
Reg. to Accumulator
EOR Exclusive OR 1ST Test
TSX Transfer Stack Pointer to Echt Usong Tone H
Index Register Sharing Computer

TXS Transfer Index Register to

INC Increment
Stack Pointer
WAI Wait for Interrupt Assemble Uhng
Cross Assembler
1

Fig. 18. Typical

Puncher,
microprocessor Assembly
Paper Tape Assembled
instruction set. F ile
Usting

Somulate Using
M6800 Systern Sonulator
On Time She ale
Motorola
Computer
Mask Shop
Computer
mulehon
Listing

with as much power as the minis of a

decade ago. Predictions, at present, are
that they could fall further to a mere
$50.
(e à Punc ,
ne?
,:eper

To make a programme microprocessor

.

system, the user has to write asoftware

programme at abasic machine- language
level. Each microprocessor has its own Enter
ROMs
instruction set built in — this tells the *Age Into Hardware
Simulator

system what to do with data. It is

written in mnemonic code using code
letters to denote operations — such alist Design ver ¡ Id atronl
Hardware S rn , at on

is given in Fig.18.
The programme, thus written in
mnemonic code, is further translated
into the circuit binary form ( object nev.serr
Does
System Meet

code) with an assembler. The object Aiterntrly

L.Snnq
Des in
Goals
form of code is then ready to be fed IFOr Documenta ion Purposes)
into the Random Access Memory
(RAM) or Read Only Memory ( ROM)
of the microprocessor system. Figure 19 Fig. 19. This flow chart
shows a system design and verification shows how the Motorola
procedure used to produce custom Company produces a
made ROMs. The peripherals are then custom-tailored ROM
ready to slip into their
interfaced to the unit and the complete M6800 microprocessor
computing system is ready to go. system.
The process may sound easy, but as 91
 ELECTRONICS it's easy!

can be imagined considerable skill is

this series, however a series of articles
needed to set up amicroprocessor. Such Imp Brief 1, 1974, is also useful as a
devoted specifically to microprocessors starting point.
skill demands extensive customer was published in ETI from August to
training material — large handbooks, "New Blocks for the Computer Builder"
October 1976.
application and hardware — see Fig.20. by D. Aspinall, New Scientist, 18
In fact the stage has been reached where REFERENCES - References listed in September, 1975 gives a basic survey
the previous part provide descriptions including some facts about production.
makers will soon have to face asituation
and illustrations of computer An extensive self-contained
where descriptive literature is as
peripherals and storage methods. introduction is " Introduction to
expensive or more expensive than the
"Computers at Work" by J.O.E. Clark, Microprocessors", H. Tireford, Motorola
hardware itself. To this end they
Bantam Book, is a worthwhile Semiconductor Products, 1975.
continually strive to reduce the
discussion on how and where computer Manufacturers of microprocessors will
complexity and to standardise design.
interface are used for all manner of freely supply descriptive data to aid the
We should soon be in the position where needs.
microprocessors are sold as standard user of their own style of unit.
Infotech International of Maidenhead, The subject of microprocessors has
tools which are each used in exactly the
same way to service the enormous UK, have recently released 12 000 pages recently been discussed in depth in
of state-of-the-art reports on computer several electronics publications —
number of custom jobs available.
operation and trends. They are, Practical Electronics, Wireless World and
Literature may degenerate to a
however, much too expensive for the Electronics Today International ( UK
short- form pamphlet. The actual
reader to procure. The sets costs over
computing part of an Everyman's edition) have each run introductory
$2000 on an individual basis!
Complete Microprocessor System is a series. " Development and Trends of the
"Microprocessors — an Introduction"
very small part of the whole. Peripherals Microprocessor" by J. Tobias, Control
by F. Horne, NS Application Note
and software are now the major cost Systems ( Sydney), 3, 17-31, 1975 is an
AN114, 1974, is abasic statement.
consideration. It is not feasible to extensive study and it includes a
"Microprocessors — Why They Evolved
describe microprocessors more fully in summary chart of dozens of systems
and What They Are" by M. Levi, N.S. offered.

Fig. 20. Array of support products for Motorola M6800 microprocessor system.
92
 Transmission

32 links and
coupling

frequency response that must be how lines can be considered as a

ELECTRONIC systems consist of
adequate for the signal bandwidth to lumped-element equivalent circuit
basic analogue and digital subsystems
be transmitted. Open-wire lines would which can be analysed more easily.
interconnected to provide the required
not normally be used beyond 10 MHz. Depending upon the factors that are
overall input-output relationships. It is
Above that coaxial cables are needed negligible for a particular case the
important for the various subsystems
— these are useful to about 5000 MHz. equivalent can be reduced to simpler
to be interfaced correctly if they are
When currents flow in a conducting circuits — see Fig. 3. For example, at
to perform as intended. But with this
line, magnetic and electric fields are very low frequencies ( less than say
condition satisfied, one cannot just
set up around the wires. Figure 1 100 kHz) amedium length line may be
assume that subsystems merely
shows these plotted for the various represented by the series resistance of
connect together without need to
kinds of cable. Open configurations the cable shunted by the capacitance
consider any other parameters in the
radiate energy, the amount increasing of the line. Typical cables may have a
interconnection process.
with the frequency of the signal. A resistance of around 0.05 ohm per
In practice the individual circuit
line is, in reality, a distributed metre and acapacitance of 100 pF per
assemblies may be geographically apart
inductance and capacitance metre. Hence a long length of shielded
— such as the remote control of
component which also has losses due or open cable could provide a
off- shore oil wells by a shore- based
to the resistance of the wire and the considerable shunting effect that
computer, the recording of test data
resistance to ground. Figure 2 shows attenuates and phase shifts the signal.
from a missile, the control of banking
accounts by a central computer centre
or the sensors of a refinery which
connect to the central control room.
Each of these required some form of
telemetry system. ( Telemetry was
introduced in part 5of this series).
When making connections it is also
important, especially when noise
sources are present that will interfere
with the signal, to ensure that the
signal is transferred from stage to stage
without significant noise pick-up or
signal degradation.

TRANSMISSION LINKS
Several different transmission (a) OPEN TWO- WIRE LINE (b) SHIELDED PAIR
methods exist in which the signal is
confined — open wires, coaxial cables
and waveguides, optical fibres etc.
Alternatively, information can be
transferred via open radiation paths —
radio, optical or acoustic links. The
required signal bandwidth is one of the
primary factors deciding which
method is used. In radiation methods
it is often necessary to use a carrier
frequency higher than the signal
bandwidth dictates because low
frequency carriers will not radiate as
well for the same amount of
transmitted power.
Confined Signal Links: The simplest
(c) PARALLEL-STRIP LINE (d) COAXIAL CABLE
links are formed using an open- wire
circuit ( supported on insulators) or a
multicore cable ( such as is used in MAGNETIC FIELD
local telephone distribution). — ELECTRIC FIELD
Although apparently trivial, lines
may, in fact, be an important part of
the sy.tem. They are not as simple as Fig. 1. Currents flowing in signal wires generate electric and magnetic fields. Enclosed
configurations can be used at higher frequencies because these fields are contained.
they first appear because they have a
93
 ELECTRONICS it's easy! bandwidth some way beyond the
inherent unloaded upper limit. This is
used, for example, to broaden the
bandwidth of submarine cables.
The coaxial cable, shown in Fig. 4,
by virtue of the surrounding external
shield ( Fig. 1) acting as the second
wire, has no external field and,
MAGNETIC FIELD OF LINE therefore, does not radiate energy.
PRODUCES INDUCTIVE EFFECTS — L
Because of this awell designed coaxial
cable will pass from dc to microwave
frequencies — that is, such acable can
have a bandwidth of about 5000 MHz.
Coaxial cable is, therefore, potentially
LCRG UNITS ARE EXPRESSED able to transfer much more
FOR UNIT LENGTH OF LINE
information than open wires. It does
however need a common earth
CONDUCTANCE LOSSES
connection ( asymmetric) and can't be
TO EARTH — G -e— RESISTANCE OF LINES — R used in a balanced mode ( see later).
The bandwidth of practical coaxial
(a) cables is limited by resistive and
dielectric losses. In practice
waveguides are generally used at
frequencies above 1000 MHz or so.
Wa vegu ides co nsist of precise
xR xL
pipework — they look as if they had
been made by a precision plumber!
Waveguides carry travelling
electromagnetic waves of very high
FROM TO SIMILAR frequency and behave vaguely in the
PREVIOUS CIRCUIT OF
same way that pipes carry water. They
LENGTH NEXT LENGTH
cannot however be used for low
frequency transmission.
The cross-sectional area of a
waveguide is inversely proportional to
LUMPED CIRCUIT FOR
LENGTH OF LINE x the design frequency. As ageneral rule
(b) of thumb guide the upper frequency
limit of a waveguide is where the
100 0.5mH
wavelength of the signal becomes one
quarter of the guide aperture —
millimetre wavelength signals ( 50 GHz
or so) being the practical upper limit.
Beyond this, a still wider bandwidth
is obtainable using optical fibre
transmission elements which will pass
radiation in the viible light region
(10 14 Hz to 10 15 Hz). At our current
state of technology, however,
(c) scientists have only been able to detect
the frequencies of far infra- red signals
Fig. 2. Transmission links are systems in which R. L and Care distributed uniformly over (around 10 11 Hz). We cannot, as yet,
the length. For convenience we can consider the line as being composed of cascaded lumped- monitor individual cycles of light with
equivalent elements.
electronic detectors.
(al A length of low frequency telephone line.
(b) Approximate lumped-element equivalent. When the losses of the line are
(c) Representation values for 1km. of medium-size telephone line with earth return. insignificant ( G=0, R=0, in Fig. 2b)
(Actual constants vary widely depending upon design of line). the lumped- equivalent of the
transmission lines reduces to L in
series and C shunting, as shown in
When connecting high example, square waves become Fig. 3b. The nett result is, rather
output- impedance sensors to lines, as rounded as well as attenuated. The surprisingly, that the line exhibits only
little as one metre of cable may be high-frequency performance of the resistance of a fixed value when
sufficient to markedly attenuate the line may be improved by " loading" it looking into the ends. This is called
signal. It's a matter of applying Ohms with inductors placed at regular the characteristic impedance, Zo, for
law to the suitable equivalent circuit. intervals. The inductance value is which Zo = ( inductance per
Because of the reactive effects of the chosen to tune out the inherent unit-length/capacitance per unit
cable the higher frequency signals capacitive reactance at the upper length) 1i. The line appears to be
transmitted will be degraded more frequency where response begins to purely resistive and the Zo value is
than the low frequencies — for fall off, a method that extends the decided by the design of the line or
94
 the power input needs rise enormously
xR for the same distance radiated in free
space. ( The Omega navigation system
uses extremely powerful VLF signals
INPUT OUTPUT because of their ability to penetrate
T XC Fig. 3. In certain practical
deep into the waters of the ocean)
o o cases the lumped equivalent Beyond the gigahertz frequency
x — LENGTH OF LINE
reduces to simpler situations region, circuitry becomes
R — RESISTANCE PER UNIT LENGTH situations. impracticable with current technology.
(a) (INCLUDING RETURN CABLE) (a) Low frequency
Even though the radiated energy
C — CAPACITANCE PER UNIT LENGTH (negligible L
assumption) short must be at a very high frequency to
line in which only operate efficiently we may not
C and R are necessarily need to use the bandwidth
dominant. The R,
C values are found
available on the carrier. The
from maker's modulation techniques we met briefly
data. in Part 5 are used to super- impose a
(b) High frequency relatively narrow bandwidth signal
lossless line
(negligible R and
onto the carrier. It might be thought
G assumptions). that optical and infra- red links use
xL
extremely high carrier frequencies
0--'ooo' • o The input and
output impedances (330 000 GHz for red light) but in
of the line are
these applications the carrier is not
equal and constant
regardless of modulated on an individual cycle basis
length. but rather as variation of acontinuous
dc link. Figure 6 shows what might be
o the first electro- optical link — its
bandwidth would have been barely
150 Hz. In contrast, Fig.7 is amodern
link designed to transmit television
plus speech commands — abandwidth
of 7.5 MHz. Acoustic links using
soundwave propagation operate with
frequencies as low as 10 Hz to well
above the 10 MHz region. These can
be modulated on the individual cycle
basis.

Skin Effect: The alternating magnetic

field produced around a wire has the
effect of causing the current flowing in
the wire to flow at agreater density in
the outer region of the wire. The
higher the frequency the more
pronounced this so-called skin-effect.
At the very high frequencies so little
Fig. 4. Coaxial cables come in many forms. Special, current flows in the centre of the cable
fully enclosed, connectors are needed to ensure no that the centre is often omitted
radiation takes place at connection pipes.
completely, thus a tube is used as a
conductor. For example, at 1MHZ the
majority of the current flows in a
copper cable to adepth of only 60µm
whereas at 60 Hz the distance would
cable, not by its length! Examples are example a typical 75 ohm coaxial be 8.6 mm depth. This also means that
600 ohm telephone lines, 75 ohm cable will have losses of the order of 2 the effective resistance of awire rises
colour TV coaxial feeder cable.This to 5dB per one hundred metres. significantly with frequency — by
means, in practice, that we can factors of 100.
interconnect units on the basis of Radiation Links: Electrical signals fed
into open wires radiate energy out into Process Industry Telemetry Links:
matching all connections to the Zo of
the surrounding medium. As well as Process plants such as oil refineries,
the cable without having to worry
this radiated energy there also exists a paper mills, brick kilns, power stations
about the cable length. If this rule is
"near field" that remains established, and aluminium refining plants are
observed, no high-frequency energy
storing energy. This is the field we monitored by using hundreds of
will be reflected at the termination to
associate with, say, an electromagnet. sensors connected to the control- room
change the information being
As the frequency rises, the ratio of area via instrumentation links. These
transmitted. ( The need for correct
radiated energy to stored energy are invariably wired using shielded
matching was also mentioned in the
increases. For this reason we are able wire or coaxial cable. Because of the
previous discussion about filters).
to build efficient radio systems extreme electrical noise level of such
However, if the line is very long
provided the frequency is kept above plants and low output signal level of
matching must still be applied to
100 kHz or so. Lower frequencies can the sensors these links could pick up
obtain maximum transfer, but account
be used as transmission systems but significant noise thus degrading the
must now be taken of losses. For
95
 ELECTRONICS it's easy! sensor information. Over the years
process instrument suppliers have
standardised the design of the control
systems, and their installation and
HELIAX Elliptical noise pick-up by the cable has been
Rectangular
Circular avoided by several methods.
The first strategy is to superimpose
CONNECTOR FLEX
SECTION .DUAL POLARIZED
the information signal onto astanding
current or voltage thus raising the
•NTENN•

wanted signal level above expected

noise levels. The two systems
ELBOW OR
ELBOW
commonly used transmit the signal
SHORT FLEX
SECTION OPTIONAL
range of the data through 4-20 mA dc
FLEX SECTIONS
OR HELo•X ELLIPTICAL or 10-50 mV dc systems. An 0-20 mA
RIGID HANGER WAVEGUIOE
system is also common. Current
DUAL POLARIZED 4,)
TRANSITION transmission has the advantage that
RIGID HANGER the circuit is of low impedance — a
RIGID HANGER
EVERY 3 FEET ----I few ohms — which reduces the level of
induced noise power. Figure 8 is an
SLIDING HANGER'''. SLIDING HANGER example of these practices —
EVERT 5 FEET EVERY 6 FEET
Honeywell's arrangements used to test
the temperature and pressure of
natural gas wells in the Leman Field of
the North Sea.

SPRING RANGER
EVERT 50 FEET SPRING HANGER
Safety Precautions: Often the sensor
EVERT 20 FEET
has to be placed at alocation where an
explosion could result from aspark or
excessive overheating of a
malfunctioning sensor circuit. The
WALL ROOF WALL/ROOF
FEED THROuG FEED THROUGH most obvious way of overcoming this
AT., RATIO
CONPENSATOR-
is to place the whole unit in an
explosion- proof enclosure. This,
SPECIAL
LENGTH however, has disadvantages: the cost is
high, and testing and maintenance
CONNECTOR difficult due to the need to shut off
---- GAS INLET
PRESSURE INLET
the power when the enclosure is
eXeN SECTION
opened.
SSORE
WINDOW vo, EsSuRF The alternative, more modern,
wrvoo.
NELà•A ELLIPTICAL method is known as intrinsic safety.
WAvEGLNOE RuNS TO
BUILDING AND EQUIPMENT As inflammables require a specific
level of energy to ignite them,
Fig. 5. Waveguides transmit electro -
magnetic energy via travelling fields that are conveyed
explosion can be prevented by
through the pipework. These are three forms of antenna feeder offered by Andrew
Antennas. ensuring that the sensor stage cannot,
under any conditions, provide enough
ignition energy. No enclosures are

_
Fig. 6. This Photophone of Bell and Tainter was designed in 1881. Sunlight reflected to the receiver was modulated by acoustic waves
vibrating the speaking tube mirror. Detection was with a selenium photo-electric cell driving earphones.
96
 BEAM B'
4 SERVICE SPEECH ONLY

BEAM A' P ROGRAMME

PROGRAMME 1 IMAGE PRESENTATION
ORIGINATING 2 SOU ND BOX P.P.B.
BOX • P.O.B. 3 SERVICE SPEECH
PROGRAMME PRESENTATION TERMINAL
PROGRAMME ORIGINATION TERMINAL

Fig. 7. Schematic with

photograph of the optohead
of a Leevers-Rich opto -
electronic communication
link. The output from a
light emitting diode is
modulated by the incoming
signal. The receiver detects
the modulation with a
solid-state photo-detector.

TELEMETRY
TO SHORE

MV/1 — MILLIVOLT TO TELEMETRY DEMAND

CURRENT CONVERTER INPUT

ADC — ANALOGUE TO DIGITAL

CONVERTER 10-50 MV DC

420 MA DC

MV /1

TYPE
THERMOCOUPLE
TEMPERATURE ONE PEN
WELL TEMPERATURE PROGRAMMER RECORDER
PROGRAMMING
RELAYS

11 OFF OTHER TELEMETRY

TEMPERATURE TO SHORE
SIGNALS
A.D.C.

1000 - 3000 MV/1

PP/1
P.S.I.
Fig. 8. Process measure-
ment link arrangements
PRESSURE ONE PEN used between the oil-
PROGRAMMER RECORDER
WELL PRESSURE well and the off-shore
platform. The A/O
units send the data to a
11 OFF OTHER
shore-based computer by
PRESSURE digital telemetry links.
SIGNALS
97
 Rl R2
FUSE
Fig. 10(a). Chart showing common com-
bination possibilities of various output to
HAZARDOUS
SAFE AREA input cascaded schemes incorporating
AREA DI 02
TERMINALS amplifier stages of various kinds between
TERMINALS
the first stage and the two commonly used
Fig. 9. Circuit used in a
o o Zener barrier.
output recording/monitoring connections.
(Courtesy Siemens Industries).

Measuring source Earthed Asymm. on Earthed Isolated Symm on Isolated

Isolated Earthed Applicabild y

asymm. asymm. v7in

c re meke symm.
COMMOn .. rnOde 10110W up•
follow-up
sY mm • voltage unit Instrument of the
13 "'''
o amplifier

qt— P
Type of amplifier F { E
\ ••• ' II '11
V:7e*Yel
asymm.
I
.. without
. i
ll '
., i restrictions
. ,:ttt /z
Iso lated
asymrn. ':'

\
\ -.
fIl' I with
4J1 ii.:
Ie
«:Ñ:' .
'
:.,\\, v. restrictions
Earthed ,'
(earth loops)
.,

"•:•:":'.. .
.1
„
-•
r i
.... '''- ' . 7',...,.kee:'. ..-.-':'• -
7.:•72.
,; with
Isolated "
:
1.>:•::::::
::::: • ..E.--
symm. • <, restrictions
f, r
tIr (height of the
It, common- mode
rt77-7.r.5r.'•'.'"". t't c't t'-- ' t .. . . .-'-- voltage)
Isolated t .,-•,«

screened
,, •ie. -. with restric-
i _ honS(meas.
"

r . Iowan
Isolated • source can

, , . . . : , . . i . i .
syrnrn . •••,, - • • , , ., i • ,:,:,::-:.:
.:.. R.,' :• be earthed)
, ,•

.„:. ..., . .
screened
,
!MI T I!

Isolating not possible

amplifier '• .-.-. .... • • -

_ A

needed and the circuit can be

overvoltage or induced earth- loop 240 V encounter will provide ashock
maintained whilst it is operating. currents cannot enter the isolated
Or the concept was at least double the fibrillation level!
hazardous area.
implemented by ensuring the sensor The instrumentation must, where the
In electro-medical instrumentation,
circuitry could not draw, or produce metal parts are earthed, be wired with
safety precautions of another kind are the active, neutral and earthing wires
via storage, more than a specified
vital to ensure the sensor does not act connected correctly. Double- insulated
power level. This level was found by as a pathway for adangerous level of
experiment in atest rig set up for the systems avoid this problem. Earth-
electric current into the patient. At
situation involved. leakage balanced — core breakers
240 Vac the human body's resistance, are worth using. These detect
The more recent idea is to use hand to hand is around 2000 ohms —
"safety barriers". At the exit from the minute difference currents in the
100 mA will flow. If totally connected
declared hazardous area, the cables active and neutral, tripping abreaker if
(as by a conducting fluid) the
terminate into a zener-diode and they rise above milliamperes.
resistance reduces to 200 ohm — 1A
attenuator arrangement which ensures The sensor attached to the patient
will flow. About 75 mA through the
that the current and voltage entering must not be capable of providing a
body will produce heart fibrillation; lethal level of energy by means of
the area are limited to safe values. only 150µA, through the heart itself,
Figure 9 shows the circuit of a zener feedback from the instrumentation.
is needed to produce this effect. A The reading list provides more detail
barrier. Another safety device uses a person can usually hold ( with the of safety factors in electro- medical
solid-state closely- coupled fingers) and release as much as a instrumentation.
electro- optic link which provides dc 10 mA, 240 Vac current — beyond
electrical isolation between its input that the muscles become paralysed. COUPLING STAGES
and output, the information being Skin moisture largely decides the hand Connection arrangements: As was
transferred from alight-emitting-diode to hand resistance. When dry it will be pointed out in the discussion of
mounted next to asilicon photo-diode (at 240 V) 2500 ohms and moist, meters, in Part 3, electronic
detector. These ensure that 1000 ohms. Thus a hand-to-hand
sub- systems must be cascaded
98
 SIGNAL/ENERGY FLOW

1- Hot pole
E- Earth pole

R2
E LOAD
Earthed asymmetrically INPUT
on one end IMPEDANCE

1,2 - Hot poles

E • Ear th pole PERFECT
1or 2can usually STAGE 1 LINK STAGE 2
be earthed
2
isolated asymmetrically

Matching requirement relative R1, R2 values

1,2.- Hot poles
E - Earth pole
Earthing of 1or 2not admissible
(connected to common- mode R1 = R2
potential)
Maximum energy transfer

Asymmetrical on common -
mode potential

Maximum voltage signal transfer R2 > > R1

12 - Hot poles
(least voltage loss across R1)
3E - Earth pole
-,i3E Earthing of 1or 2not admissible

2 Symmetrical center
point earthed Maximum current signal transfer R1 > > R2
(least current change due to R2)
12,3 - Hot poles
E Earth pole
Earthing of one pole, preferably
no 3, usually possible

Test voltage
Isolated symmetrically
oCommon- mode
voltage to which Fig. 11. Summary of impedance values for
1.2,3 - Hot poles the test voltage various matching requirements.
E - Earth pole is applied
Earthing of 1or 2.or 3
not permissible

Symmetrical on commoi - (b) Extra detail

mode voltage of source
arrangements.

intelligently or loading of the output amplifier and output device. of ten to one hundred times is usually
of a stage by the input impedance of Not- possible situations usually arise sufficient.
that following may degrade the signal. because the earth connection shorts The opposite situation, that is,
Output configuration of the various out one of the source arms. loading a high output impedance stage
stages involved in instrumentation can with a low input impedance, arises
Matching: Three basic matching
take many forms depending on how when the maximum current transfer is
criteria exist when connecting two
the earth is connected and if the signal required.
stages together. Figure 11 summarizes
is symmetrical or assymmetrically In many cases the appropriate buffer
these.
connected. The six commonly amplifier is required to provide the
encountered source output schemes If the need is for maximum power desired matching condition. In certain
are shown at the top of Figure 10. On transfer, as when driving a ac coupled systems — those which do
the left-hand side are seven common loud- speaker from an output stage of not require adc path between stages —
kinds of amplifier connection ( any an amplifier, the output impedance a transformer can provide an adequate
other form of black box could be (usually thought of as an average value impedance match in an economic way.
regarded similarly). On the right-hand of resistance) of the driving stage must Transformers, however, have limited
side are leader lines that show a link equal the input of the stage being frequency response and must be
between the output of the chosen driven. When maximum voltage chosen carefully to suit the signal
amplifier and one of the two most transfer is required, as occurs when a requirements.
commonly used instrument pick-up cartridge or other voltage
connections — fully isolated circuit generating transducer is used or when Eliminating noise: In the ideal
with case only grounded, or one pole measuring a voltage in a circuit, the situation any circuit added after
grounded to earth. Using the legend, rule is to ensure the connecting stage another should add no more noise
the chart shows the applicability of has a much higher input resistance energy to the signal than is fed to it.
connections between chosen than the output resistance of the stage We specify the ratio of the two as the
combinations of source arrangement, producing the voltage signal. A factor signal/noise or S/N ratio. In practice
99
ELECTRONICS it's easy!

RECORDER CHASSIS
RECORDER CHASSIS

SIGNAL SOURCE OSCILLOGRAPH

SHIELD srdr Lip

F - F;0,1•Su
F- FLOATING
G-- CHASSIS
G- CHASSIS GROUND
POTENTIAL DIFFERENCE • --1
POTENTIAL DIFFERENCE -e"
SIGNAI GRMIND StsIl , 1'.1.IND sll;NAL GULA,ND
SYSTEM GROUND

RIGHT WRONG
Fig. 12. Correct and incorrect methods of joining a sensor to a recorder. Most output instruments offer the user the choice of leaving the
instrument floating above ground or grounding it.

all circuits, including connections, will operation, source resistance and connection to mother earth is a good
add a finite amount of noise — temperature — the correct choice of ground or earth point. But this is not
degrading the S/N ratio. An amplifier components to yield a low NF is a so for instrument stages operating at
or other cascaded stage should ideally skilled task. millivolt and microvolt signal levels.
increase or modify the signal Connections between stages are most Signals as large as volts can be induced,
amplitude without reducing the S/N common source of noise addition. or dropped, between two points of a
ratio ( input noise will be amplified Observing several basic rules will metal chassis! The rule for avoiding
equally with the signal). usually greatly reduce the noise this ground loop problem is to attach
A common unit used to describe the pick-up in wiring between and within all circuit points required to be
degree of degradation is the stages. grounded to a substantial size copper
noise- figure NF which is calculated as Grounding and Shielding: When wiring bus bar — the circuit ground — that is
NF = 10 log ( Signal in/Noise in, circuits and inter-connections the grounded to earth at one place. Better
divided by Signal out/Noise out); the circuit diagram shows a signal ground. still, use a single common connection
ratio being expressed in decibels ( dB). (Terms ground and earth are used point.
The perfect additional stage has NF = somewhat synonymously). This line Shields of cables are too often
0 dB, So-called low-noise amplifier is assumed to be at exactly the same assumed to have the same potential at
stages will have noise figures better potential at all points where a ground each end, both ends being presumably
than 3 dB ( S/N ratio reduced to half). symbol is indicated. From the at ground potential. This is often
NF is a function of device electricity supply authority's incorrect for the shield becomes an
characteristics, frequency of viewpoint any good low resistance earth-loop having a finite resistance
when both ends are grounded. Only
one end, the input end, should be
earthed and the shield should be
-r-
IMTGRbl NG WIRE
insulated against earth at all other
points. Figure 12 shows the right and
C2
wrong ways to connect two stages
together with ashielded two-core lead.
AMPLIFIER
Special quality low-level signal cables
Fig. 13. Common-mode are available. These incorporate an
pick-up of noise is balanced inner twisted- pair that is wrapped
out by the differential inside a multi-layer metal foil along
„, GROUND PLANESSt7
system.
with a bare copper drain wire, the
whole being well insulated.

Common- mode rejection: Before other

— DISTURBING WIRE aspects of connections with cables can
be appreciated we need to study the
principle of common- mode signal
rejection.
Fig. 14. Twisted wires We begin by looking at the noise
provide best common-mode pickup from supply mains radiation by
rejection when making two open wires used to complete a
connections between
SOURCE link, as shown in Fig.13. If both wires
stages in which a symmet-
rical signal source is joined are at the same potential above earth,
to a differential input that is, neither is earthed, the noise
amplifier. pickup in each wire will be closely
100
similar. One wire, however, passes ROOM NOISES, BRUSHES
MOTORS.ETC.
NUMERIC
INDICATORS
BROADBAND
NOISE
INIE
ABOVE 1GHz
signal currents in the opposite 50 INTERNAL SHIELD
REOUIRED
direction to the other so noise induced
in each wire will add to the signal in
one wire and subtract in the other — HP SYSTEM 11 STANDARD CABINET
IWITH ADDED SHIELDING)
the result is that the noise just about 40

balances out. This is known as a

common- mode rejection arrangement.
HP SYSTEM 11 STANDARD CABINET

It is a balanced system as far as

unwanted signals are concerned
because of the use of a differential
arrangement.
HP SYSTEM 111PLAST C CABINET
(WITH ADDED SHIELDING)
The same concept is used in
low- noise, high gain, dc amplifiers —
see part 11 — to eliminate transistor
defects. A slight disadvantage of
differential configurations is that
many testing instruments operate with 10
HP SYSTEM 1STANDARD CABINET
one grounded input. Connecting an
oscilloscope to probe a
differential-mode circuit may short
out a line to ground in certain 4 6 . 8 10 10 20 50 100 200 300 400 500 600 700 800 900 10 MHz

connections. For such work a

differential input amplifier is essential
Fig. 15. Shielding of RF energy by various designs of enclosure used for H.P. instruments.
in the oscilloscope.
The actual value of aparticular unit depends upon the need for holes and shafts through
Once the signal level has been the panels.
amplified well above the ambient noise
levels the symmetrical dual output can signal levels that they can handle. If are to be found in works on
be converted to a single pole with the induced signals are too great in telecommunications. "Transmission
earth output, using a suitably amplitude they may saturate the lines and Networks", W.C. Johnson,
connected operational amplifier. amplifier, removing its ability to 1950 is a theoretical treatment of the
operate correctly. It is, therefore, use of lines.
For the best low-level signal transfer,
wiring between stages should observe always best to reduce interference at Modulation methods are discussed in
the common- mode principle, the aim source rather than attempt to most general text books on
being to make each wire of the pair eliminate it by common-mode electronics. Safety in medical
appear as identical as is possible to the rejection alone. electronics practice is the subject of
interfering noise sources present. "Hazards of Electrical Apparatus",
R F Shielding: Mains frequency
Figure 14 demonstrates why the J.M.R. Bruner, Anaesthesiology, 28,
interference ( 50 Hz) is comparatively
twisted pair is better than two separate 396, 1967, and " Electrical Hazards
easy to eliminate from or retain within
lines to connect a symmetrically- Associated with Cardiac Pacemaking",
equipment by using low conductivity
connected source to a following R.E. Whalen and others, Ann. N.Y.
enclosures. RF interference, however,
differential input stage. The Acad. Sci., 111, 922, 1964.
tends to penetrate the best designed
distributed capacitances of the two
enclosure — remember waveguides
wires are different ( with resultant
transmit RF — through apertures of
different pickup noise) in the A useful short guide to the practice of
size similar to wavelength. Cracks,
open-wire case than they are in the interconnecting system units is to be
where covers join, may act as
twisted line. found in " Elimination of Noise in
waveguides for UHF signals. As
Low- Level Circuits", Clevite
Shielded two-core cables used with a modern circuits operate with
Corporation ( Brush Instruments
symmetrical outputs source should transition times of nanoseconds they
have the shield grounded at the source, too generate considerable quantities of Division). A more extensive treatment
not at the following stage. The latter RF energy. By way of example of is " Grounding and shielding
option degrades the common- mode what can be achieved by careful techniques in instrumentation", R.
mechanical construction Figure 15 Morrison, Wiley, 1967. The topic is
rejection capability.
compares different instrument en- also covered in a treatment on low-
Common- mode principles must be level techniques by PAR — see ETI,
closure designs of a manufacturer.
carried through completely in exacting February 1972 for " Signal to Noise
Slots introduced into frame elements
low-level signal applications, even to Ratio — its Optimization in Precision
form wave- traps ( as opposed to wave
providing identical terminating Measurement Systems" by T. Coor, A
guides) when the metal covers are
conditions at the wire ends — similar similar article is " Modern Signal
bolted in. Modern instrument
length open wire ends, similar, Processing Technique for Optimal
enclosure design is as much a case of
dissimilar- metal, conditions at terminal Signal to Noise Ratios", R.D. Moore
containing RF radiation inside the unit
posts with identical temperature for and O.C. Chaykowsky, Princeton
as it is to prevent it entering.
each to ensure identical Allied Research Corporation,
thermo-electric currents are generated Technical Bulletin, No.109, 1963.
in each lead. REFERENCES "Taking noise out of Weak Signals",
Active devices, such as amplifiers, In depth discussions of the theory R. Brower, Electronics, 41, 80-90,
have a limit to the common- mode and practice of transmission methods 1968 is also relevant.
101
 Oscilloscopes

33
OF THE MANY INSTRUMENTS fields. Prior to 1897 interest had been
Figure 1 shows the three stages in
required to service, test and maintain developing the basic cathode ray tube. in physical-science investigation — not
electronic systems, the cathode-ray in the measurement of electronic
Fig. 1(a) is a thermionic diode — a
oscilloscope must be the most versatile signals. Then in 1897 K.F. Braun
valve diode. The cathode, heated by
and useful. Other names are derivatives produced the first basic measuring
the current passing through it, emits
from the full name — the C.R.O., CRO device from the CR tube.
electrons into the space around it.
(pronounced crow), oscilloscope and These, being negatively charged, are
scope. Early works also refer to it as
attracted to the positive anode. FURTHER DEVELOPMENTS
an oscillograph.
The greater the voltage between the However the CRO to become a
The basic workings of the oscil- cathode and anode the greater the
loscope were introduced in Part 4 useful, practical instrument more
velocity of the electrons. If a hole is
where it was explained how they are development was needed. From
made in the anode, as in Fig. lb, many Fig. lb it can be seen that the beam of
used to view signal- level variations as of the electrons will pass through,
time-versus-amplitude graphs drawn on an elementary device is badly defined
forming adiverging beam on the other
the screen by electrons. This part and floods over the entire area of the
side of the hole. When a phosphor
provides a deeper coverage of the phimphor. A tube or grid arrangement
powder is placed on the inside of the placed between the cathode and anode
capabilities both simple and advanced
tube the electrons reaching it cause it
units. We do not delve in the circuitry; causes the beam to pass through the
to glow as they give up their kinetic anode more cleanly, because of the
although electronic system builders
energy. The powder re-emits this
will always need to use an oscilloscope negative repulsive effect of this tube
energy as photons of visible light.
they will rarely need to build one. assembly. The whole assembly —
Early researchers' tubes did little more
cathode, anode, grids and tube — is
than this. The nature of cathode rays called the electron gun. Its full design
THE CATHODE RAY TUBE was studied in the early 1900s by such
is quite complex: Other elements are
The first cathode-ray tubes were famous names as Goldstein, Braun,
used to make electron-lenses ( akin to
experimental, designed to investigate Crookes, J. J. Thompson, Rontgen,
optical lenses and light) to focus
the nature of beams of particles Coolidge and Dumont. Experiments
control and an intensity control, the
produced in thermionic-diode arrange- showed that the beam could be
former adjusts the spot shape and size
ments operating at extremely-high deflected by apermanent magnet and
on the screen, the latter the current
voltages. by electro- magnetic and electrostatic flowing in the electron beam.
GLASS ENVELOPE
ANODE

FOCUSING
STREAM OF TUBE
—Ve CHARGE ARRANGEMENT
ELECTRONS

HEATED
CATHODE
Fig. 1. The basic cathode ray tube developed
(b) A hole in the anode allows some
through stages to provide agun aimed at (c) A tube arrangement or grid is added
a screen. electrons to pass through to the phospho- to form a more concentrated and smaller
rescent screen. size beam.
(al Therm ionic diode in which the cathode
is self-heated to cause liberation of electrons
which move to the positive anode.

Phosphor Fluorescence Phosphorescence Persistence

European/ Burn Relative Comments
resistance luminance
U.S. code

GP/P2
Bluish- green Green 10 its-- 1ms Medium high 55%
GM/P7 Purplish- blue Yellowish- green 100 ms-1s Medium 35% Medium speed oscillography
BE/P11 Low speed oscillography.
Blue Blue 10 lis-1ms Medium 15%
GH/P31 Best photographic writing speed.
Green Green 10 us-1ms High 100%
General purpose oscillography.
Brightest available phosphor.
GR/P39 Green Green 5-100 ms High 50%
Sampling oscillography.
Fig. 2. Chart showing characteristics of oscilloscope screen phosphors.

102
 Ne AXIS
The choice of phosphor on the DEFLECTOR
screen determines the persistence ( the CATHODE PLATES
GRID
FILAMENT
length of time the spot glows after
removal of the beam) of the display.
The storage effect of various
phosphors enables CROs to be made
so that beam energy can be dispersed Fig. 3. The electron beam can be
as light over time durations varying deflected by voltages applied to
from microseconds to milliseconds. deflection plates.
Fig. 2 is a guide to the selection of a
phosphor. Manufacturers often offer a
choice of screen persistence values to to be read in conjunction with the
the instrument viewpoint the differ-
suit various applications. Fast moving schematic of Fig. 5 and the panel
ences have been improvements in
spots, where the spot is likely to layout of a portable unit shown in
frequency response, spot control,
reappear on the same point in ashort
linearity of sweep and awider choice Fig. 6.
time, require short persistence. Long-
of phosphors. In addition the develop- Time- base: If the X plates are driven
persistence screens are suitable for
ment of tubes with more than one gun by a signal voltage that increases pro-
slowly changing signals. ( See the
and deflection system ( some dual- portionally with time the electron
discussion of storage methods in the
beam oscilloscopes, but not all, use beam will be deflected across the tube
next part.)
separate beams for each channel) and at a steady speed. When the signal
storage tubes which enable the returns to its original value the spot
ELECTROSTATIC DEFLECTION
effective persistence to be varied at
The next refinement provides a returns to begin the next sweep. The
will have greatly improved the versa-
method by which the beam can be waveform required to produce such
tility of today's instruments. linear deflections is a sawtooth.
made to deflect under the control of
electrical signals. Fig. 3shows how this (During return the beam is normally
blanked out.) This provides a sweep
is done for one axis, using the electro-
TURNING THE TUBE INTO A function. The period of the sawtooth
static method. A voltage difference of
MEASURING INSTRUMENT determines the time taken to cross the
zero between the deflection plates
In the niajority of cases the CR screen; this is expressed in the units of
allows the beam to pass along the tube
tube is used to produce a graphical time per division ( screens are divided
axis undeflected. Any voltage
display with the amplitude of asignal into a grid of centimetre squares by
differential will cause the beam to be
being expressed in the vertical ( Y) means of plastic graticule or by
deflected towards the more positive
direction and its variation with time engraving the inner face of the tube).
plate. Thus we have away to cause the
being along the horizontal ( X) A selector switch in the time-base
beam to move in the vertical direction
direction. The following description is section of the panel ( an example is
(called Y-axis or Y plates). A further
two plates set at right angles to these
(the X plates) will cause the beam to
deflect in the horizontal plane when a
voltage is similarly applied to them.
Beam-intensity control by electrical
means is defined as the Z control.
Electrostatic deflection is the
easiest to deploy for voltage measure-
ments because deflection is pro-
portional to applied voltage. Small
cathode ray tubes usually use electro-
static deflection. Large tubes, such as
those used in television systems or
large-screen teaching oscilloscopes,
usually use magnetic deflection
because electrostatic deflection would
require very high deflection voltages.
These do not have deflection plates set
inside the tube, but make use of
magnetic fields created by electro-
magnet coils placed around the neck
of the tube. The deflection in this case
is approximately proportional to the
current in the coils.
Cathode ray tube design ( for CROs
and TV) has remained reasonably
static since the late 1930s, the only
obvious differences being in the
linearity of beam sweeps and the
shortness of tubes for a given screen
Figure 4. The insides of a modern oscilloscope — the Trio CS- 1562. The cathode ray tube and
size in television applications. Figure 4 its cover have been removed. Note the tube is much longer than the tubes used in TV sets
is a modern oscilloscope with the (when you consider the small screen area). The cover screens the electron beam from the
cover removed to show the tube. From influences of stray magnetic and electric fields.
103
 ELECTRONICS it's easy!
CATHODE RAY TUBE

n INTENSITY
CONTROL ( z)
Fig. 5. Schematic diagram of essential
sub-systems added to the CR tube to
make it a measuring instrument.

VERNIER y AMP
A AMP
VERNIER

TIME BASE

/wm
SWITCH

DEFLECTION
SENSITIVITY TIME BASE
SWITCH DIRECT T.B. GENERATOR

INT.
o
TRIGGER
EXT./ LEVEL

HORIZONTAL
VERTICAL
IINPUT
INPUT (X) 2
(Y) EXTERNAL
TRIGGER
".=
INPUT
VERTICAL CIRCUITS
- L HORIZONTAL TIME BASE CIRCUITS
—J

Fig. 6. The controls of this mini-portable oscilloscope are typical of basic units. The position of controls around the screen will vary from make
to make.

OPERATING CONTROLS
POSITIONED ON FRONT PANEL VARIABLE TRIGGER
INPUT COUPLING SELECTOR LEVEL CONTROL

4 x 10 DIVISION GRATICULE
WITH 10%, 90% RISETIME
VERTICAL STEP ATTENUATOR
REFERENCE LINES
WITH VERNIER CONTROL

VERTICAL POSITION LED POWER ON LIGHT AND

FOCUS CONTROL ON/OFF/CHARGE SELECTOR

ROPIZO NAL F1 SWEEP RATE

VOLTS, DIV DR
CONTROL WITH
X10 VERNIER/
MULTIPLIER
FL 00
INTENSI r

XI IN1

%/ L
i - DATA
CarPCOliCKON
am.%
gad,
ri ralth TRIG IN
- DI DC AIM nen-

HORIZONTAL
INTENSITY CONTROL
POSITION EXTERNAL TRIGGER
(1 pF)

GRATICULE AND/OR FILTER EASY TO

INTERNAL/EXTERNAL TRIGGER,
REMOVE AND INTERCHANGE WITH CUSTOM
SLOPE AND AUTO/MANUAL
GRATICULE OR SPECIAL FILTER
MODE CONTROLS

VERTICAL INPUT BNC CONNECTOR;

104
 Fig. 7. Controls of abasic
î le

TRIGGER SELECT time-base unit include those

VERNIER
shown on this plug-in. POSITION

r „
NORMAL
Terminology is generally

oN lU
ti the same for all makers
but layout and controls
will vary. eC e
AL

CAL
LE VU
TB VERNIER

NOR= • :: VEnt
SENS / 01V

DC
TIME / DIV
BAL
.5 . 2

5 20

10 ,0

20 5

rr
AC
OUP
Fig. 8. Single- trace amplifier or, eiti HPUPpo
unit. These can be used for
both vertical and horizontal GLIC DC

amplification. bwd 6B 50MHz AMPLIFIER

bwd 61. TIME BAST

Time Base Amplifier: The voltage (Attenuator probes also enable high
given in Fig. 6) enables the sweep rate
to be chosen to suit the period of the required to deflect the beam over the voltage signals to be investigated.)
signal being examined. Basic units will full X ( or Y) traverse is of the order of The application needed from Y-
have time bases which range from hundreds of volts. The time- base amplifiers can, therefore, rise to
0.5 µs to 0.1 seconds per centimetre; generator therefore requires an ' X' 100 000 on the most sensitive range.
sophisticated units can go as slow as voltage amplifier between it and the In addition it is important that the
10 seconds per division to as fast as plates. gain be constant over the bandwidth
ins per division. ( Special "sampling" In certain applications the X plates of the signals being monitored.
plug- ins can provide 10 ps/division.) are used with signals in the same way Basic units provide amplifier
The time- base sawtooth generator is as Y plates — that is without atime- response flat from dc to amegahertz
an integral part of all CRO measuring base signal. In such cases considerable or more. ( Bandwidths are defined
instruments. The accuracy of the rates amplification may be needed. More between points 3dB down from
are determined by circuit components versatile CROs offer plug-in facilities maximum.) Magnetic-deflection
— more expensive units can provide for the X input to give the user awide display monitors will only reach
choice of functions from the one unit. 20 kHz whereas sophisticated high-
more- accurate information. A further
control in the time- base section ( See Simple units however, have the ' X' quality instruments have bandwidths
Fig. 7) allows the switch-selected amplifier wired in permanently. rising to 350 MHz_ Sampling plug- ins
provide bandwidths equivalent to dc
sweep rate to be varied continuously.
This is usually referred to as avernier Vertical Inputs: At the same time as to 1GHz.
the time- base circuits sweep the line Vertical amplifier controls are
control. When making time measure-
across the screen the 'Y' - plates are usually grouped together on the front
ments, such as waveform period, it is
driven with a voltage proportional to panel, as are time- base controls. Figure
important to set the vernier control to
the amplitude of the signal of interest. 8 shows the panel layout of a50 MHz
the calibrated position.
This causes the beam to be deflected bandwidth amplifier. From this it may
To obtain a static display (where
in the vertical direction whilst it is be seen that the vertica: sensitivity is
each cycle of aperiodic signal overlays
swept across the screen. The result is selected by a switch and that the y
the previous one) the time- base must
the graphical display of signal amplifier has a 'vernier' sensitivity
be synchronized with some point of
amplitude versus time. control which must be at the
the input signal. That is, the time- base
Again an amplifier is needed to 'calibrate' position when measure-
is caused to begin its sweep across at
increase the signal level so that auseful ments of signal amplitude are being
the same point on the waveform being
vertical deflection results. Such an made.
viewed. The circuit which does this is
amplifier must be able to amplify the The position of the trace on the
called the triggering circuit. Triggering
incoming signal without distortion to screen depends upon the standing
can be taken from either an internal or
provide vertical sensitivities up to voltage applied to the plates. On both
an external source. When switched to
10 mV/centimetre ( typically the most Y and X axes extra controls, ( as shown
'internal' it is possible to vary the
sensitive range of educational units), in Fig. 6) enable vertical and hori-
voltage level of the signal operating the
or maybe as high as 10 µV per zontal shift of the trace position by
trigger. Thus the sweep may be
division ( in sensitive oscilloscopes). adjustment of the bias applied. When
adjusted to commence at a chosen
The insensitive end of the range will using the CRO to probe circuits
point on the waveshape. An 'auto'
control position provides automatic usually be around 50 V/cm but special involving ac signals combined with
selection of the voltage level for most units ( for electrical supply authority standing dc levels — as is the case in ac
use) provide for much higher levels. amplifiers for example — the dc level
reliable triggering.
105
 ELECTRONICS -it's easy!
on the Y signal causes the trace of the
some applications the CRO must be
ac signal to be displaced vertically and, the phase shift between two signals
matched to reduce reflections — in
perhaps, to go right off the viewing across a filter stage. Single- beam
such cases the input might be 50 St or
area. This difficulty is overcome if you oscilloscopes are very limited because
600 S 2. For fast rise- time studies it is
-
couple the circuit signal to the Y- they cannot provide as much
necessary to ensure that the capacitive information to the user as a unit that
amplifier via a capacitor. The ac signal value presented does not reduce the
then centres itself on the screen at the can compare the waveforms at two
overall bandwidth by shunting the
position chosen by setting the vertical points simultaneously. Three distinct
device to which the CRO is connected. alternatives are available to provide
shift control. This method is
In exacting cases, needing high input dual beam operation:
acceptable provided frequencies below
impedance and small capacitance,
the cut-off of the RC filter produced
special probes are used. These are Separate gun: These use two,
are not wanted. Measurement of very-
described later.
low frequency to dc signals must be dc physically-separate, electron beams
coupled on the ac/dc selector switch Calibration of the Time Base and and deflection systems that are
provided. A further switch position Y- Amplifier: The value of electronic mounted inside the tube envelope. The
enables the input to the plates to be components may drift with time, beams may be generated by splitting
brought to its dc zero position. This altering the sweep rate and vertical the beam from a single gun. These are
helps the operator to establish where amplifier values from those indicated generally referred to as dual- beam
this level is on the screen. The switch by the selector switch. To enable the units ( dual-trace is a term reserved for
for this function is in the middle of operator to check these, more the next method described).
the unit shown in Fig. 6. advanced oscilloscopes incorporate a Each beam has its own Y- input
special circuit that provides a fixed- panel with a complete set of controls
frequency, fixed- amplitude square as described earlier. Dual- beam units
Signal Input Connections: Oscillo-
wave signal for calibration purposes. A drive both X- scans with acommon set
scopes for use with frequencies below
typical signal would be 1volt peak-to- set of deflection plates ( as in Fig. 9)
about 1MHz can make use of separate
plug-in/screw-down banana- plug peak. As it is derived from the mains 5ut some ( rather rare) oscilloscopes
terminals. More usually, however, the frequency ( 50 Hz or 20 ms period) its enable each time- base to scan at a
time duration is also quite accurate. different rate.
input to the Y amplifier, and perhaps
In general, dual- beam units are less
to the external trigger, will use
standard BNC connections. These are MULTIPLE TRACE common because of the higher
expense compared with the next
designed for use with coaxial cable and OSCILLOSCOPES
method.
coax should be used for all except the
Measurement situations involving
shortest end connections to the
oscilloscopes more often than not Electronic switching — chopped mode:
circuit. The input impedance require display of comparative
characteristics are usually quoted — The deflection response of an electron
information between two points in a beam is rapid enough to allow it to be
1 megohm with 20-100 pF shunting
system — the relative input and output directed from one position to another
capacitance being typical values. In
signals in an amplifier response test, or at aspeed exceeding the scan rates used
with the signal being viewed. Fast
electronic switches are used to switch
eN.J the common single beam between two
(or three or four) Y-inputs. Figure 10
YA illustrates this. Appropriate blanking
EAT 0-0
(that is reduced Z intensity) is applied
MAINSCSO
YB when needed, when the beam is
chopping from one trace to the other.
If the chopping rate is chosen to be at
r•J least 100 times faster than the highest
frequency to be viewed the two traces
appear as separate traces. Hence the
Fig. 9. Schematic of Philips 3232 dual beam oscilloscope. Common x plates provide scan for name "dual- trace" for this method. In
both beams, separate y plates deflect the two distinctly separate electron beams that are derived reality the traces are not continuous
from a common gun.
but are made up of dash-spaces. A
hundred dashes across a screen
produces a virtually continuous trace
to the eye. The limit of usefulness is
reached when the inbuilt chopping
rate comes close to the upper
frequency being viewed thus
producing a dashed- line trace in which
the dashes are of length equal to
wanted signal features. At this point
information is lost.
As far as the user is concerned there
are still two groups of Y controls —
Fig. 10. Electronic-switching enables asingle-beam and deflection system to provide dual- trace
operation. just the same as for a dual- beam
arrangement. The difference arises as
106
 another two- input unit that is for
,Mplemout.

POI.TTITErIe
single trace operation. This is the
differential input amplifier unit; it is
AMPLIPER
.0 1 • normally provided as an optional plug-
VOL ' SLTOL
in.
Two inputs are amplified by the
• high-gain differential arrangement of
a dc amplifier — refer to Part II for
CII2 POLARITY
INPUT SII ,;rnR (
01 the basic concept. These are used
K EW
when common- mode noise rejection is
W MUTT
needed and when the difference

0 v -▪▪-Are
Tmr,
between two fully floating inputs must
be studied. Figure 12 is the panel of a
high-gain differential amplifier.
CH2 (' Position and gain controls are used as
---,
e * , t for an asymmetric input Y- amplifier.
' The input selector provides choice of
0.es•
7A is ah.CUAL TRACE AMPLIFIER
Channel A — channel B, A- B; each of
Fig. 12. Differential amplifier these being ac or dc connected. The
Fig. 11. Example of dual- plug-in panel. balance controls enable the two
trace plug-in showing the
basic controls. amplifiers to be trimmed to remove
circumstances where it is desirable to offset and gain differences.
the position chosen on the selector
switch where a 'chop' mode must be trigger from the channel 1signal.
selected — see the panel of adual-trace The electronic-switching method FINDING THE TRACE
plug-in shown in Fig. 11. enables more than two traces to be Even experts can experience
Chopped operation ensures that the displayed— three and four- trace units temporary difficulty when confronted
time relationship between the two are available. with an unfamiliar oscilloscope —
signals is faithfully presented: phase especially when it is complicated.
measurements are also accurate ( that DIFFERENTIAL AMPLIFIERS Naturally it takes training to get the
is, providing the input amplifiers to Generally the dual- trace oscillo- best from a unit. A basic difficulty is
each are identical). scope is recognised by two sets of often finding the trace! These steps
Chopped operation will also display input terminals. There is, however, provide an efficient procedure that
two simultaneous, non- recurrent
signals, such as transients induced at
various points when a complex
resonant system is excited by an
impulse. It is quite suitable for low-
frequency signals but less convenient
as the frequency rises.

ELECTRONIC SWITCHING-
ALTERNATE MODE
Switching can also be employed on
a full alternate trace- by- trace basis.
The first trace is a scan of channel 1,
the next of channel 2 and so on. This
does not suffer from the dotted defect
with high-frequency viewing but it
suffers from another deficiency in that
the phase relationship between the
two signals may not necessarily be as
indicated on the screen.
The method is unusable for obser-
vation of " once- only" dual events
because the second transient signal
may have gone to zero by the end of
the trace of the first simultaneous
transient signal. The panel shown in
Fig. 11 is typical of dual trace units.
The selector switch enables choice of
alternate, chop, channel 1, channel 2,
and channel 1plus channel 2 modes.
With two channel operation it is
necessary to decide which input will
synchronize the time- base scan. A
switch provides the choice of
appropriate internal triggering.
Fig. 13. Front panel of an oscilloscope that represents medium sophistication. The text contains
Although only channel 2, for example,
a self test about its controls.
may be being viewed there are
107
 should be learned. Begin with the BASIC MEASUREMENTS USING AN The current passing through induces a
input to the Y-
amplifier unconnected. OSCILLOSCOPE voltage that feeds the CRO. For ' ac
only' measurements a transformer
device suffices but for dc measure-
Voltage amplitudes of a waveform: ments it is necessary to use something
1. Ensure that the power is on. The
The amplitude vernier control is set to
on- off switch control is usually more sophisticated, such as a Hall
the largest position. The range selector
built in with the intensity knob, effect probe. All probes have the
is set to obtain the largest practical
but not always. conversion constant clearly marked on
vertical size of the waveform. Set the them.
waveform with the lower value on a
2. Turn the intensity to 75 per cent convenient graticule line using the
clockwise. vertical position control. Read off the Waveshape: Controls are adjusted to
number of graticule divisions over- obtain the largest possible waveshape
3. Switch the time-base ( horizontal) lapped by the waveshape, to the nearest on both time and amplitude scales. It
estimated tenth. Multiply the number is sometimes easier, however, to use
to a medium speed — say 1ms/cm.
by the range setting value to get the a slower time base value to obtain
This ensures that the trace displays
peak to peak voltage. Example . . . more stable triggering. The shape is
a full line across the screen rather
7.8 divisions for range setting of then transferred to a hard copy form
than apoint which occurs when the
20 mV/cm = 156 mV pp amplitude. by use of a camera, by longhand,
scan speed is on the slow ranges.
copying using the gratioule lines as in-
dicators, or by tracing the waveshape
4. Switch to auto triggering. This may Voltage level of a dc signal: Connect directly on the screen.
be a marked position of the trigger the two probe leads together. Set the
control or a separate switch. This horizontal scanned line, created with
ensures that the trace is being a suitable time base setting, to alower Frequency: Again the waveshape is set
triggered. graticule line using the vertical to have the largest size with stable
position control. Set to calibrate triggering. The horizontal distance
5. Switch to internal triggering. This is position, as above. ( Advanced units
from peak to adjacent peak is a
necessary for ( 4) above. Relying on have a control that shorts the probes measure of period. Frequency is cal-
an external signal to trigger the scan internally). Attach the probes to the culated from this by dividing the
is unreliable — it may not be of circuit. The dc level is given by the period into unity.
adequate magnitude. number of graticule divisions above or
below the line used as the zero defined Measurements made with the CRO
by the first step.
6. With this done slowly vary the are only of low order. The linearity of
vertical position control about its To read the dc level of an ac signal scanning and the estimation of traces,
mid range point widening out to get superimposed on a dc signal, carry out with the finite width trace, limit
the trace on screen. a similar process but use the centre of reading accuracy to around 5-10%.
the ac waveform as the vertical The CRO is not the tool to use for
7. The above may still not produce the reference point. accurate measurements.
trace. If not put the vertical position
in its middle point and the gain at
Currents: As they stand, few CRO
an insensitive value and begin ascan
units can measure currents. They are
of the x- position control. This
built to measure voltages so some
should be somewhere mid range.
conversion method is needed. Two
Too much x-shift can cause the
methods are:— SELF TEST
trace to slide off screen.
If the above facilities are under-
Complicated oscilloscopes will (a) Addition of a small value series stood it should be a straightforward
invariably incorporate a variety of resistor. The circuit is broken and a matter to master any oscilloscope with
controls that may also need adjust- resistor added in series. The voltage the degree of sophistication that has
ment to find the trace. Space prevents across this gives the current using been introduced so far. Figure 13 is
a full guide to spot finding. Fortun- Ohms Law. It is, however, not always the front view of a modern unit. It is
ately the more expensive units often possible to use this method because an given as an exercise.
provide a spot- finder button. Press it oscilloscope is usually grounded at one
and the spot appears on screen terminal meaning that one end of the
enabling the controls to be adjusted resistor must also be at earth potential 1 Locate the controls that are
accordingly to bring it back from the to preserve proper operation. If that is associated with the quality of the
direction it flies to when the button not possible then a differential input dot produced by the beam.
is released. CRO is needed.
2. Does it provide two trace display,
When the trace is located in mid An alternative, but not very accur-
and if so by what method?
screen the intensity and focus are then ate method is to probe the voltage at
adjusted by switching the scan to the each end of the resistor taking the
3. Which is the time- base control area?
slowest rate to produce a spot. These difference of the two voltages
obtained.
should be adjusted to produce a small
4. Where is a Y- amplifier vernier
round spot without halo: stationary
control to be found?
spots on screens should be avoided as (b) Current probe. Special probes are
this shortens the life of the phosphor available that clip over the circuit line 5. What are the input terminating
at that point. without the need to break the line. conditions?

108
 Oscilloscopes-

34
the refinements

needs. Until the advent of the permanent record for reports which no
MANY MEASUREMENTS IN other storage system can provide, and
Polaroid- Land process this involved a
electronics can be handled by the the price of a camera is not as great as
time-consuming development process
relatively unsophisticated oscilloscopes
before the operator was certain of the extra cost of the variable —
described in the last part of this series.
having even recorded the trace. Most persistence storage units to be
More capability can be provided at
oscilloscope makers now offer discussed later.
greater cost and this can be valuable if
specially built trace- recording cameras
the user understands how to make the
that fasten onto the large bezel STORAGE OSCILLOSCOPE
most of it. This part describes refine-
surrounding the screen.
ments that will be encountered in Most of the objections to the above
Such cameras use a Polaroid- Land
more advanced oscilloscopes. storage methods, with the exception
film pack of some kind and often
of permanent photographic repro-
incorporate a 35 mm roll film facility
IMAGE STORAGE duction, are overcome by using an
also. A Dumont unit is shown in Fig.1.
Screen persistence: Repetitive signals, advanced form of the basic CRO tube.
The user sets the CRO controls until
It is called a variable persistence
such as a sinewave signal, can be made satisfied that the trace will be as
storage tube and is a development of
to repeat on the screen overlapping the needed. This is done using the viewing
early 1950's more basic storage tubes
previous trace produced. If the time- aperture which reflects the screen
in which the waveform could only be
base frequency is sufficiently high — image to the observer via amirror. It is
from thirty or forty hertz upward — held at a constant intensity ( without
essential that the camera has the
the feature of gradual fade out). In
the screen provides an apparently correct focal distance set for the CRO
fact variable persistence is afeature of
stationary signal of constant and concerned, so in general cameras relate
tube operating circuitry not the tube
adequate intensity. This is primarily to specific units. Some models incor-
itself.
because the eye cannot detect individ- porate adjustable object- image ratios;
ual scans ( as in motion pictures and a few are fixed ratio. With experience The construction of a typical
television) and secondly because the it is even possible to capture multiple storage tube is given in Fig.2. The
phosphor, at frequencies above a few trace events ( by multiple exposure) for phosphor viewing screen ( having 0.1 s
hundred hertz, re- energizes before its comparison purposes. persistence time from P31 material)
light emission due to the previous A considerable amount of film and and the writing electron gun shown are
scan, has decayed away. patience can be consumed trying to similar to those used in the simple
Phosphors with large time-constants record once-only events. Cameras can cathode ray tube. Additional com-
are available ( such as P2, which takes be quite expensive — several hundred ponents are the flooding electron gun
one second to reduce to 10% of dollars — but they do provide a system, astorage mesh which is coated
original brightness and P7 which takes
three seconds) and oscilloscopes have
been manufactured which use these to
enable signals of less than one hertz to
be studied. This feature, however,
largely restricts the use of the instru-
ment to low frequency work because
medium and high-frequency signals
that are not well synchronised will
produce separate traces which remain
and add up with time to produce an
unclear picture. This method of study-
ing slow-transient phenomena has not
been developed to any great degree
because of this and other factors ( such
as poor resistance to burning). In
addition the retained- image times are
still inadequate for many applications.

CAMERAS
Storage requirements fall into two
classes — those where the transient is
unique and therefore needs to be
recorded only long enough to allow
the trace to be studied and those
where apermanent record is needed.
Fig. 1. Recording camera using Polaroid film pack.
The oscilloscope fulfils both these
109
 ELECTRONICS it's easy! on to hit the phosphor.
The collector mesh is provided to
STORAGE
help accelerate the flood electrons; to
MESH repel the positive ions generated by
PHOSPHOR the flood guns; ( which would other-
\ / 0(
wise write the whole screen bright)
4
1

and to absorb the emitted secondary-

emission electrons produced whilst
a writing is in operation. It is not
possible to store the trace in the view
mode for as long as in the store mode:
one to ten minutes of viewing time is
Fig. 2. Schematic typical for various makers' designs.
construction of
Erasure is done by applying a large
storage cathode
WRITING GUN ra y tube. positive voltage to the storage mesh
FLOOD AND
which charges capacitively to the same
STORAGE SYSTEM
value. The mesh voltage is then
brought back to a small positive value
whereupon the flood guns reduce the
voltage to zero. A small sudden
with a non- conducting, highly- resistive negative excursion is finally applied to
be held for at least an hour ( one maker
material such as magnesium fluoride, the mesh making it ready to write.
offers four hours) in a reduced
and a collector mesh which is held at a (This procedure is automatically
intensity mode.
positive potential. initiated at the single action of a
Io make the trace visible, low switch.)
To store a trace the writing gun is velocity electrons are sprayed by the
scanned over the storage surface. Variable persistence is incorporated
flood guns onto the entire mesh
Where the beam strikes the storage by changing the time taken to erase
surface. These electrons are allowed to
mesh electrons are knocked loose the picture. In the Hewlett-Packard
pass through to the phosphor in
leaving a positive- charge pattern. The unit, shown in Fig.3, this is achieved
proportion to the amount of positive
high- resistivity of the surface prevents by using a variable-width pulse
charge at each aperture of the storage
the charges moving toward a neutral generator that applies erase voltage
mesh. The positive field pulls many
state: the scan is thus stored — and can pulses to the storage mesh. The
electrons through causing them to pass positive- ions created by the flood- guns
limit this mode to a maximum of 10
minutes persistence.
Storage oscilloscopes can be used as
conventional units by applying about
30 volts to the storage and collector
144-1
meshes. Long persistence has many
virtues — it enables successive traces
resulting from adjustments to asystem
response to be overlaid together for
comparison purposes. It also allows us
to see very low- frequency scans, and
to plot scans of spectrum analysers.
Long persistence also finds use in time-
domain reflectometry where the time
between send and receive pulse needs
measuring.
By stacking sweeps on top of each
other a long persistence time can be
used to integrate or average a set of
traces. Variable- persistence storage
oscilloscopes are extremely versatile
4.111111111.111M11.1111 jp
but the high price restricts their use to
large laboratory groups.

° STORAGE USING DIGITAL

fee? • 9
MEMORY
Figure 4 shows a unit marketed

teeli, el -
around 1972. The transient recorder

to
unit accepts the analogue signal, con-
verts it to adigital equivalent with res-

lle IáÏn
pect to time and stores the values in

te
digital registers. Readout can be ob-
tained by using digital- to-analogue con-
mom en Fig. 3. Hewlett

III nUNI Packard Model

184A variable per-
version of the stored increments which
are scanned sequentially, the resultant
sistence and storage
analogue voltage being fed to an
be
t
oscilloscope. Con-
trols for storage are oscilloscope or chart recorder. Digital
print-out is taken direct from the
Maw on the left-hand
side of the screen. scanned store locations.
110
 8

9
2

A. SIGNAL BEING SAMPLED

6 78

Idee
,

2 9

8. PRESENTATION OR CRT

Fig. 5. The sampling oscilloscope builds up the waveform on the screen from sampled
values taken from the original.

C.)

Fig. 4. Storage can also be obtained using a

digital memory to capture the event which
can be displayed at will on an oscilloscope
Fig. 6. 1GHz dual-
channel sampling
plug-in. (Hewlett
o
Packard Model SUM
or recorder ( Datalab DL 905).
1810A). The con-
trols are designed to
provide operation as
1
<:›
-

for those on normal

This method is less common than real-time oscillo-
the storage oscilloscope alternative but scopes. With such a
unit it is possible to

COO
the ever- reducing cost of digital
view nanosecond
methods may put this technique into a rise time signals of
competitive price region. repetitive nature.
Another method of capturing
difficult to see, once-only transient
signals, and very slowly changing wave- deflect the beam at such speeds. The applications for the sampling unit is
forms is to record the level of the sampling oscilloscope offers asolution effectively smoothing the unknown
signal, increment by increment, as the to these problems. true original signal). Sample and hold
signal occurs, using a digital memory. The sampling oscilloscope makes methods were discussed in the
The concept is simple and the method use of the stroboscope concept to look previous part discussing D- A and A- D
offers certain advantages. These at a waveform, which must therefore conversion.
include ability to speed up or slow be repetitive ( as shown in Fig.5). The In practice a sampling oscilloscope
down the timescale of the original beam is set to illuminate the screen at is a normal high quality scope which
event, ease of providing a permanent point 1 in the diagram, waiting there can accept asampling plug-in. Figure 6
numberical printout and the facility until the next cycle where it moves to is the panel of adual sampling unit.
to process the signal before display. point 2 — and so on. The trace there-
fore gradually works its way through
DELAY FACILITIES
SAMPLING OSCILLOSCOPES the complete cyclic waveform and
because the scan speed is slower than Often one needs to study a certain
How to capture a very fast repet-
with a conventional sweep system the part of a repetitive - waveform — the
itive event, say near to the GHz region
cathode-ray tube system can operate very beginning, for instance. An
where scan times of 0.1 os/division are
with a lower bandwidth than the example is the ringing of a non- ideal
needed, is a problem because the
signal. The waveform produced is an square wave shown in Fig.7a. The
electron beam cannot transfer enough
average of many so the display is not trace is triggered, to begin the sweep,
energy into the phosphor to obtain a
only sharper but more uniform. ( This by a fast-going edge. Due to circuit
useable trace brilliance. Further it
may be a disadvantage in some response-times, the trace does not
becomes increasingly difficult to

TRACE BEGINS HERE

(a)
IaI (b)
o

Fig. 7. Inherent LOST TRACE DUE

trigger delay, if not TO DELAY
compensated for,
will lose the leading
OSCILLOSCOPE SCREEN
edge of a waveform.
111
 ELECTRONICS it's easy!

SPIKE OF
triggering circuits and this is often
INTEREST provided within the circuits. A slightly
better method is to provide an ad-
justable delay control on the trigger
panel.
(al A more difficult problem is cap-
o (a)
turing a point on the signal train that
is remote from the triggering transient.
Consider the signal shown in Fig.8(a),
Fig. 9. Trace brightening is used to show
where the problem is to investigate the which part of the waveform is to be
spike transient on the pedestals of the expanded. In this display the expanded
square wave. Triggering is best portion is also displayed on the second
achieved by using the edge ( a). But trace of the CRO.

ZERO TIME DATUM

this means that scale expansion puts
FOR TRACE the spike off scale when the horizontal
expansion scale is great enough to
S,, 0, Stdsré, res,,c.ro
provide information about the spike
structure. II Ii

Variable delayed sweep is the

answer. The trigger circuit is set by the
(a) edge but trace scan does not begin
until after a period, as in 8(b). Thus
(b) the trace captures the spike at the left-
o
hand side of the screen and scale
expansion will now be possible as in
8(c).
(al
To make this workable in practice
the operator must know just where
OSCILLOSCOPE
SCREEN triggering occurs for there may be
several somewhat similar events along
the trace. It is vital to know which one
is being viewed. A refinement pro-
vided in variable delay circuits is to
brighten the original display from the
SPIKE SHOWN TO BE
RINGING TRANSIENT
point where triggering will begin.
Taking the idea one step further leads
to a second delay that effectively
decides where the trace stops
o Figure 9 shows the waveform bright-
(c)
ened to show the portion that will be (b)
expanded and the second trace of the
dual- beam unit is used to show the Fig. 10. Use of dual delayed triggering point.
expanded part. Another useful feature (al original (14 expanded.

A SWEEP • • B SWEEP r •,'

Time/cm Time/cm
Fig. 8. Use of introduced delay in triggering rn se<
,
0 Gate
Length Out Length se
to enable an event away from trigger

e
20
transient to be investigated.
fal Original spike on pedestal of square wave. •
Sweep
(b) Delay introduced to bring spike back to
time origin.
(c) Scale expanded to reveal true nature of ec 0,
spike. an Delay
Display D

A-
Àl0 Off
, • r77Position
Hor
tal AmpliVer

x10
Avar
e.
begin to sweep at exactly that time
T
but begins a little later. The result is .....

loss of the leading edge region of the

dip
wave as shown in Fici.7b. The follow- Trigger
ing waveform may provide the infor- DC AC Level
dr
i ill gi
c AC
Slope
mation sought but attempts to widen Slope Level
+/rage —
the waveform in the horizontal
Est
direction lead to the second front p
OPP Sou rce
Est
Linei
disappearing. The simplest solution to
mr,

Ent
this problem is to incorporate an
Fig. 11. Marconi L
ine Est
TM 9220, TM 9221
appropriate fixed delay into the sweep units. 4tie TM 9220
'NIS 61)

libe TM 9221
112
Fig. 12. Most oscilloscopes have this input Fig. 13. Chart for obtaining reactance of
equivalent circuit. Although the values capacitors at various frequencies of
seem insignificant, at high frequencies they operation.
become dominant requiring the use of
special probes.

is to be able to use atrigger point not Probes with division ratio of 100:1
on the origin of the first trace set up — also are manufactured — these can
as in Fig.10. Here a marker dot is provide equivalent termination con-
provided to help the operator. Figure ditions of 5K/0.7 pF, 10M/1.8 pF,
11 shows the panels of Marconi plug- 1M/1 pF. The reason for different
ins which provide these and other pair combinations arises from the
variable delay features. need to alter the trade-offs between
rise time and signal loss in high-
frequency and very fast transient
PROBES
measurements.
Passive probes for voltage measure-
There is no easy answer to the
ment: In part 32 the importance of question of which attenuator probe to
providing the right matching con-
use. These guides are the start. For
ditions between two electronic
systems was stressed. This is also
important when connecting an oscil-
loscope to a circuit, for each output
and input has certain resistive and
reactive conditions which must be
LOSSY COAXIAL
properly combined to get realistic
CABLE 1M
signal transfer.
The oscilloscope can be represented PROBE
CONNECTIONS
as an ideal termination shunted by a
B.N.C.
large R and an adequately small C CONNECTOR
(a)
value — or at least they appear this
way at first sight. Figure 12 is the PROBE OSCILLOSCOPE
CABLE
most common approximate equivalent
circuit. ( Others used include 50 ohms
with negligible reactance in certain
applications). Referring to the chart
in Fig. 13, it can be seen that with
20 pF at 10 mHz the circuit being
9M —
measured must have an equivalent
output resistance of no more than
8 ohms! (b)
For high frequencies, those above
100 kHz say, we therefore need a CORRECTION
BOX
better connection method. To further
compound the problem the oscillo-
scope input leads can easily increase
the equivalent C value to 100 pF —
leads for 1:1 connection must there-
fore be carefully designed to ensure
known loading conditions which can
be allowed for in signal measurement
corrections. It is very bad practice to
use any piece of coaxial cable and
connector for frequencies beyond
100 kHz.
The first improvement is to use a (c)
probe which has 10:1 attenuation
built in, for these are designed to have
a lower effective cable capacitance —
see Fig.14(a). Still better is a special
correction arrangement that balances
the shunt against series capacitance to
provide a wider bandwidth — see Fig. 14. Schematic of atten-
uator probes (a) simplest
Fig.14(b). By the use of inductive concept (b) capacitive load-
tuning afurther improvement in band- ing to extend bandwidth
width can be obtained — Fig.14(c). (c) inductive.
113
 ELECTRONICS-it's easy!

amplitude measurements select a

minimum- impedance source point to
measure from. The best probe to use
here is one with the highest impedance
at the frequency of interest. Capac-
itance is less important here than
resistance for it alters edge shapes, not
amplitude.
For fast risetime measurements
again select a low impedance source
point and use a probe with lowest
effective capacitance — signal atten-
uation is less important than transient
edge shape changes.

ACTIVE PROBES FOR VOLTAGE

MEASUREMENT
The above probes make use of
passive matching arrangements. But
for the extremes of frequency and/or
risetime measurements the values of
components required in passive probes Fig. 15. This FET probe uses active coupling to overcome oscilloscope connection problems.

become impractical. However active

amplifiers interposed between the
circuit and the oscilloscope can be situations where measurements across by the current flowing in the wire into
used to improve performance by high source impedances are needed. a Hall effect transducer which
increasing input resistance and lower-
generates a voltage equivalent to the
ing capacitance ( short loads). FET OTHER PROBÉS
current flowing. These will also
probes are marketed to meet this. Voltage measurements are by far measure ac currents. The maker
Figure 15 is a Phillips PM 9354 FET the most frequent measurements made specifies the conversion constant —
probe which can be used for dc to but in some instances it may not be typically 1mV/mA. AC only, current
1 GHz measurements. These need an possible to determine voltages, and probes are also made using acurrent-
additional power supply to operate, current measurement is used instead. transformer principle.
and problems of dc drift ( 0.5mVPC) An example is the current flowing in a Probes for use in digital circuits are
and added amplifier noise direct-coupled Darlington pair con- also available. These may incorporate
(60 pV — 1.5 mV) may be disad- figuration where no significant resist- a logic gate that combines the outputs
vantages in certain applications. As ance exists over which avoltage can be from up to 6 circuit points as shown
well as being the choice for very high- developed. DC current probes ( see in Fig.17. Power for the gate is
frequency work, FET probes also find Fig.16) clip over the wire in question obtained from the circuit under test.
useful application in low-frequency coupling the dc magnetic field created
SPECIAL PLUG- INS
The oscilloscope, due to its
extensive flexibility, can form amajor
part of many test systems, thereby
reducing the overall price of advanced
measurement systems where asuitable
CRO already is available. Special plug-
Fig. 16. D.C. current probe. ins are offered (to suit certain main-
frames) that will convert an oscillo-
scope into aspectrum analyser or into
a semiconductor characteristic-curve
tracer. Another plug-in is offered that
converts the CRO into a four-trace
unit.
A basic need in manual measure-
ment is the provision of output form
that best suits the operator. In many
tasks a visual output in the form of a
picture or graph is better than having

Fig. 18. Display monitors can now provide

an exhaustive arrangement of visual output
forms. The next step is to provide this
capability for routine use in oscilloscopes
Fig. 17. Digital by making plug-ins available to go into
circuit logic probe. suitable main frames.
114
 extending the information rate at entirely to the oscilloscope.
to view many traces of a time General considerations are discus-
sequence taken over the whole system. which the operator can be informed
about asystem via aCRO. sed in " Test and measuring instru-
Already we have logic analysers which
The only weak link in present ments — 1974 Catalogue", ( Philips).
display space- plane information on the Tektronix, Hewlett-Packard, Dumont
systems ( as far as robustness, life and
CRO screen, multi- meter CRO units and Marconi outlets also provide basic
that write digital values on the screen cost is concerned) is the CRT itself for
it is just about the last remnant of articles on the selection and use of
and units that provide axes inform-
thermionic device technology remain- oscilloscopes.
ation on screen graphs. With the
ing in general use. This too will soon The principles of storage tubes are
reducing cost of advanced processing
be replaced by a solid-state equiva- explained in the Philips catalogue
it will not be long before the micro-
lent. Perhaps this will take the form of and in " Variable persistence increases
processor and memory ( already in use
a matrix of three-colour, LEDs in a oscilloscope's versatility", R.H. Kolar,
in very sophisticated units) are intro-
flat display — making maximum use of Electronics, November 29, 1965,
duced into quite moderately priced
the low-cost production advantages of 66-70.
oscilloscopes for converting the
LSI techniques. The subject of correct probe choice
information taken from the circuit
is quite extensive. The Philips Oscillo-
into better forms of display. Display
scope series referred to above has two
monitors are already available with REFERENCES
articles by PFVV Zwart on probes
.nany display forms — see Fig.18. Due to the versatility of the oscillo-
(Nos. 3 and 4 of series). V Bunze in
The next stage must be the marrying scope most books on electronic "Matching oscilloscope and probe for
of the basic CRO unit to such capa- instrumentation include basic descrip-
better measurements" Electronics,
bility via a wider range of sophisti- tions of how oscilloscopes work and
March 1, 1973, p 88-93 discussed
cated plug-ins. The colour oscillo- how to perform basic measurements
voltage measurements. •
scope will also soon be with us with them. Many books are devoted

4.10AS 0.10106 • 0.1060 *

.1.2ffle * kS11 . a feleill •
re. Idea •. 29,9, • •. 11443 •
edge» 4. 3/44. . 0.1..U1 •
tl, ire, 4 • 'So • 4 • 4w% ,
••6411.. .1.%•' • o• On J •
de ,...4 lea., e 4..444 •
4.440 11... , • 4•414* •
4 ......, a ..., , • •• i-e. .•
1.44de e 4.1 , . ... S.» •
I • 1e.ed S.44, .
i• iedie • . 934.1 • d.)62. •
&. 3.4. 4.14..s. • • . 211•3 •
1.Ni/el 11.8444 • 4. tIle •
i• 544a 11.44 , • aAPI, ••
t. 6ffll • • Ilfile .- 0.4214 •
0-1244 •
/eke./ 4 . 3, 34 • r•22,2 •
...ea» o , 3,4, ... • . R.I.3 T
2. XV» a.ffl_ 0.115Z .
2.1w •. 111%), •. 4.501

115
 Chart

35 recorders

IN GENERAL, chart recorders are plant. The earliest chart recorder was
designed to accept electrical voltage the two are so closely related that the
probably Lord Kelvin's 19th century
signals as these constitute the majority response depends on adjustment of
paper-tape siphon-recorder used to both disciplines of thought.
of signals produced by sensing equip-
record electric telegraph signals.
ment. Occasionally the chart recorder
Because of the large and varied Chart Recorder Formats:— Chart
is more appropriately connected to a
demand for chart recorders, manu- recorders are designed to display a
mechanical output without electrical facturers have developed numerous
signals being involved: in some circum- signal in a graphical form that is
alternatives. Figure 1 shows a number convenient to the user. There are
stances there is no need for electrical
of recorders installed to monitor an two basic types: those which record
circuitry.
oil rig.
Chart recorders are, therefore, one or more variables with respect to
In fundamental terms chart-
electronic system units which accept a time ( commonly called x-t recorders)
recorders are electro - mechanical con- and those which plot one variable
voltage signal converting it to an
verters — electrical signals are changed against the other ( x-
y recorders).
equivalent graphical representation on
into equivalent mechanical ones which Strip-chart:— In these recorders a
paper. The recorder can be put to use
are used to make a permanent record continuous roll of suitably scaled
in any application where an electrical
on a paper-chart. For this reason there paper is motor driven at constant
signal is produced. Examples are
are two aspects to a chart recorder — speed past the marking head. The
measurement of fluctuations of the
its mechanical design and its electrical paper drive is usually driven by a
power mains voltage, records of body
design. For convenience we look at synchronous or stepping motor as this
currents in medical diagnosis and
each more or less separately but in ensures accurate paper- speed. Where
changes in temperature in a process
designing and operating the recorder
mains supply is not available dc
governed- motors and clockwork
alternatives can be used. Chart speed
changes are commonly obtained by
altering gear ratios. Figure 2 shows the
construction of a typical panel
mounted strip-chart x-t recorder. The
module shown withdrawn from the
housing is the paper drive unit, the
housing contains the electronic amp-
lifier driving the pen which contacts
the top of the paper when the drive
unit is plugged in.
Strip chart recorders designed for
bench top use are also common —
Fig. 3. Some strip chart recorders take
up the used paper by rolling it or by
folding it in a concertina. The latter,
known as z-fold, is very convenient
when the need to refer to the record
arises. Chart speeds vary widely —
from metres per second in fast-writing
recorders used to capture kilohertz
bandwidth transients, down to milli-
metres per hour for industrial process
and slow- scientific phenomenon
recording. It is not usual, however, to
find a range as wide as this in the one
unit.
Process industry strip- chart
recorders generally run at one speed
only; units for scientific use usually
have switched speed capability. The
choice is decided by matchin9 the
resolution required with the amount
of paper consumed.
Fig. 1. Chart recorders are used in many varied applications. The panels of this control room
Paper sheet:— The flat-bed style lends
contain a number that are used by the operators to see how the process is behaving.
116 itself to x-y operations where the
axes are driven by two independent
Fig. 2. Typical strip-chart x- trecorder
variables. Examples are plotting the designed for industrial panel mounting.
properties of a material, as shown in (Record Electrical Co.).
Fig. 4, and charting antenna field
strength versus position. In this style
the recording paper is a single sheet Fig. 4. Plotting an hysteresis curve
which is attached to the platen. The for material under test in the large
magnet shown at the rear.
pen moves both in the x and y
directions. The paper may be held by
clips or by electrostatic attraction. If
the x axis input ( horizontal) is fed
with voltage that rises lirearly with
time ( a ramp function) the x axis will Fig. 3. Flat-bed strip-chart
move across the chart with time recorder (Philips).

117
 ELECTRONICS it's easy!
making the unit an x-t format re-
corder. Plug-ins generating appropriate
ramps are often provided as an acces-
sory — one is illustrated in Fig. 5.
Circular:— Where the Geometry of the
measurement task is circular, such as
recording out-of- roundness of a
ground shaft, or where the measure
has a cyclic time function, such as
daily temperature changes, a circular
form of chart is easier to use. The
chart rotates under the marking device
at a rotational velocity locked to the
geometrical position or the
appropriate sub-unit of time — hours,
days, weeks and months. An example
of a circular-chart recorder is given in
Fig. 6.
The size of chart papers varies
Fig. 5. Plug-in used to
convert x- y flat bed greatly from recorder to recorder.
recorder to x-tmode Strip charts are used from 50 mm
of operation. width to around 800 mm with lengths
as much as 150 m. The duration of the
.
-naximum record that can be taken on
Fig. 6. Circular chart. a roll is decided by the chart length
recorder. and the chart speed. Flat bed units
begin in paper size at about 200 by
300 mm ranging to huge computer-
controlled automatic-draughting units
with beds as much as 6 m x 4 m. Cir-
cular charts rarely exceed 300 mm
diameter.

Fig. 7. Kelvin's 1873 siphon recorder introduced the use of

continuous ink marking in chart recorders.

118
Pen Tuoe

(a) 1 Filter
2 Capillary
3 Permanent magnet
4 Pole pieces
5 Field winding of pole pieces
6 Connection (b)
Fig. 8 (a). Capillary action ink pens. for measuring signal
(b) Schematic of pressurized ink recorder 7 Pump
(Siemens).

Supply of chart papers can be a factor of the bore of the pen, paper
difficult at times because stockists absorbency and ink viscosity.
find difficulty in holding large stocks Figure 8a shows pen details.
of the numerous options available. It A second ink feed method uses a
combination of gravity feed and
is wise for the operator to hold a
generous supply in hand at all times. capillary action through small bores.
When reading values from paper These are the ballpoint and fibre-tip
charts care must be exercised in ensur- pens. A third ink method pressurizes
ing that inaccuracies caused by paper the ink, recording being performed by
size changes, paper wander across its a very fine ink jet. This method is
platen and marking mechanism offsets suitable for fast writing speeds ( as high
are allowed for. Good quality charts as 60 metres per second compared
are a necessity with high-quality with around 1m per second for un -
measurements. pressurized ink feeds). There is no

PAPER MARKING TECHNIQUES

In these units an electronic amp-
lifier coupled to a mechanical drive
moves a mechanical point across the
chart. It is then necessary to mark the
paper in order to show where the
point has travelled. Five commonly
used techniques will be encountered.
Ink pen-- Samuel Morse's telegraph
recorder shown in Part 5 ( Fig. 1) used
a pencil to mark the paper strip. A
limitation is that the lead wears away
making a feed mechanism necessary.
Ink can flow from a reservoir con-
tinuously: Kelvin introduced the
siphon system in 1873 — see Fig. 7.
This system is used extensively today
in one form or other. Ink feed rate is

Fig. 9. Recording multimeter using

pressure sensitive paper (Goerz).

119
 ELECTRONICS-it's easy!

Cooper bar

le !items

Peer &PO ree

cart leell frier

Pleasuring point selector

Fig. 11. Dotting recorders offer the advantages in

Fig. 10. UV recorders provide traces by exposure of photographic paper. Further slow-speed applications of being suitable for multi-
exposure a needed to bring the latent image into view. channel multiplexing.

mechanical contact with the paper in producing the mark by applying ahigh mechanisms: Fast writing speeds
pressurized systems, the fast writing voltage between the stylus and the require continuous marking and for
rate arising because of the very small platen.
these the writing mechanism functions
size of nozzle built into the deflecting Another method electroplates onto continuously. For very slow speed
system. Figure 8b shows the schematic the surface of paper made conductive needs, as are found in process plant
of such a recorder. The pressure is by saturation, with salts. It requires monitoring an alternative, in which a
automatically adjusted to suit the wet paper use but will operate with dot is produced on the paper at regular
chart speed set. lower voltage levels than the above periods, has certain advantages.
The correct choice of ink and paper carbon paper method. Figure 11 shows one form of mechan-
for the speed of operation is essential. Zinc oxide reduced to free zinc is ical arrangement. A separate motor, or
Water- based inks are to be avoided as the process used in another kind of pick-off from the chart drive causes a
the record can be destroyed by recording system. Metallized papers in point to periodically press on the
accident. Fast drymg inks are needed which the metal film is fused to its paper, marking it by the appropriate
or else the trace may be rolled- up paper backing are another. Yet method used. By incorporating a
before the ink is dry. In short, another is based on providing achange geneva mechanism ( one that rotates a
although the alternatives to ink offer in the paper surface which takes up shaft in steps) the input signal can be
certain advantages we are still forced toner ( similar to the Xerox process) — switched sequentially over a number
to use ink as the best all-round choice it is fine for very fast systems but not of different signal channels ( six and
in many applications. those that occur slowly. twelve are usual). Also synchronised
Pressure sensitive papers-- Black paper
to the channel changing action is an
treated with tiny wax beads appears Heat senstitive papers: Yet another
inking system that steps from colour
white until the beads are flattened to method of making the record is to use
to colour to provide a different
form a transparent cover window a heated stylus melting a wax- like
coloured dot for each channel. Inking
thereby exposing the black. Pressure coating on black paper. These papers
may be as shown ( different ribbons)
sensitive papers are marked by the can oe manufactured with greater
or may be provided as individual pads
action of a gentle pressure exerted by resistance to marking ( during
each soaked with ink. A multipoint
the stylus. The relatively high contact- handling) than the pressure sensitive
dotting head wipes through this ink.
force needed restricts these to slow papers. Stylus temperature can also be
One maker uses a multicolour single
response application. Pressure-sensitive varied with ease to suit the writing
ribbon, akin to atypewriter ribbon.
papers are more usually used with speed concerned.
Multi-channel operation is also
marking mecharisms that are Photographic paper: The earliest
provided in some continuous trace
periodically pressed against the paper photographic systems used negative
recorders. This is almost always
to form a dot. Figure 9 shows a re- film. Such systems are still in use
achieved by incorporating separate
cording multimeter which uses this today but the majority of the highest
recording heads for each signal.
latter method of marking. Another speed recorders ( 30 kHz is possible)
Figure 12 is afour pen recorder of the
limitation is that the record can be use ultraviolet light to expose specially
type in which the pens do not overlap:
marked during hand1Mg. treated paper. Exposure produces a
each trace is contained within a
Electro-
sensitive papers: Some re- latent ( invisible) image which needs
quarter of the full chart width. Multi-
corders use paper which is marked further exposure to form the visible
trace recorders in which each trace has
when an electric current is passed image. This is shown in Fig. 10: the
the full paper width capability are also
through it. The earliest was carbon fluorescent lamp intensifies the traces.
available. Mechanical drives have the
impregnated; dielectric breakdown Continuous versus dotting disadvantage in that the traces must be
120
slightly out of phase so that the pens
can pass one another without fouling.
Optical recorders do not suffer from
this drawback.

RECORDING MOVEMENTS
We now look at the methods used
to transduce the electrical input signal
into an equivalent mechanical
movement.
Moving coil mechanisms: Basically
these use modified moving coil and
pointer. The end of the pointer
carries an ink pen or acts as amarking
point when forced onto the chart
paper in dotting styles (see fig. 11).
Simple systems trace an arc across the
chart giving a non-linear record.
(curved markings on the paper over-
come this but complicate the platen
design). This can be linearized to
provide better accuracy by various
means such as that shown in Fig. 13.
Optical recorders also use amoving
coil unit on which amirror is mounted
to reflect a high intensity focussed
beam across the paper. These units
have their origin in practical oscillo-
graphs designed by Duddell ( to
Blondel's ideas) at the turn of the
century. The choice of galvanometer
unit largely decides the frequency
response. Today they are supplied as
robust plug-in units like that shown in
Fig. 14. The application, in many
units, decides which galvanometer is
Fig. 12. High-speed four pen recorder. In this style the pens do not cross over each other
used and the optimum terminating
limiting the trace width to aportion of the paper width.
resistance value in order to know the
deflection and sensitivity for a given
frequency of signal. ( Refer to reading
list for guides). These recorders offer
the ability to modulate the trace
intensity producing 2-D half-tone
chart records.
Potentiometric recorders: Around
1898 Professor Callender devised his
recording resistance pyrometer
(Fig. 15) and in doing so provided
instrumentation with the potentio-
metric or self- balancing recorder. This
method makes use of a closed- loop
system that causes the pointer to
follow input signals. Referring to
Fig. 16 the recorder has a drive motor
mechanism which translates the
pointer in one direction or the other
depending upon the polarity of the
signal driving the motor. Attached to
the shaft driving the pen is a rotary Fig. 13. Special linkages are used to
resistance balancing potentiometer, as linearize the non-uniform movement
shown in Fig. 16a. Schematically this produced by a moving coil pen drive.
can be shown as a linear equivalent
(the more recent design style used) as
shown in Fig. 16(b). The potentio-
meter wiper moves across in unison Fig. 14. Galvanometer unit for UV recorder
with the pen and generates achanging (Hathaway Instruments).
121
 ELECTRONICS-it's easy!
Balancing evienliomerer

Pecerchng caffiage

Fig. 16 (a). Arrangement

,9f motor form of potentio-
metric recorder.

MVI

(1)) Schematic using linear

motor.

Fig. 15. Callendar's original potentiometric recorder was devised

around 1898 to record furnace temperatures.

(c) Circuit schematic of potentiometric

method (simplified.

value signal. The potentiometric

The advantages of recorders such as by the amplifier gain than mechanical
system circuit layout is represented in those described above are that the
Fig. 16(c). A reference voltage is constants. The majority of flat-bed
mechanism plots a linear scale, and recorders use this principle: at full
supplied across the potentiometer.
there is considerable power available trace movement their writing speeds
Voltage from the wiper is compared
to move the pen against frictional can reach several metres per second.
with the input signal voltage to be forces. The system, being potentio-
recorded. If a difference exists this The method also overcomes the
metric, draws little current once the restriction on traverse length suffered
constitutes an error which causes the
unit has achieved balance and, as by rotationally driven recorder
drive motor to move accordingly to
considerable drive power is available mechanisms. Although a simple dc
correct the error. The input signal
under closed- loop control, the pen servo control is shown, potentio-
and reference signals are suitably
response can be made tighter than for metric recorders, especially those
attenuated to provide the sensitivity
the open- loop pointer-type moving built before around 1970 more usually
needed at full-scale deflection.
coil units. Sensitivity is decided more used ac control systems.
122
Fig. 17. Future style of chart recorder.

amplification because the majority in Fig. 17, will use a linear array of
CRT — Fibre Optic Recorders: A
of signals to be recorded, have in- LEDs to expose a spot on photo-
recent design concept couples a CRT
linear sweep trace to photosensitive sufficient power to provide an graphic paper in the appropriate place.
adequate response. Recorder sensi- This method would eliminate mech-
paper via an optical fibre connection.
tivities may be fixed in manufacture, anical manufacturing problems, have
This provides the highest response of
as in process industry dotting re- excellent response characteristics and
all chart recorders so far available — dc
corders, or have adjustable ranges. The be readily multiplexed to provide
to 1 MHz.
manufacturers of recorders usually multi-channel traces. Using LSI manu-
provide the amplifiers as part of the facturing methods, the cost of the
DYNAMIC RESPONSE recorder, the purchaser only has to array head and analog-to-digital
A point commonly overlooked is make the selection. converter would be minimal.
that chart recorders have a certain
Event-marking recorders: In many REFERENCES
dynamic response and are effectively
recording applications the variable Books containing chapters on chart
low-pass filters of the input signal.
remains constant for more of the time recorders include:
The response of a recorder to a sine
than it varies. An example might be "Basic industrial electronic
signal, that is, the recorded trace, will
recording rainfall in dry areas. If the controls" J. H. Ruiter and R. G.
look like the original but will lack
record must provide fine time- Murphy, Rolf Rinehart and Winston,
adequate amplitude if the pen cannot
resolution the chart must run fast 1962.
follow fast enough. When quoting
which means using immense lengths of "Measurement systems"
response rates it is therefore necessary
paper for little data recorded. An E. O. Doeblin, McGraw-Hill, 1966.
to state amplitude as well as fre-
approach, slowly finding acceptance, is "Principles of instrumentation",
quency. For example, moving-coil
to use atime/date printer which prints J. T. Miller, United Trade Press, 1968.
recorders with short pen arms — as in
a value each time an increment of A brief survey of recorders is given in:
Fig. 12 — have a typical response that
event occurs. Each increment print- "Instruments-electric recorders",
is flat from dc to 100 Hz at 10 mm
out causes the chart to advance aunit Siemens, September, 1968. Liquid jet
peak-to-peak deflection for asinewave.
The result is arecord chart completely oscillographs are discussed in detail in
If the frequency is increased the
filled with non- zero data. It is harder Siemen's pamphlet MS7/200e, 1967 of
recorder will still operate but the
to interpret but much more efficient that title.
amplitude of a sinewave record fallS
for spasmodic data situations. At A review of the merits of various
off. Plots of complex waveforms may
present, however, this form of equip- writing systems is given in:
be severely distorted for the funda-
ment is hard to procure commercially. "Graphic recorder writing systems",
mental may be recorded at full
amplitude with harmonics attenuated D. R. Davis and C. K. Michener,
progressively. A square-wave input
THE FUTURE Hewlett-Packard Jnl. October, 1968.
may be recorded as anear sine-wave if The design of recorders is decided Chart inaccuracies are discussed in
the response is inadequate. It is better largely by cost, reliability, sensitivity, "Recording charts", L. Briggs Dunn,
to use a smaller signal amplitude in and packaging to suit the application Instruments and Control Systems,
such cases. Response and accuracy cost money. July, 1969.
Simple moving-coil chopper- type The weakest points of inexpensive When using optical oscillographs
recorders will roll off from as low as recorders seem to be the reliability 01 the correct choice of galvanometer
the marking arrangement, and poor head and source impedance usually
1 Hz. Ink jet units extend to 800 Hz:
beyond that optical recorders are response. Optical recorders eliminate requires calculations to be made.
needed providing up to 1MHz in the marking problems but still ( with the (There is a trend toward elimination
CRT design. Frequencies above this exception of the CRT types) require of this by providing suitable amp-
must be viewed by oscilloscopes fine electro -
mechanical mechanisms to lifiers). Manufacturers usually provide
deflect the trace. We can confidently such detail. A paper " The Theory of
using cameras to record the image.
expect to see solid-state " deflection" recording galvanometers" by M. A.
Faithful response is also a function
of amplifier characteristics. With the systems marketed in the near future Le Gette, Consolidated Electrody-
exception of simple moving-coil which are based on semiconductor namics, Pasadena, California provides
technology. Units, like that depicted in depth detail.
recorders most units have built-in
123
 The control

36 of power

IN PART 10, WHEN DISCUSSING which they are mounted. A rectifier collector current. The heat loss is,
the types of amplifiers, we briefly
stage using flat- plate heatsinks is therefore, 100 W and this must be
mentioned the power stage found at shown in Fig. 1. These metal struct- liberated in order to keep the tran-
the output end of electronic systems. ures are needed to rapidly conduct sistor temperature lower than its
Typical devices requiring amplifiers away and dissipate to the air the heat recommended maximum value.
to drive them are loudspeakers, elec- generated at the junction of the All materials resist the conduction
tric motors, and heaters. device — this is a critical design re- of heat to some extent — this property
The power handling capability of quirement. The approach to designing is called 'thermal resistance' and its
the various designs of these special heatsinks is common to all power value depends upon the material
amplifiers can range from one watt to components. (copper is less resistive to heat flow
many kilowatts. In this final part we than iron) and the cross-sectional
introduce the special semiconductors All semiconductors used in ana- area ( increasing the area decreases
and techniques used in electronic logue control will have heat losses the resistance). In practice catalogues
power control. (the power lost as heat equals the for power components usually quote
current through the device multiplied
the thermal resistivity O ( which has
by the voltage drop across it) which units ° C/W ) between two points on
HEATSINKS will cause the junction temperature to
the device. For example, typical
rise above the case outer temperature. measured temperatures for a certain
As some power is lost as heat in For example, a transistor power amp- power transistor mounted on a heat-
power transistors they may usually be lifier stage may have at half output sink are as shown in Fig. 2. From
recognized by the large heatsinks on power ( say) 10 V drop and 10 amp these temperatures we can see that:—

(
SOURCE OF HEAT )

1JUNCTION

2CASE

3 HEATSINK
4 AIR

(
SINK OF HEAT )

THERMAL RESISTANCES
1 2 3 O IN °CM
4
Tj Tc TH TA
95°C 79.5 ° C 75 ° C 25 °

(20w (20w
GENERATED ) HEAT FLOW LIBERATED )
Fig. 1. Power handling stages are easily iden-
tified by the heatsink assemblies on which
they are mounted — a 35 A. 60 VDC rec- 2. Thermal resistances for a 20 W power transistor.
tifier stack is shown here.
124
TABLE 1

= ( 95 — 79.5)/20 = 0.77 ° C/W Thermal

OC -H = ( 79.5 — 75)/20 = 0.23 ° C/W Resistance 0 C- H in ° CAN

0 H- A = ( 75 — 25)/20 = 2.5 ° C/W

Material used between device with heat con-
0 J- A = ( 0 J- C 0 C- H + OH- A) = 3.5 ° C/W
and heat sink ( for insulation) Dry ducting grease
Where J = junction, C = case of device,
H = heatsink and A = air. Direct contact ( T03) 0.20 0.10
Teflon insulator shim ( T03) 1.45 0.80
Mica shim ( T03) 0.80 0.40
From this example we can see that Anodized aluminium ( T03) 0.40 0.35
the thermal resistance within the 0.25in stud mount ( direct) 0.40 0.25
device — the parameter the user has 0.50in stud mount ( direct) 0.12 0.07
no control over — is larger than the 0.75in stud mount ( direct) 0.07 0.04
case-to-the-heatsink value. This means
it is not worth improving the contact
and heatsink material. The important
thermal resistance is that between the Fig. 3. Table of thermal resistances Oc_H for typical mounting methods. Values can vary
junction and the air ( presumed to be widely.
at constant ambient value); in many
cases a different shape heatsink, one
ature remains less than a specified values. In switching designs the base to
that transfers heat better to the air
maximum limit. Beyond this quoted emitter junction voltage of a tran-
(finned for example) would make an
value the junction will be destroyed. sistor is significant.
improvement. The thermal resistivity
A practical difficulty is that the Step 2: Establish Tjmax, TAmax from
(heatsink to air) can also be reduced
junction temperature cannot be data sheets and expected ambient
by forcing air past the heatsink and/or
measured to ensure that the design is conditions. This enables the minimum
by increasing the heatsink surface
adequate so selection of mounting and required value of Tj_A to be cal-
area. The latter measure, however,
heatsink type must be made with care culated.
also has its limits because the thermal
using manufacturers' quoted thermal Step 3: Calculate the overall thermal
resistance between the device con-
resistance values as the basis of a resistivity needed from 0j- A
nection point and extremities of
larger plates rises with increasing design. The following steps are given TJ-A/Wmax.
as a guide but full detail should be Step 4: Establish Oic and BC- El from
dimensions ( reducing the effectiveness
sought from more detailed accounts — device table charts and the mount
of outer areas).
see reading list. thermal resistivity for the device
The above example illustrates how
a heatsink stage can be designed using Step 1: Assess the maximum power clamping method. Fig. 3 lists typical
(W max) to be dissipated by the O values for various clamping methods.
the concept of series thermal resist-
ances. In practice the design procedure device. This will be the worst case of Step 5: Calculate 0H_A required
must be worked in reverse. The aim is V.I product remembering to allow for OH- A = 0 J- A — ( OJ-C 0 C- H)
temperature effect's and maximum Step 6: Use heatsink tables to find
to ensure that the junction temper-

Fig. 4. Range of heatsinks for dissipating excess heat in semiconductors over arange from milliwatts to kilowatts.
125
 ELECTRONICS it's easy! for several hundred volts operation
and current levels exceeding a 1000 A.
Cut-off frequencies into the gigahertz
suitable design having OH_A value or Figure 4 shows a wide selection of region are available ( with less gain than
smaller. heatsinks including units for fluid that of lower frequencies). At RF
In general if OH_A needs to be less cooling applications. Fins should frequencies gains range from 4-13 dB
than 2 to 5 ° C/W the heatsink becomes always be positioned to assist the for powers in the range 0.1-80 W.
prohibitively bulky. Design of the vertical convective flow of air over There are few power applications that
whole system is usually limited by the the surfaces. Total immersion of the transistor devices cannot handle. In
manufacturer's value of 0J- c, which electronic circuit in cooling liquid is practice, however, certain other semi-
cannot be reduced. The interface not used. conductor devices are often a better
coefficient BC- F-1 is usually around
choice.
0.15-0.20 °C/W for direct contact
using the recommended heat conduct- POWER TRANSISTORS
ing silicon grease. Mica insulation SCRs, THYRISTORS AND TRIACS
Power transistors are little different
degrades this value a little, poor heat to small-signal devices in their basic Semiconductors and diodes have
conducting insulators should be semiconductor principle of operation: one p- n junction and transistors have
avoided as they contribute a quite the distinguishing factors are the two junctions, p- n- p or n- p- n. A logical
high value of 0C- H. heavy-duty design which enables high progression is the three-junction device,
Heatsinks for analogue control collector currents and voltages to be p- n- p-n. This family contains such
power units will need to be much controlled. The junction areas are devices as the silicon-controlled
larger than those of switching de- much larger and the case design is rectifier SCR, the silicon-controlled
signs such as the switching regulator made to keep the thermal resistivity switch SCS, the gate-turn-off switch
and normal rectifier stacks. This is as low as possible ( around 0.8 ° C/W) GTO, the light- activated, silicon-
because the latter need only dissipate in order that the losses can be controlled switch LASCS, and the
the V.I product of the two extremes removed. Collector currents being Shockley diode. Of these, the SCR
of V and I. The voltage drop across a higher and the gains being lower than (also called a thyristor) mainly con-
power diode running at many amperes small- power transistors means the base cerns us as it is able to control high-
is around one volt: when reverse currents are also large. Thus, high power levels (they were introduced in
biassed the voltage is high but the power stages have to have lesser power Part 16). The SCR has an anode and
current negligible. stages driving them. They are available cathode and a gate lead (which when
held positive prevents the unit from
conducting).
By controlling the gate voltage it is
possible to control when power begins
a‘dNee to flow during an ac cycle. Once the
SCR is triggered (or fired) it remains
on until the anode-cathode voltage
SCR
falls to zero again. SCRs are, there-
fore, extremely useful when an alter-
nating current source is available as
this automatically provides the
necessary switch-off conditions at each
CONTROLLED HALF- WAVE CONTROLLED HALF PLUS half cycle.
FIXED HALF- WAVE
TR IACS are special SCRs that can
be switched on to allow both positive
and negative half cycles to pass. This
action can also be arranged by using
two SCRs.
This class of device cannot control
the flow of dc power from adc source,
because once turned on they remain
on, acting like an adequately low-
resistance contact. They are, however,
CONTROLLED FULL WAVE CONTROLLED FULL WAVE invaluable for controlling loads which
can be energised by ac power —
heating coils, motors, lighting and
furnaces.
The operating circuitry for an SCR
BRIDGE RECTIFIER is designed to provide the appropriate
gate on-voltage level at the correct
time during the half cycle. Fig. 5
shows five basic forms of phase con-
trol. A typical trigger circuit is given in
Fig. 6. One difficulty in this kind of
control is that large line transients are
generated, along with RF interference,
when the power begins to flow during
CONTROLLED FULL WAVE FOR AC OR DC each cycle.
A more refined type of control
derives the required average output
Fie 5. Five arrangements by which a load can be fed with power flow controlled by SCR power as the mean of a series of
devices.
complete whole-cycles rather than in
126
the mean of many partial cycles. This
method generates substantially
reduced line transients and RF inter-
ference because switching always
occurs at the zero voltage condition:
Figure 7a shows one form of pro-
portional zero- voltage-switching con-
troller using a TRIAC to control the
heat produced in the element.
Figure 7b is a typical output signal
burst of gradually increasing power.
Capabilities of SCR devices range to
hundreds of amperes, reverse voltages
to as much as 2000 V. The maximum
voltage drop across the turned-on SCR
lies in the range 1.3-2.5 V, with leak-
age currents being in the region of Fig. 6. This typical phase-
40 mA in the turned-off state. control trigger circuit for
controlling an SCR uses
These characteristics may make a unijunction transistor.
SCR devices appear extremely robust.
Design of reliable, high- power, units,
significant advantage of very small
however, is a matter for a specialist. iably means carrying the correct spare
losses in the regulator stage. The tran-
Many pitfalls can occur if their ready to use.
operation is not understood in detail. sistor is either fully-on ( high current
but very low voltage) or fully-off
Designed properly they will, however, SWITCHING REGULATORS AND (highest voltage but minimal current).
give utmost reliability.
CONTROLLERS As well as reducing the losses the
Fuses for SCR circuitry also need
Parts 15 and 16 discussed methods method also can use asmaller capacity
special consideration because semi-
used to regulate dc power supply transistor. The price paid is the need
conductor junctions when overloaded
will blow more rapidly than simple output voltages or currents. For small for a filter stage and for a pulse
power levels — afew watts — the series generator to drive the switch.
wire fuses or electromagnetic circuit
regulator and zener diode arrange- Switching regulators are especially
breakers. The criterion is that the
1 2t rating of the SCR must be greater
ments are acceptable because the necessary when the voltage drop
than that of the fuse. 1 2 t values are
power they dissipate is not an between the source and the load re-
usually provided in maker's data sheets. economic factor. The controlling tran- quirements becomes large.
sistor ( as is shown diagrammatically in Modern designs often make use of
During the turn- on period of the SCR
this value may drop significantly. Fig. 8) can, instead, be used as a an integrated circuit as the basic
Selection of adequate protection fuses switch varying the on-to-off time ratio control unit adding an additional
is a matter that must be studied in (mark-space is the term used) in order switching transistor to cope with the
some depth. Care must be taken to to vary the average dc power obtained output current needed. Fig. 9 is a
after smoothing. high-current switching regulator which
mend blown fuses in SCR units with
the correct replacement -- this invar- The switching method has the can supply 3 A continuously at 30 V
input with losses sufficiently small to
allow the use of quite small heat-
sinks.
TO HEATER Switching is also a suitable method
ELEMENT
to efficiently control output loads —
the difference between this and
TRIAC regulator design is that the feedback
loop ( dotted in Fig. 8) is not used; the
mark-space ratio of the generator
being controlled instead by the input
AC MAINS
INPUT

CONTROLLED
CYCLES
ZERO-VOLTAGE
SWITCHING IC

(a) "
01--THERMISTOR

AVERAGED
POWER
OUTPUT

TIME --•»-

Fig. 7.
(a) Zero- voltage-switching temperature controller using a
zero voltage switching IC driving a TRIAC from a therm-
«AM istor sensor.
(b) Output signal with gradually increasing power.
UNIJUNCTION
127
 signal to be amplified. This principle is
SOLID STATE FILTER FOR
used in high-current dc motor control SWITCH 11 _Inn SMOOTHING
and in advanced forms of audio UNREGULATED
amplifier. ric. SUPPLY
>
TO LOAD
INVERTERS AND CONVERTERS (AS REGULATED
OR CONTROLLED
A converter, in the electrical power OUTPUT)
engineering sense, is a machine ( or a
VOLTAGE
circuit) that changes current from one INPUTIFOR n _
— J
kind to another, or from one FEEDBACK CONTROLLING
USE)
frequency to another. An inverter, in MARK-SPACE RATIO ( FOR
MARK-SPACE REGULATOR USE)
the same sense, is a machine that GENERATOR
specifically converts dc to ac — being (VOLTAGE- CONTROLLED)
Fig. 8. Schematic of switching
one kind of converter. Originally regulator or controller (see
rotating machines were used but today text).
the trend is to use static solid-state
equipment. Figure 10 shows a number of con- wave oscillator circuit causing flux
figurations used to produce ac power changes in the transformer which
There are many instances where
from adc supply. induce the higher output voltage
these are required — providing a240 V
ac 50 Hz supply when only 12 V bat- Switching produces square- wave needed for rectification back to dc
teries exist, providing a 200 V dc energy after inversion and in many current.
;upply from 12 V dc and to change instances this roughly square-wave AC to AC: Some mains equipment can
frequency such as where a 240 V ac output waveform is satisfactory. run on either 50 or 60 Hz frequency
Where the output must be sinusoidal with little change in performance.
50 Hz mains might be needed to drive
aircraft equipment operating at more complex circuitry is required to Occasionally, however, it is necessary
400 Hz. obtain an undistorted wave shape. to use the correct frequency specified.
The basic principles used in each When choosing a commercially made To change frequencies the simplest
are based on the technology discussed inverter it is important to verify if the procedure is to convert the original ac
output waveshape is suitable for the supply to a suitable dc value inverting
before in this part. These are now
summarised with examples of the task. this back to ac at the other frequency.
Crystal oscillators can be incor- This procedure is easiest to implement
procedures used.
porated into an inverter design where because it makes use of standard
AC to DC: This conversion path has the output frequency must be kept rectification and inverter packages.
been discussed when we dealt with within exacting limits. The cost of semiconductor con-
rectification in Part 14. A transformer DC to DC: The procedure here is to verters has fallen rapidly over the
is used to obtain the required ac volt- first form the dc to ac conversion. 1970 decade. This has brought about
age; this is then rectified with diodes After transformation to the correct new philosophies in power electrical
and smoothed to provide dc. voltage ( usually the need is a voltage engineering. In the future tnere will
DC to AC: This path first changes the increase) with a double-wound trans- be more use made of dc electrical
dc into a suitable ac signal which can former the output is full-wave rectified transmission. Speed-changing motors
then be transformed to the desired and smoothed. The transformers used are becoming easier to implement
signal level. The frequency of the ac use special iron laminations material to using frequency- varied supplies to
signal is decided by the output load get the best out of the square- wave drive conventional ac machines. Large
requirement for once produced it must input waveforms. Figure 11 is atypical dc motors are also becoming useful
remain at that frequency. ( In some 200 W dc to dc up-converter. The again because regenerative braking of
cases it is preferable to make use of a transistors, in conjunction with the large units — using them as agenerator
higher frequency than 50 Hz). transformer primary, form a square- driving into a load — can be put to use

100
90

80
70
EFFICIENCY I%)

60

50

40

30

20

10

O
0 1.0 2.0 3.0 4.0 50

OUTPUT CURRENT IA)

Fig. 9. Switching regulators provide the smallest overall package due to the
greatly reduced heatsink needed. They are also the most efficient form of
regulator.
128
to charge power into the ac mains by
the use of dc to ac inverters.
Revolutions have occurred in both LOAD

power and signal electronics. Attitudes

to problem solving are now quite
different to just a decade ago. No
I CHOPPER 3 CENTER TAPPED SUPPLY
doubt this trend will continue. 2 CENTER- TAPPED LOAD

REFERENCES
Heatsink Design: Several articles were
published in Electronics October —
December, 1973. Most power semi-
conductor device manufacturers have
design sheets available. Heatsink
makers usually provide graphs and 5 THREE PHASE HALF WAVE
LOAD
4 BRIDGE
design procedures. Forced- air cooling
is discussed in " Cooling high- power 6. THREE PHASE BRIDGE
equipment by forced-air convection",
Fig. 10. Various inverter configurations using SCR switches. Triggering methods have been
L. Katz, Electronics, September 21,
omitted for clarity.
1964. A good basic treatment of
"Power transistor Handbook 1961" Switching regulators: Low to medium
design is " Heatsinks for power tran-
sistors" Miniwatt Digest, December, deals with transistors. SCRs are power units based on an IC are dis-
1962. A deeper discussion is " Design covered in " SCR Manual", General cussed in " Designing switching
of cooling fins for silicon power Electric, 1964. regulators" R.J. Widlar, National Semi-
Converters and Inverters: An early conductor AN- 2, 1969. See also
rectifiers" Mullard Tech. Comm. Vol.5,
June, 1960. " Silicon Rectifier Hand- paper " Self-excited inverters" W.J.R. "Switchin§ regulators: the efficient
Farmer and R. J. Spreadbury, AEI way to power", R.S. 011a, Electronics,
book" Motorola, 1966, devotes a
Engng., March, 1965 provides a August 16, 1973, and application
chapter to heatsink design.
PovVer Devices The above Motorola tutorial summary. Many power engine- notes of Motorola, Delco Radio, RCA,
ering companies can provide papers Kepco, and others. •
handbook is most useful on power
diode design. A companion volume and articles.

Fig. 11 Circuit of dc up-converter. SECONDARY

300Vd.c.

FS1
OUTPUT

150Vd.c.
DC
INPUT
02 C2
COM.

CONCLUSION
In selecting the material for this series, it would have been helpful to have had powers of prediction. Even the
greatest minds have been wildly wrong about the near future. Lee de Forest who invented the triode valve in
1906, had this to say in 1937:-
"While theoretically and technically television may be feasible, commercially and financially I consider it an
impossibility, adevelopment of which we need waste little time dreaming".
Having knowledge of materials does not ensure that they will find immediate use. Germanium was
predicted well before it was isolated in 1886. Silicon was discovered in 1823. Both were available at the time of
the emergence of the valve in the 1900's; so was the idea of the transistor. History shows time and time again
that ideas seem to sit and wait. Few people in the 1950s foresaw the great impact that the transistor would
have on everyday life. In 1955, Sir George Thomson, a great physicist, said this about the germanium
transistor:-
"It is possible that ashort-range ' walkie-talkie', light enough to be carried regularly, may in the course of a
few decades replace agood deal of telephony over wires".
Who could have forecast the great changes to electronics over the following two decades. One wonders
what is to come in the next two!
129
 A
Index computer - advanced organisation 82
absolute code disk 66 - controlled testing 72 E
ECL 17 LASC 126
active - probes 114 - in analysers 67 to 68
EDP 81 latch 41
A to D convertor 5, 47, 65 - programme 81
efficacy of light source 34 large character display 38
addition 52, 53 contact bounce 64
control of power 124 electro-
mechanical switches 6 LED 25, 34 to 36, 43, 44, 84
addressing amatrix 45
electro-
sensitive recorder 120 - increasing operating voltage 36
algebra of logic 10 control with SCR's 126
alpha-numeric display 34 conversion time of DAC 51 electronic data processing 81 Leibnitz 76
converters 128 to 129 event marking recording 121 light emitting diode 34
alternate mode switching 107
cooling devices 124 electrostatic deflection 103 linear motor recorder 122
amplifier - connection modes 98
ENIAC computer 80 linearity of DAC 52
-CRO 104 core store 88
ERTS 87 line printer 86
analogue signal 49 correlator 70
analogue to digital conversion 49 counter system 41 to 44 excess-3code 39 to 40 liquid-crystal display 37 to 43
exclusive-OR 13 loading - cables 94
analogue to frequency conversion 52 coupling - units 93, 98 to 99
analysers 66 CPU 80, 84, 89 logic, algebra of 10
AND gate 11, 16 to 17 crazy digital 24 logic- gates 1010 13, 16
arithmetic - unit 80 CRO 102 - probe 69
fail-safe design 83
- with DAC 52 - controls 104 - state analyser 69 to 70
fan-out 13
ASCII code 39 - plug-ins 107 LSI method 78
FET 17
astable 21 - tube 102 lumped circuit 94 to 95
- switch 55
asynchronous counter 24 to 25 - trace finding 107 - probe 114
automatic - drawing machine 86 current - source 52 to 53 fibre optic recorder 123
- test systems 72 machine language 81
field effect transistor 17
- wiring tester 59 Fitts list 76 magnetic - core store 88
- disc store 89 to 90
DAC 49 to 50 flash conversion 52, 54
- drum store 90
- resolution 50 flatbed recorder 116-117
Babbage 76, 78, 79 - tape units 85, 89 to 90
data-processing 75 flip-flop 19 to 21, 23 to 24, 28
banking, using computers 76, 77 matching units 99
decade counter 28 flow chart 81
',etching counter 61 matrix of LED's 36, 43
BCD 30 decimal point 41 frequency - domain 67
mechanical - switches 6
- to binary 39 to 40 - counting 28 - measurement 63
memory stage 41
- to decimal 40 decoder-driver unit 42 to 43 - spectrum 67
Michelson and Stratton 80
decoding 32 - synthesis 71
- to 7segment 40,42 microprocessor 84, 90
binary - counting 24 - BCD 41 Fourier - analyser 68
minicomputer 76
- gates 40 fuses for SCR circuits 127
- multiplication 27 minimizing logic 15
- to BCD 39 to 40 deglitching 52
MODEM 87
- weighted resistor chain 48 delay in CRO 111 G Moir fringes 65
bi-stable 19 to 21, 41 delay lines 89 galvanometer for UV recorder 121 monostable 21, 23
bit 80 difference engine 77, 79 gates 10,40 MOS device 18
blanking 41, 44 differential - amplifier 107 glitches 51 to 52, 55 MOSFET multiplexer 56
Boole 10 digital - accuracy 50 graphic display 86 to 87, 115 moving-coil recorder 121
Boolean algebra 10 to 11, 14 to 15 - acquisition system 49 grounding 100 MTBF of devices 82 to 83
buffer amplifier - stage for display 40, 42 - cw analogue 5
multi-channel recorder 120 to 121
- store 44 - computer 10,74 H multimeter 5
byte 80 - controlled source 52, 71 harmonics 67 multiplexing - analogue 49, 56 to 57
- correiator 70 to 71 hardware, of computer 80 - displays 45 to 46
- display 32, 40 half adder 13 to 14 multiplication 52 to 53
cables 93 -encoded disk 66 heat sensitive recorder 120 multi-segment display 33
calculator 75 to 76, 79 - IC 9, 13 heat sink 124 to 125
calibration standard 72
Callendars recorder 121 to 122
- instruments 58 to 59, 65 - calculations 124 to 125 N
- instrument history 58 -types 125 NAND function 12 to 13, 16 to 17
camera for CRO 109 - linearization 65 hexadecimal format 34 neon indicator tube 32 to 33, 38
card reader 84
- machine history 76 high speed switch 9 Nixie tube 32 to 33
cathode ray oscilloscope 87, 102, 109 - multimeters 59,61 history of digital displays 32 noise -error detect in counting 63
central processor unit 80 - numbers 24 Hollerith - person 80 -error reduction in counting 64
character generation 69 - panel meter 49, 59 - card code 84 non-linear A to D conversion 55
chart recorders 116 - readout 66 horizontal amplifier in CRO 104 NOR function 12
- circular 118 - sensors 72 hybrid computer 49, 74 NOT function 11
- dotting 120 - systems 5
- formats 116
- heat sensitive paper 120
- to analogue conversion 49 to 50
- voltmeter 20 Image display 88
o
octal code 39
- ink jet 119 - germanium characteristics 8 image storage 109 on line operation 75
- ink pen 119 - logic 16 infinite hold sampler 56 open wire line 94
- linear motor 122 - silicon characteristics 8 ink-jet recorder 119 optical grating 65
- moving coil 121 - switch 7 ink-pen recorder 119 optical link 96 to 97
- paper sheet 116 to 117 discrimination pulse 70 instruction set 91 opto-
Mechanical display 32
- paper marking methods 119 display driver 39 to 40 integrated circuit - counter 27, 29 OR gate 11, 16 to 17
- photographic 120 displays 32 - computer 80 oscilloscope 102 to 103
- potentiometric 121 to 122 direct-digital control 75 - current sources 51 - special plug-ins 114
- pressure sensitive 119 divide by twelve 30 - display 37
- stripchart 116 to 117 dot generated characters 36, 43 - logic 16to 19
- ultra violet 120 dotting recorder 120 intensity control of display 42 parallel - conversion 52
- x-t116 to 118 driving - displays 40 interactive display 87 Pascal 79
- x-
y 116to 118 OIL 16 intrinsic safety 96, 98 Pegasus computer 68, 78
chopped-mode In CRO 106 dual beam CRO 106 inverter- digital 11, 13 Peirce 11
clock - see 'astable' 21 dual-In-line package 19 inverters 128 period measurement 62
- in computer 80 dual ramp integration 52 to 54 peripherals of computer 80, 84, 88
CMOS - counter 29 dual trace CRO 106 J, K phase control of SCR 127
- family 18 duty cycle 43 J-K flip-flop 21,27 - measurement 63
coaxial cable 93 to 94 DVM 20 Kelvin 78 phosphors 102
code converters 39. 46 dynamic resistance- of diodes Kelvin's siphon recorder 116, 118 photographic recorder 120
columnar display 32 dynamic response of recorder 123 keyboard entry 79 photophone 96
common-mode rejection 100 earthing shields 100 knitting - computer controlled 77 plotter 86

130
potentiometer recorder 121 to 122 RF — elimination 101 spectrum — analyser 67 to 69 TRIAC 126 to 127
power control 123 — shielding 101 storage by memory 110 trigger level 63 to 64
— rating, switches 6 ripple blanking 44 storage in computer 88 triple ramp integration 54 to 55
power transistor 126 ROM 45, 65, 89 storage oscilloscope 109 to 110 truth table 12 to 14, 24, 40
pressure sensitive recorder 119 to 120 RTL logic 16 strip-chart recorder 116 to 117 TTL 17
printer 85 to 86 R-2R ladder 50 to 51 strobing 43, 45 — gate 9
probes CRO 113 subtraction 53 turbine flow meter 66
process control computer 75 to 76 successive approximation 54 to 55
process telemetry link 96 to 97 sample and hold 49. 5510 56 swept-tuned analyser 67 to 68 U
programme, computer 81, 91 sampling CRO 111 switched tuned analyser 67 to 68 unijunction device 127
projection display 32 to 33 scanning 43. 45 switched resistor DAC 50 UV recorder 120
pulse — height analyser 68 Schmidtt trigger 23 switching regulator 126 to 128
— output transducer 66 Schottky diode 36 — resistance 6 V
— time modulation scientific computing 74 synchronous counter 25 variable delay sweep 112
— width measurement 62 SCR 126 to 127 synthesis 67 VDU 87
punched cards 77.84 sensors 5 synthesizer — frequency 71 Venn diagram 15
punched tape 84 serial conversion 39 vertical amplifier in CRO 104 to 105
serial to parallel conversion 39 visual display unit 87
servo DAC 54 to 55 tape — punched 84 to 85 voltage— to frequency conversion 49. 53
quartz crystal thermometer 66 to 67 settling time of DAC 51 — reader 84 to 85 V to Fconversion 49,53
seven-segment display 33 to 34. 37, 43, 45 telemetry — link 95. 97
to 46 teletype 84 to 85
radio — link 95 Shannon 10 to 11 temperature coefficient — DAC 51 waveguide 94 to 95
reading references 15, 23, 30, 38, 46. 55. 57, shielding 1CO to 101 test instrument 58
64, 72, 82. 83, 92, 101, 115, 123, 129 shift register 24 to 26 thermal resistance 124 to 125 X, Y
read-only memory 45 single trace CRO 105 — of mounting methods 125 X-trecorder 116 to 118
real time analyser 67 to 68 simulator 87 thermocouple 65 X-
yrecorder 116 to 118
real time working 75 skin effect 95 thermometer, digital 65 X-
yplotter 86
recorders 116 solid-state — display 34 time base 103, 105 Y-
tplotter 86
redundancy in logic 15 — multiplexer 56 time— delay unit 11110 112
redundant circuits 82 solid-state recorder 123 time — interval measurement 6110 62
relay 7 — store 89 — sharing 75 zener barrier 42
— computer 80 — switch 7 totalizing 61 to 62 zener diode 36
reliability 82 software of computer 80 transistor — switch 8109 zero voltage crossing control 127
reverse bias, diode 8 speed-up capacitor 9 transmission — links 93 Zuse 80

Both volumes of this book are available from all major newsagents, technical bookshops and electronics
suppliers, or direct from the publishers, Electronics Today International, 15 Boundary Street, Rushcutters Bay
NSW 2011 (price $5.95 plus $ 1.00 post and packing). Electronics it's Easy is also available in two-volume
hard-cover form. It may be obtained from technical bookshops and electronics suppliers or direct from the
publishers at the above address ( price $ 12.95 per volume plus $ 1.00 per volume post and packing).

131