Original from

Digitized byGoOgle UNIVERSITY OF MICHIGAN

 ttOPBRTT Of

ARTBS SC1ENT1A VERITAS

Digged by ^OOgK UNIVERSITY OF MICHIGAN

 ADVANCED
ANTENNA
THEORY

SERGEI A. SCHELKUNOFF
MEMBER OF THE TECHNICAL STAFF
BELL TELEPHONE LA BO PA TOBIES

NEW YORK. JOHN WILEY & SONS, INC.

LONDON. CHAPMAN & HALL, LIMITED
1952

80
Original from
igitteed by GoOglC UNIVERSIT/ OF MICHIGAN
 Copyright, 1952

Bell Telephone Laboratories, Inc.

AU Rights Reserved
This book or any part thereof must not
be reproduced in any form without
the written permission of the publisher.

Library of Congress Catalog Card Number: 52-5084

PRINTED IN THE UNITED STATES OF AMERICA

Original from
UNIVERSITY OF MICHIGAN
 To my wife

Jean Kennedy Schelkunoff

Original from
Digitized by ^OOgie UNIVERSITY OF MICHIGAN
 PREFACE
Hertz is the founder of antenna theory. Three quarters of a cen-
tury ago — only a few years after publication of Maxwell's treatise on
electromagnetic theory — Hertz solved the first, the simplest, and the
most basic antenna problem : the problem of a small capacitor antenna.
In those days the progress of antenna theory was slow. Another quarter
of a century had passed before the next significant advance was made
when Pocklington showed that the current and charge on thin perfectly
conducting wires are propagated approximately with the speed of light
and that, between any two points of monochromatic excitation, the
current distribution is approximately sinusoidal. Until recently the
practical theory of antennas and antenna arrays was based on these two
fundamental results. Hertz's solution enables one to calculate the field
and distribution of radiation for any given distribution of electric current.
Jointly with Pocklington's result and with the aid of the principle of
conservation of energy, it makes possible the solution of many other
practical problems.
During the last ten or fifteen years, however, it has become in-
creasingly evident that the time is here for further extension of funda-
mental antenna theory. This realization has stimulated work in this
field, and the end of it is not in sight. For this reason a unified account
of recent developments is timely. In this book I present a compact but
general exposition of Hall6n's method of obtaining asymptotic solutions
for tinear antennas, Stratton and Chu's theory of spheroidal antennas,
and my theory of biconical antennas and thin antennas of arbitrary
shape. I have included much hitherto unpublished work, and for rea-
sons of space have condensed some developments which can be found in
fuller form elsewhere. Perhaps needless to add, there is no routine
method for solving antenna problems. At present we have to make the
most of isolated solutions. Fortunately these solutions bring out dif-
ferent aspects of antenna behavior and together present a satisfying
theory.
While my principal object in this book is to present mathematical
methods for solving antenna problems, I also stress the resemblance of
antennas to common circuits and transmission lines. Just as science is
not a collection of unrelated facts but a body of knowledge in which facts
are grouped together on the basis of their essential similarity, so applied
mathematics does not consist merely of methods for solving problems
vii

Original from
UNIVERSITY OF MICHIGAN
viii PREFACE
but is a way of recognizing the common features in apparently
also
dissimilar phenomena. Mathematics helps one to evolve physical pic-
tures which bring order into a multiplicity of facts. When the quanti-
tative solutions of given problems are hard to find, these pictures often
give qualitative answers. Maxwell's equations themselves express a
physical point of view —
the point of view developed by Faraday and
formulated by him in geometrical terms. In Faraday's picture of elec-
tromagnetic phenomena, even a vacuum is a kind of "medium" in which
electric and magnetic forces cause electric and magnetic "displacements."
It is not necessary to endow these concepts with their literal meanings;
the picture remains just as useful if they are thought of as expressions of
what will happen when appropriate experiments are performed. Max-
well's equations are the relations between the spatial rates of change of
the forces and the time rates of change of the displacements. They are
reminiscent of the equations for forces and displacements in springs and
at once suggest that electromagnetic disturbances are propagated from
place to place. From this point of view, the arms of an antenna form
the banks of a channel in which the" waves excited by the source are con-
fined before they emerge into unlimited space. In this sense antennas
are similar to waveguides. This simple picture led me to a successful
solution of certain antenna problems which is presented in one of the
chapters in this book.
Hertz's analysis of electromagnetic waves excited by an oscillating
charge gives automatically the forces existing between two such charges.
From this point on, waves in the medium may be ignored. Instead
one's attention may
be concentrated on the currents in the various sec-
tions of the antenna as is usually done in the case of electric networks.
Mathematically, Maxwell's equations with various boundary conditions
become converted into integral equations. It was by this method that
Pocklington obtained the important sinusoidal approximation to the
current in thin antennas, and the natural frequencies and damping con-
stants of circular loops. More recently Halten discovered a method of
deriving asymptotic solutions of integral equations for thin antennas.
Still more recently Schwinger discovered variational properties of cer-
tain integrals which are very useful in approximate calculations. I have
devoted two chapters to the essential features of this method of approach;
but for actual applications I refer the reader to the original papers by
Hall6n and his followers.
An alternative to this approach is a direct solution of Maxwell's
equations by the method of separating the variables. Unfortunately
there are only a few coordinate systems in which these equations are
separable. Of these, the important systems are Cartesian, circular-

Original from
UNIVERSITY OF MICHIGAN
 PREFACE ix

cylindrical, elliptic-cylindrical, spherical,and spheroidal. The first

three are very convenient for solving waveguide problems but are not
well suited to antenna problems. All their coordinate surfaces are
infinite in extent whereas the b oundaries of a ctual antennas are finite.

The spKeroidal system (and the spherical as its special case) is the only
system that contains coordinate surfaces of finite dimensions. For this
reason spheroidal antennas received considerable attention in theory.
As far back as 1897, Abraham analyzed free oscillations on thin spheroids.
Spheroidal functions, however, turned out to be very complex, and it
was only in 1941 that Stratton and Chu succeeded in obtaining and
interpreting the complete solution in the case of forced oscillations.
Although spheroidal antennas are not used in practice, their theory is
important because it exhibits one aspect of antennas: the resemblance
to leaky resonators. The impressed electromotive force excites various
modes of oscillation of current in the antenna. In each mode the cur-
rent excites waves in surrounding space, and energy is lost by radiation.
In the middle '30's I considered several possible methods of dealing
with cylindrical antennas. It was then that I recognized the possibility
of developing a method that would conform to one outstanding physical
characteristic of all thin antennas regardless of their shape: The waves
excited in the center of such an antenna are guided by its arms and are
almost totally reflected from the ends. Spherical coordinates were
clearly indicated for expressing radial propagation from a point. To
simplify the boundary conditions I assumed at first that the antenna
arms were conical. The surface of the antenna was divided into the
major conical part and the small complementary ends. The boundary
conditions over the major part were to be satisfied first. In fact, for
sufficiently thin antennas the end surfaces could be neglected. Sub-
sequently the conical boundary was deformed into an arbitrary shape.
The effect of such deformation happened to be easily calculable. Thus
it was no longer necessary to rely on spheroidal coordinates as sole

coordinates suitable for analysis of thin antennas by the method of

separating the variables. I have assigned a long chapter to this method
because my early papers were too sketchy.
Many years passed before Hertz's and Pocklington's theoretical
results became fully exploited. Lt may take as many years again for full
exploitation of the new results, although even now there is ample
evidence that recent theoretical work has not been in vain. Practical
engineers are skillful in squeezing the most from a new idea or from a
particularly significant solution. I take no chances in recommending
antenna theory to applied mathematicians as a fertile field, abundant in

problems requiring imagination, ingenuity, and analytic prowess. I

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 X PREFACE
hope that this book will interest them as well as radio engineers. Mathe-
maticians need not fear that they may fail to understand it for lack of
specialized engineering knowledge. And the type of mathematics needed
in advanced antenna theory is beginning to look less fearsome to the
new generation of engineers than it did to their predecessors.
I am very grateful to Miss Marion C. Gray who has checked the
text and equations in the original manuscript and in proof, and who has
prepared the index. I thank Miss Dorothy T. Angell for her assistance
in proofreading. Mr. B. A. Clarke made many helpful suggestions in
connection with illustrations, and Mr. H. P. Gridley did the drawings.

S. A. S.
• New York, December 1951

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 CONTENTS
1 SPHERICAL WAVES
1.1 Classification of antenna theories 1
1.2 Maxwell's electromagnetic laws, differential equations, and boundary
conditions 2
1.3 Maxwell's equations in orthogonal coordinates 4
1.4 Maxwell's equations in spherical coordinates 5
1.5 Transverse magnetic waves 6
1.6 Transverse electric waves 10
1.7 General expressions for spherical wave functions 11 ,

1.8 Waves in free space 11

1.9 Waves in regions bounded by perfectly conducting conical surfaces
with the apices at the origin 12
1.10 Orthogonality
1.11 Transverse electromagnetic waves
1.12 Dipoles and solenoids; capacitors and loops
........ 12
15
19
1.13 Infinitesimal electric circuits, small antennas, and transmission lines . 23
1.14 Biconical transmission lines 26
1.15 Quasibiconical transmission lines 28

2 MODE THEORY OF ANTENNAS

2.1 General mathematical formulation of the antenna problem ... 32
2.2 Two general methods of approach to the solution of the antenna
problem 34
2.3 Antenna analysis in terms of spherical coordinates 35
2.4 General expressions for the field around a biconical antenna . 39
2.5 Antenna current and transverse voltage 43
2.6 Fundamental theorem 44
2.7 General formulas for the terminating admittance 45
2.8 Auxiliary formulas for calculating the terminating admittance . 45
2.9 Symmetric spherical antennas 49
2.10 Thin biconical antennas
2.11 Wide-angle conical dipoles
.

... ... ....

.... 54
60
2.12 Antennas in dissipative media 65
2.13 Extension of known solutions 66
2.14 Input regions 68
2.15 Thin antennas of arbitrary shape 71
2.16 Asymmetric spherical antennas 72
2.17 Asymmetric conical dipoles . 76
2.18 End-fed antennas . .79
2.19 Current element above a discoid ground 83
2.20 Treatment of reflection phenomena by successive approximations . . 90
2.21 Miscellaneous problems 92
2.22 V antennas, fans, porcupines 97
2.23 Variational methods 99
2.24 Spherical waves on infinitely thin wires 102

-
Z *-,,-,,,!-, Original from
Digged by ^OOgie UNIVERSITY OF MICHIGAN
xii CONTENTS
3 SPHEROIDAL ANTENNAS
3.1 Prolate spheroidal antennas . Ill
3.2 Oblate spheroidal antennas 125

4 INTEGRAL EQUATIONS
Nonhomogeneous Maxwell's equations 128
4.1
4.2
4.3
Special solutions ...
Equations for surface currents
. ....
... 129
130
4.4 External and internal oscillations on closed surfaces of revolution . 131
4.5 Integral equationsand Kirchhoff's network equations . . 134
4.6 Ring source, Green's function, and the input impedance 135
4.7 Solutions in terms of orthogonal functions 136
Variational theorems 136
4.8
4.9 Asymptotic solutions .... .... 138

5 CYLINDRICAL ANTENNAS
...
5.1
5.2
5.3
5.4
5.5
Fundamental equations
Properties of the kernel
HalleVs method of iteration
Special asymptotic expansions
Approximate integral equations
....
.... 140
141
142
144
149

6 NATURAL OSCILLATIONS
6.1 Forced and natural oscillations . 152
6.2 Equivalent networks 154
6.3 Small antennas 160
Thin conductors — ... 163
6.4
6.5 Thin conductors
I
—
II ... .... . 169

Problems . . 175

APPENDIXES
I
II
Integrals in Antenna Theory .

Inverse Radiation Impedauce of a Thin Cone

.

.
....
. .
191
195
III Inverse Mutual Radiation Impedance of Two Arms of a Thin Bi-
conical Antenna 196
IV The R
Inverse Radiation Resistance, a and Reactance,
, a , of a Thin X
Biconical Antenna with Equal Anns Making a 90° Angle, and the
Mutual Radiation Resistance, between Its Arms . . . . 197
V Inverse Radiation Impedance, Za = Ra + iX a> of Two Thin Cones
Making an Angle & J9g
VI The Inverse Mutual Radiation Resistance of Two Inclined Cones . 202
VII The Inverse Mutual Radiation Resistance of Two Thin Cones, of
lengths li and k, Making an Angle 180° 203
VIII M(fil)and N(fil) for Cylindrical Antennas, for Use in Equation 1-136 . 204
IX M(l) and N(t) for Tapered Antennas ! 205
In^ex 207

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 4

SPHERICAL WAVES

1.1 Classification of antenna theories

Passive electric structures are usually divided into circuits, transmission
lines, waveguides, resonators, and antennas; hut structurally there are
no sharp dividing lines among these subdivisions. Any electric circuit
dimensions) will radiate some power and thus will act as
(of practical
an antenna, although unintentionally. At low frequencies the power
radiated by an ordinary circuit is negligibly small ; at high frequencies
special precautions (shielding, for instance) must be taken to reduce
the radiated power, or else the circuit will not have the intended per-
formance. Structurally some antennas are merely sections of transmis-
sion lines, short or long; they are also resonators. As resonators
antennas are naturally poor ; and, since in antennas the resonant prop-
erties are usually undesirable, efforts are often made to eliminate these
properties as much as physically possible.
Fromthe point of view of electromagnetic theory, there are also
no sharp criteria which enable one to distinguish among circuits, trans-
mission lines, waveguides, resonators, and antennas. Mathematically
we are confronted with the same general problem: that of solving
Maxwell's equations; but the most convenient method of solution
depends on the type of structure and the most significant characteristics
of the solutions that are being sought. For instance, the most significant
characteristic of the field associated with a capacitor is that the electric
intensity at a typical point is substantially in phase with the electric
charge on one of the plates. Similarly, the magnetic intensity of the
field of an inductor is substantially in phase with the electric current.

To derive "circuit equations" from Maxwell's equations most directly,

we should express the solutions of the latter in terms of the electric
charge and the current in the conductors of the circuit. On the other
hand, in a waveguide it is usually more convenient to concentrate first
l

Original from
UNIVERSITY OF MICHIGAN
2 SPHERICAL WAVES 1.2

on finding the field, and from it the current and charge in the walls.
Then in calculating the field the walls are considered merely as bound-
aries of the field.
In these examples it is possible to interchange the methods of
solution; but only at the price of making simple solutions unwieldy.
Linear antennas, however, have equal affinity to circuits, to cavity
resonators, and to waveguides. On the one hand, they may be con-
sidered as radiating circuits, or rather as radiating transmission lines
on the other hand, they, and the surrounding space, may be thought
of either as cavity resonators or as waveguides. Antenna theories
may thus be classified broadly into two types: (1) circuit theories, in
which the emphasis is on the current and charge in the conductors;
and (2) field theories, in which the emphasis is on the fields, and the
conductors appear only as the boundaries. Field theories may further
be subdivided into resonator theories depending on the affinity between
antennas and cavity resonators, and into mode theories exhibiting the
resemblance of antennas to waveguides.

1.2 Maxwell's electromagnetic laws, differential equations, and

boundary conditions
In their most general form, the laws of interaction between electric
and magnetic fields, discovered by Ampere and Faraday and generalized
by Maxwell, may be stated as follows:
Ampire-Maxwell law. The total electric current (the sum of
1.

conduction, convection, and displacement currents) passing through

a given surface equals the magnetomotive force (the line integral of the
magnetic intensity H) round the edge of the surface.
2. Faraday-Maxwell law. The total magnetic current (the sum
of the magnetic displacement current and the convection current, when
the condition of magnetized bodies is expressed in terms of equivalent
magnetic charge) passing through a given surface equals the negative
electromotive force (the line integral of the electric intensity E) round
the edge of the surface.

It is assumed that, when

the handle of a right-handed screw is turned
in the positive direction of integration round the edge of the surface
of
integration, the screw advances in the positive direction of the normal
to the surface.
In antenna theory we are not concerned with the internal structure
of electric generators. Each generator is enclosed by some surface on
which the electric field is assumed to be given. Under some conditions
the details of this "impressed" field distribution are either unimportant

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
1.2 •
MAXWELL'S ELECTROMAGNETIC LAWS 3

or relatively unimportant, and it is sufficient to know only the voltage

across the terminals of the generator. In any region in which free
electrons are confined to conductors with no moving parts, and in which
the electric intensity E and the magnetic intensity H are differentiable
functions of time and position, the electromagnetic laws of interaction
between E and H may be expressed by partial differential equations,

curl£ = curltf = gE+e-?^-, (1)

at at

where n are, respectively, the conductivity, the dielectric constant,

g, e,
and the permeability of the medium.
If E and H are of the form E(x, y, z) exp(pt) and H(x, y, z) exp(pt),
where the oscillation constant

P = {+/<- (2)

may be any complex number, equations 1 become

curl E= -pnH, curl H = (g + pi)E. (3)

lip = jo3, we have the Equations 3 are Laplace

steady-state equations.
transforms of equations 1, and hence are equally general. In antenna
theory, however, we are interested mostly in the steady-state case,
and occasionally in transient oscillations in which the growth constant
£ is negative. Taking the divergence of equations 3, we have

div H = 0, div E= 0. (4)

At a boundary between two media with different electromagnetic

properties, the tangential components of E and H are continuous. These
conditions enable us to connect solutions of equations 3 across a surface
of discontinuity in the medium. At a perfectly conducting surface
the tangential component of E (and, hence, the normal component of H)
vanishes by definition. Across such a surface the tangential component
of H
is discontinuous. Let n be a unit normal to that side of the surface
on which the tangential component t H
is positive in the direction

CXn, where C is the linear electric current density then, ;

H + - Hr
t =C X n. (5)

The normal component of the electric displacement density is also

discontinuous; thus,
e+£?„+ - *-E«- = <r, (6)

where a is the surface density of electric charge.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
4 SPHERICAL WAVES 1-3

At a surface of discontinuity moving in a homogeneous dielectric

medium, the boundary conditions are

Ei+ _ = Br - nHr, # + = Hr, E n + = Br, (7)

where

(8)

is the intrinsic impedance of the medium. The positive directions

of E t and H
t (which
are perpendicular to each other) have been chosen
so that the surface of discontinuity is moving in that direction in which
90°
a right-handed screw advances when its handle is turned through

M
from E to H%. Equations 7, 8, and the following expression for the
t

velocity with which the surface of discontinuity is expanding normally

'
v = W-H, (9)

are obtained directly from Maxwell's electromagnetic laws.

1.3 Maxwell's equations in orthogonal coordinates

Antenna problems are considerably simplified if all conductors are
assumed to be perfect. This assumption does not detract from the
practical value of the results. In the first place, the effect of the con-
ductivity of good conductors on the field distribution is small. In the
second place, the principal part of this effect can be evaluated a posteriori
from the solutions of idealized problems. When the conductors are
perfect, they become merely the boundaries of fields in dielectric media
where g = 0. The steady-state case is the one of principal interest,
and in most subsequent equations we shall assume that p = jw. In
some cases, however, we shall consider transient fields, and then we
shall replace ju by the more general parameter p. As far as algebraic
details are concerned, there is little difference between p and ju.
In orthogonal coordinates {u, v, w) Maxwell's equations become

d(e3 E w ) d(e 2 Ev )
dv dw
d(e3H„) d(e2 H v)

dv dw
d(eiEu ) d(e3E»)
—jo)H€3€iH9
dw du ,

(10)
d(eiHH ) 5(e3 // w )
dw du

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
1.4 MAXWELL'S EQUATIONS IN SPHERICAL COORDINATES
d{e2 E9 ) d{e x Eu
du dv
)
= -jo>ueie2 H W)
d(e2 H v) d(e x H u)
= jueeie2 Ew .

du dv

To these often convenient to add the divergence equations 4, even

it is

though they follow from equations 10 and do not constitute independent

equations,

idu
(e#zBu ) + ±- {emE,) + £
dw ( ei e2 E„) = 0,

(11)
3
du v 2 e3
(e
"
Hu +)
' '
-jL
dv
(eaeiff.) + {e x e2 Hw ) = 0.

The coefficients, ei, e 2 , ez, are defined by the metrical form

rf8 2 - ei 2 du 2 _J_ C2 2 dv 2 + e3 2 du 2>

,
(12)

Fia. 1.1 Cartesian (z, y, z), cylindrical Fia. 1.2 Vector component* in spheri-
(p, v>, z), and spherical (r, 0, *>) coordinates, cal coordinates.

In antenna theory the most important coordinates are spherical

and spheroidal. In this chapter we shall consider spherical wave
functions.

1.4 Maxwell's equations in spherical coordinates

In spherical coordinates (Figs. 1.1 and 1.2),

da2 = dr* + r2 W+ r2 sin

2
d^. (13)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
6 SPHERICAL WAVES 1-5

Hence,

(14)

d^> dr

^ - sin $ 4~
or
= jcoer sin E«.

The divergence equations are

^W) 4^ + ft)+r
^ =0 '
(15)

sin 9 -|r (f*Hr) + r (sin * //,) + r = 0.

1.5 Transverse magnetic waves

magnetic (TM) if the radial
Spherical waves are said to be transverse
component of magnetic intensity is zero. In this case we
conclude
transverse components
from the second divergence equation that the
thus,
of H are derivable from a stream function II ;

Similarly, from the first equation in the set 14, wo conclude that the
transverse components of E are derivable from a potential function V,

rB, = -^-- rrin»*,= -4£" (17)

Substituting from equations 16 and 17 in the fourth

and sixth equations
of the set 14, we have
= -jmV. (18)

radial electric intensity may be expressed

from the remaining three
The
equations in either of the following two forms,
i r i a / . . an \ ,
1 d 2 n~|

(19)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 1.5 TRANSVERSE MAGNETIC WAVES 7

where

= ufa)* = £. (20 )

isthe phase constant. Comparing the two expressions for E„ we find

that the stream function satisfies the following equation

r2
i^ + w^( 8me
^J + Tar's? (21)

The complete field has thus been expressed in terms of a single scalar
function.
Consider now fields for which

U{r,$,p) =R{T)T{e,<t>) (22)

so that the field pattern is the same on all spheres concentric with the
origin. Substituting in equation 21, wc have

sin A (sin 8 + 0. -_„(„+ 1) sin 2 9 7\ (23)

where y is the separation constant. From equations 22 and 24, we have

«.[^ + 2feja.]a (25)

From equations 18 and 25, we obtain

Hence, V and II vary as the voltage and current in a transmission line

with a distributed series inductance m henrys per meter, series capaci-
tance zr2 /v(v + J) farad-meters, and shunt capacitance e farads per
meter. The corresponding transmission line diagram is shown in
Fig. 1.3a. If we differentiate equations 26 with respect to 6 and substi-

tute from equations 16 and 17, we find that rE 9 and rH v satisfy equa-
tions 26. Likewise, r£„ and— rH$ satisfy these equations.
When the stream function may be expressed as in equation 22, the
ratios Et/H v and E 9 /H$ are independent of and p. These ratios are
called radial wave impedances. The algebraic signs are chosen so that
the transverse components E t ,
H
and the direction in which the imped-
t ,

ance is taken, form a right-handed triplet of orthogonal directions.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
8 SPHERICAL WAVES 1.5

Hence, the radial wave impedances in the directions of increasing and

decreasing r are, respectively,

* + = 7Z7 = -^7' (27)

The reciprocals of the wave impedances arc called the wave admittances
and denoted by Y.

T 1 T % T I

(a) (b)

Fig. 1.3 Transmission-line diagrams representing radial propagation: (a) TM

wavea, (6) TE waves.

From equations 16, 17, 19, and 22, we have the following expressions
for the field components:

rdnfffi-jgi rH r = -R^> (2g)

tf# = Z+H p ,
E 9 - -Z +i/,, j«er»tf r = r (p +
where

Solutions of equation 24 may be expressed in terms of the normalized

Bessel functions defined in terms of regular Bessel functions as follows:*

/»,(*) = HrtJH/^W, AM*) = (J*0 M iW*).

H (30 >
*n,(x) = (-^) W. = (*«:)« J,+M (*).

These functions are related as follows:

In,(jx) = j""Jn,{z),
'
Kn,(jx) = (-j)'+V»>G«0 -jNn,(x)]. *

•In the reference texts by the author, Electromagnetic Waves and Applied
Mathematics (D. Van Nostrand, New York), the normalized functions were distin-
guished from the corresponding regular functions by the addition of a circumflex
accent. The notation introduced here now seems preferable.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
1.5 TRANSVERSE MAGNETIC WAVES 9

In particular, for integral values of v,

Knofr) - e~*, Km(x) = *r*(l + ar 1 ),

+ 3x-» + 3x"2
Kn 2 {x) - e~*(\ ), (32)

+
7 ^)--\t w .(i- ^(L)-
-

When n is not an integer, the series for Kn is asymptotic. The In

function is given by
In,(x) = \[Kn,{-x) + (33)

The general solution of equation 24 is

= A Jn,{fir) + ivn,(0r) - A' £»,(#r) + £' In,(jfr)- (34)

The first form is particularly convenient in the steady-state case and

the second in the transient case (and also for waves in dissipative
media). In the transient case, for instance, = i«Gie)
M is replaced
jfi

by a more general expression p(//s)

w where p is complex, and

R = A'Kn,(prV&) + B' In,{pr\fa). (35)

Separating the variables in equation 23, we obtain

T(eiV) = e(*)*(*), (36)

where

sin 6
-J-
(sin B
^+ [»<> + 1) sin
2 - = 0, (37)

^ = -M 2
*. (38)

Equation 37 is the associated Legendre equation. In most cases

general solutions of equations 37 and 38 are*

$ = E cos \up + F sin

But, when v +
m is an integer, P/(cos 5) and /V( — cos 5) are not lin-
early independent, and, for the second solution, we may use the limit of

Q,-(co89 )=-l
^(cos^cos^ + mV -P^-cosO) t

2 sm{p + H)TT

as v + m approaches an integral value.

If * = 0, we introduce a new independent variable,

P = cot |# f
(41)
We do not expect any confusion between the separation constant n and the
*
permeability p.

Original from
Digged by Google UNIVERSITY OF MICHIGAN
10 SPHERICAL WAVES 1.6

in equation 37 and obtain

>£('f) -"* = <> < 42 >

The solution of this equation is

6 = Cp" + Dp" = C cot" \0 + D tan" J0, n * 0,

= C log P + Z) = Clog cot \B + D, M = 0. (43)

In this case E r, as well as H r, vanishes, and the waves are transverse

electromagnetic (TEM).

1.6 Transverse electric waves

Spherical waves are said to be transverse electric (TE) if the radial
component of E vanishes. The field expressions are similar to those
for TM waves, with the parts played by E and H interchanged. Thus,
E is derivable from a stream function and the transverse component
of // from a potential function U. Omitting the details of the deriva-
tion, we have

r Bin E9 = —
dip
> rE r9
30

rH$=
"W 9 rsmeH 9 = --|^- (44)

(0 »)
The stream function satisfies equation 21. When * is of the form 22,
dl
r sin E, = -R -|^- > rE,
~* = R
d9 60
(45)
H, = - K+£ ri H v = Y+Eh jo>pr*H r = y(v + 1)RT,
where the radial wave admittance in the direction of increasing r is

1
jupR dr <46 >

The stream function * and the potential function U satisfy the

following equations:

_ = _JttMCr f _ = - + -VJ^-J *. (47)

Likewise rE 9 ,
rH 9 and -rE„, rH
, t satisfy equations 47. Hence, these
three pairs of functions vary with r as the voltage and current in a
transmission line with a distributed series inductance p., shunt capaci-

Original from
Digiiized by Google UNIVERSITY OF MICHIGAN
 1.8 WAVES IN FREE SPACE 11

tance e, and shunt inductance nr2 /v(v + 1) all per unit length. The
corresponding transmission line diagram is shown in Fig. 1.36.

1.7 General expressions for spherical wave functions

The most general spherical wave is the sum of transverse magnetic
and transverse electric waves. Hence,

1 d 2*i l d* 2
tE
jioe drdd sin dtp
(48)
1 d*i . 1 d 2 *2
r** 9 '

sine dtp jco/j drd9

jtM sin 6 dr dip S0 d& jvfi sin 8 dr dtp

where both * functions satisfy equation 21. To prove the generality

of equations 48, consider any given field. From E we r can determine
*i and the corresponding part of the total field. Subtracting this part
from the total field, we obtain a field for which E r = 0.

1.8 Waves in free space

sources of waves are confined within a shell bounded by concentric

If all

spheres r = n and r = r 2 > n, the field intensities outside the shell

must satisfy the homogeneous Maxwell's equations 14 and the following
boundary conditions: (1) All wave functions must be regular, (2) they
must be periodic functions of <p, and (3) at infinity they must represent
waves traveling from the origin. Theperiodicity condition requires m
jn the expressions. 39 to be an integer. When v is not an integer, the
associated Legendre functions in equations 39 become infinite, the
first at = t and the second at = 0. Hence^ for waves in free space v
jnust be an integer, and the characteristic T function is of the form

T = P n m (cos 6) (E cos rrup + F sin rrup). (49)

If n = 0, the only regular T function is a constant. The corresponding

fieldvanishes identically. Hence, in free space there can be no trans-
verse electromagnetic waves.
Since Nn„(x) becomes infinite at x = 0, in the region < r<n
B in equation 34 must vanish. In the region r 2
the coefficient < r < *>
we must have B = — jA in order to obtain a progressive wave traveling
from the origin. In the second form of the solution, A' = in the
region r < n, and B' = in the region extending to infinity. If there

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
12 SPHERICAL WAVES 1.10

is another surface of discontinuity, at r = r$ > fi, for instance, than

in the region r2 < r < r3 we must retain both A and B.
1.9 Waves in regions bounded by perfectly conducting conical
surfaces with the apices at the origin
Consider now a region bounded by one or more perfectly conducting
surfaces
e = /«. (so)

At each surface the tangential component of E (and, therefore, the

normal component of H) must vanish. In the case of TM waves,
the radial and transverse components of E are given by equations 28,
and E r will vanish if either v = or T vanishes on the boundary,

The transverse component E t is derivable from a potential function

which is proportional to T. Hence, the tangential component of E t
will vanish if T is constant on the boundary, and, in particular, if it
vanishes. Thus, equation 51 is the boundary condition for waves, TM
when v 7^ 0.

For TE
waves the transverse component of H is derivable from a
potential function U. Its component normal to the boundary is
—dU/dn. Since U is proportional to T, the boundary condition is

These boundary conditions define a set of characteristic values and

functions for any given set of perfectly conducting conical surfaces.
If v = 0, the waves are transverse electromagnetic. Such waves
can exist only in the presence of at least two perfectly conducting cones.
In this case we express the electric intensity as the gradient of a potential
function, and the boundary condition is

T = constant (53)

on the surface of each cone.

1.10 Orthogonality
Green's theorems for a spherical surface are

JJ grad (7-grad V dn = Ju ~ ds -
JJ U &V da, (54)

ff V&V-VAV)*-f(u£-Y-%)
(
ds, (55)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 1.10 ORTHOGONALITY 13

where effi = sin $ d$ cbp is a differential solid angle, the line integrals are
taken over the boundary and the normal
of the area of integration,
derivatives are taken along the outward normals. On the surface of a
unit sphere, the gradient of a scalar function T and the divergence of a
vector function F are

-'-(£•=•-£)• (56)

Hence, the Laplacian is

d 2T
AT = div 6
grad T « -4— 4r (sin + ~k (58)
sin B d$ \ BB ) sin'2 e i d<p>

When T is a solution of equation 23,

AT = -v(v + 1)T. (59)

For a pair 7*i, T2 of such solutions, equations 54 and 55 become

jyVad7Vgradr2 dG=
f Ti ^ ds + v2 {v 2 + \)fj TfoaXl, (60)

Mvi + i)-v2 (v2 + i)]ffT T2 da= 1 ffr^-^^jdn. (61)

If neither n nor v2 vanishes, the line integrals vanish on the boundary

of each perfectly conducting cone. If there are no perfectly conducting
cones, the line integrals vanish when the surface integration is extended
over the entire unit sphere, because T\ and T 2 are regular. Therefore,
if v\ V2,

= =
JJT1T2 rfO 0,
Jjgrad 7Vgrad T 2 dil 0, (62)

and Ti, T2, as well as their gradients, are orthogonal. If v\ = v2 = v

and Ti = T2 = T,

JJ |grad T\ 2 d& = v(v + 1) jj 2 da. (63)

If there are two or more linearly independent T functions for the

same v, they may or may not be orthogonal. If they are not, we can
always obtain linear combinations of these functions which are orthog-
onal. Let T\ and T2 be nonorthogonal functions, and let
Tz = Ti + kT2 . (64)

Original from
D,g,t ( ze0 by ^OOgie
f~" r^s~* ,-%\s*
UNIVERSITY OF MICHIGAN
 14 SPHERICAL WAVES 1.10

This function is orthogonal to T\ if

jj x Tz da = Jj Ti*<to + kjj TiT 2 <Kl = 0, (65)

that is, if

k = - (66)

By our assumption the denominator does not vanish and k is well

defined. Similarly,a third solution which is not orthogonal
if there is

to either of two orthogonal solutions, we form a linear combination of

the three and determine the coefficients to make it orthogonal to the
first two.
If Ti belongs to v\ \* and vanishes on the boundary, while 7*2

belongs to v% = 0, then, from equation 60, we obtain

grad Ti -grad T2 dti = 0. (67)

These orthogonality conditions enable us to determine the field

when the components of either the electric or the magnetic intensity
tangential to a given sphere r = a are given.* Suppose that we know
E$(a, 6, ip) and E f (a, 0, ip). From the first equation in the set 14, we find

Hr(a,
x
'
0, *>}
'
= -.—K-r
Java sin |_
dip
- 4s
60
(sin
v * r/
J
- (68)'
v

From H r we can determine the transverse electric component of the

entire field. We take the general flux function,
* = E a, R.{r)
*
T,(0, *), (69)

where the summation extends over a complete orthogonal set of charac-

teristic functions appropriate to the region under consideration. Using
equations 45, we have

favr* H r (r, B, <p) = E p(p + IK R>(r) T,(0, <p). (70)

Ut
H,(a, 6, v) = E A, T,{6, v ). (71)

By virtue of the orthogonality,

A, = N,' 1
ffH,(a, B, *>) T,(B, „) da,

(72)

N =
'
ff [T,{6 '

* Another solution of this problem is given in Problem 34 at the end of the book.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 1.11 TRANSVERSE ELECTROMAGNETIC WAVES 15

From equations 70 and 71, we have

" Kv + i) *.(«) '

(73)

Hence, we have the flux and the complete TE component of the field.
Incidentally we have also E$,te(o, Q> <p) and E ViT E(a, B, <p). We subtract
these functions from the given functions E e (a, 0, <p) and E v (a t 6, ip) and
obtain E 9 , TM (a, $, <p) and E+tTU (a, 6, <p).

Now, if we take a complete set of stream functions appropriate

for TM waves (including the TEM wave corresponding to v = 0), then,

II = £a,fl,(r) T,($, v),

(74)
jmrE = t E a, B/(r) grad T.(6, *>).

Using again the orthogonality properties, we have

a, R,'(a) |grad T,| 2 dU = jcoea

fjErffsd T, dU. (75)
jj
When v is different from zero, we can use equation 63 to simplify the
calculation of the square of the gradient. we can use When v = 0,
equation 60 to reduce the surface integral to a line integral. From
equation 61 we can obtain the integral of T2 by letting v l —*v2 = v.

1.11 Transverse electromagnetic waves

We shall now consider TEM waves in greater detail. If we start with
TM waves and assume that E T vanishes, we find that v = 0, and equa-
tions 26 for the potential and stream functions become

— - -»M n, = -joxV. (76)

Hence, V
and II are sinusoidal functions of fir. Equation 23 for the
T function becomes

sine — (sin*—j+^-y =0. (77)

Both V and II also satisfy this equation. On the surface of each per-
fectly conducting cone emerging from the origin, we must have
T= constant. (78)

The field components are given by equations 16 and 17. Since V

and n are sinusoidal functions of /3r, the amplitudes of all field compo-
nents vary as 1/r. Hence, the transverse voltages (the line integrals

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
16 SPHERICAL WAVES
of E along curves in spherical surfaces concentric with the origin)
remain finite as r approaches zero. TEM
waves are the only waves
for which this is true; for any other spherical wave whose source is
at the origin, the electric intensity approaches zero more rapidly than
1/r, and the voltages become infinite. Hence, an electric generator of
infinitesimal dimensions with a finite internal electromotive force produces
no external field in free space; and, in the presence of two or more cones
connected to its terminals, it produces afield which depends on the voltages
between the terminals of the generator but not on the details of the internal
field distribution. These ideal generators are approximated by small
generators, where the term "small" signifies that, if a is the radius of
the smallest sphere that can enclose the generator, then, /3a <^ 1.
Introducing a new independent variable

P = tan \B (79)

in equation 77, we obtain Laplace's equation in plane polar coordinates,

If we imagine a plane tangent at the north pole. 6 = 0, to a sphere of

unit diameter, then equation 79 gives a correspondence between points
(6, on this Neumann sphere* and points (p, >p) in the plane.
tp) The
corresponding points lie on the radii drawn from the south pole, = tt.

The corresponding solutions are given by typical functions of a complex

variable z, where
z = pe J * = e^tan \6. (81)

Hence, we have a correspondence between two-dimensional electrostatic

fields for planes and spheres, and for systems of parallel cylinders and
systems of cones emerging from a common apex. Likewise, for every
TEM wave on a system of parallel cylinders, we have a corresponding
TEM wave on a system of cones. In particular, circles on the Neumann
sphere correspond to circles in the tangent plane.
The potential V of a line charge of linear density q and passing
through a point z = z\ and the stream function n of the current J are
t

v= S log|z n -
i log 2l
" "I* (82)

These functions are also the potential and the stream function of the
radial filaments of charge and current passing through z = z\. The
point z = oo, which is the point at infinity in the plane and the south

* Applied Mathematics, p. 29.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
1.11 TRANSVERSE ELECTROMAGNETIC WAVES 17

pole 6 = ?r on the Neumann a singular point. For every

sphere, is also
straight filament of charge or current, there exists an equal and opposite
every radial filament of charge or
parallel filament at infinity, and, for
current, there exists an equal and opposite filament along the radius
$ — w. For a charge q and current I on the radius (0i, <p\) and the
opposite charge — q and current —I on (02, *pz), we have

where

di = \z - zi\ - [p
2 - 2p,p cos(*> - v>i) + 2
pi ]^

= [tan
2
\B~2 tan \0 tan \9X cos(*> - *>i) + tan 2 JtfJ«. (84)

The expression for di is similar. In the present case, the charge and
current filaments along = w cancel.
Consider now n +
1 perfectly conducting conical surfaces emerging

from the origin. Let h, In, and — Iv, In be the h— h

currents in these conductors. The last conductor may be called the
"ground cone." We are free to choose n so that it (and, hence, V)
vanishes on the ground cone. The stream functions on the remaining
conductors will be linear functions of the currents,

III = P11/1 + P12/2 + • * •

+ PlJn,
n2 = V2lh + P22/2 + h 7>2 Jn, (85)

II,, = p n] /l -h Vrtl* + • • 4" P-Jn.

Equations 76 are on a typical radius {6, ip), and, hence, on each

satisfied
conductor. Substituting from equations 85 and calling the correspond-
ing potentials of the cones Vj, V2,' V nt we obtain the equations for
the voltages and currents.
In particular, for two cones, we have

dr
1

-j»LI,
'
'
^= dr
-jaCV, (86)

where / is the current in one cone and V is the transverse voltage from
this conductor to the other. The series inductance L and the shunt
capacitance C, per unit length, are

L = mu pn
(87)

More generally the inductances and the potential coefficients of a system

of conical conductors are proportional to the corresponding p'a.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 18 SPHERICAL WAVES 1.11

To calculate L and
C, consider current / along the radius (7, 0)
and —I along the radius (y, ir). The stream function II is given by-
equation 83, where

di = (p
2 — 2cp cos <p + c2 )^, c = tan %y,
(88)
d2 = (p
2
+ 2cp cos <p -+• c 2 )« p = tan \9.

Let« = log(rf2/di); then,

p
2 - 2cp coth w cos tp + c2 = (89)

is the equation of the cone u = constant. It is also the equation of

its traceon the Neumann sphere and of the projection of the trace on
the tangent plane. If u = ui and u = 112 are two such conductors, then,

L .£ („,-«,), C-tJS^- (90)

Next we have to express u\ — m

in terms of the dimensions of the cones.
The cones u = u\ and u = W2 are circular. To show this, let us
start with two circular cones whose axes are 0) and (#2, jt). Let a
and be the cone angles (the angles between the axes and the generators
of the corresponding cones). Then, the equations of the cones are

cos cos #1 + sin sin t?i cos ip = cos a,

(91
cos cos #2 — sin
.
sin d 2 cos <p = cos 0.
Since

cose = 8ine =
TTf' TT^' (92)

equations 91 are readily transformed into the form 89, and reciprocally
equations of the form 89 may be transformed into the form 91. Thus,
we find

c coth ui = (cos a + cos sin di,

c coth U2 = — (cos + cos 02)" 1 sin #2,

c = (cos a — cos t?i)

w (cos a + cos t?i)~ w
= (cos - cos # 2 ) M (cos + cos 2 )~ H
t? .

The equation shows that, when 7 is assigned, the angles a, 0, #1, t? 2

last
are not independent. Simplifying the relationship between them,
we have
cost?! ^ cosa
COS t?2 COS '

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
1.12 DIPOLES AND SOLENOIDS; CAPACITORS AND LOOPS 19

From this equation we can obtain the

between ratios of the differences
the numerators and denominators to the corresponding sums, and then
we find

tan - $2 ) = tan J(a - 0) tan J(<* +W cot £0, (95)

where t? = t?i 4- 1?2 is the angle between the axes of the cones.
Hence, if we know the cone angles ol, & and the angle t? between
their axes, we can find the angles t?i, t?2 that these axes must make
with the bisector of the angle between the focal lines (7, 0) and (7, ir),
which is in the present case the z axis. From equation 93 we obtain,
successively, coth tti, coth 112, sinh ui r sinh U2, cosh ui, cosh U2,
cosh (wi*- U2), and, finally,

. . cos a cos — COS 1?

wi - u2 = cosh" 1
sin
:
^—z
a sin (90)J
v

It should be noted that each cone has two axes: 0), (?r — flj, *),
for example. A particular choice of the axis defines the interior of
the cone, in the sense that the axis itself is in the interior. In the
foregoing discussion we have chosen the axes and, hence, the cone
angles so that one cone is in the exterior of the other. If we choose the

axes and the radii so that one cone is in the interior of the other, then,
in equation 96 we should replace a and # by n — a and w — t?.

For coaxial cones, # = w, and

ui - u2 = log(cot la cot |/3). (97)

This expression, however, is more readily obtained directly from equa-

tions 82 if we let z\ = 0.

If a and /3 are small,

2 sin 2
Ul -u 2 ^ cosh-» , . \&
^ 01
2 log
2 sin \$
•
(98)

1.12 Dipoles and solenoids; capacitors and loops

As the frequency of oscillations approaches zero, electromagnetic fields
approach either magnetic fields. Reciprocally,
static electric or static
we can start with static fields and determine the corresponding electro-
magnetic fields. A simple pole, that is, a point charge Q, generates
the simplest electrostatic field: E r = Q/^wer 2 E$ = E v = 0. The ,

corresponding electromagnetic not simple, however, since the

field is

law of conservation of electric charge requires an infinite electric current

filament supplying charge to the pole or abstracting charge from it.

In fact, the simple pole in the nonstatic case may be considered as one
end of a semi-infinite chain of dipoles, with the other end at infinity.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
20 SPHERICAL WAVES 1.12

The two point charges, Q and — Q, an infini-

electrostatic field of
tesimal distance I apart and situated at points whose Cartesian coordi-
nates are (0, 0, JZ) and (0, 0, -§/) is

Ql cose - Ql sin

The product is the moment of the dipole. This field is independent

of the ip coordinate. When Q varies with time (by passing from one
pole to the other), there are radial displacement currente which, in
accordance with the Ampere-Maxwell law generate magnetomotive f

forces in circles coaxial with the dipole. Hence, we have the component
of the magnetic intensity. The variable field is thus a transverse
magnetic From equations 28 we see that T (0, <p) = 0(0) must
field.

equal cos 0, at least when « = 0, if we are to obtain equation 99. From

equation 37 we find that T is independent of w and that 0(0) — cos
satisfies equation 37 if v = 1. This, together with the boundary condi-
tion at infinity, gives R(r), except for an arbitrary constant. In this
way we obtain

w-w) trm *~*< (100)

„ = ^ ( 1 + -r^ ) I
'

As £r —* 0, equations 100 approach equations 99 if

Ae*"' = ja>QZ = Ilt (101)

where / is the current passing between the poles.
A point source of magnetoetatic field is provided by one end of an
infinitely thin and infinitely long solenoid whose other end is at infinity.
If $ is the magnetic displacement (or "flux") issuing from the open
end at the origin, then, H r = */4?r^r2 , Ht = H r
= 0. We have seen
that, except in the limiting case Max-
o> = 0, there are no solutions of
well's equations that are independent of the angular coordinates. Next
in the order of complexity is the field of a magnetic dipole formed by
an infinitely thin solenoid of infinitesimal length I,

Hr ~
2*jir» '
H$ ~
4r/ir» '
H* °' (1 °2)

This is the magnetic counterpart of the electrostatic field defined by

equations 99. The corresponding field at all frequencies may be ob-

Original from
Digilizefl byGoOglc UNIVERSITY OF MICHIGAN
 1.12 DIPOLES AND SOLENOIDS; CAPACITORS AND LOOPS 21

tained from equations 100 if we note that Maxwell's equations remain

invariant when we interchange p and e, and E and -H. Hence,

&( +-k-w) r~~ 1 Bin9

'
(103)

s '=-^( 1 + ir)^ + '" ,sin *-

As /Sr —* 0, equations 103 approach equations 102, if

Ae*«* = ju>i>l = VI, (104)

where V = ju* = d$>/dt is the magnetic displacement current through

the solenoid.
The dimensions of a three-dimensional structure may be made to
approach zero in different ways. In the electric dipole we start with
point charges: that is, with one limiting process already completed.
We can also start with charged particles of radius a, much smaller
than the distance I between them, and let both a and / approach zero,
either keeping their ratio constant or permitting a/l to approach zero.
In the first case the voltage between the charges is infinite, and in the
second it approaches infinity. But suppose we start with a capacitor
formed by two parallel plates of area S, an infinitesimal distance I

apart. If V is the voltage across the capacitor, the charge on the

positive plate is

Q = CV, C = y- (105)

The dipole moment of the capacitor is

Ql = eSV, (106)
and equations 99 become
SV SV sin
Er _
„ cos 6 , .

, E9 - 3
1 E9 _
-0. (107)
47rr

We shall call the product SV the area moment of the capacitor. Equa-
tion 106 gives the moment of the dipole producing the same external
field as the capacitor whose area moment is SV. Suppose now that,
starting with finite values of S and I, we let them approach zero in such
a way that the capacitance C remains constant. The external field of
the capacitor will approach zero with S at all frequencies. Thus, we
have one ideal element of electric circuit theory. For finite values of
S we have a "capacitor antenna."

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
22 SPHERICAL WAVES 1.12

The inductance of a thin solenoid is L = nn 2 S/l, where n is the

number of turns, I the length of the solenoid, and S the area of the cross
section. This can be kept finite as S and I are allowed to approach
zero by increasing n. Since * = plnS/l = LI/n, the magnetic dipole
moment #J = Lll/n approaches zero. Hence, the external field of an
infinitesimal solenoid of finite inductance vanishes. Thus, we obtain
another ideal element of circuit theory, an "inductor."
The field of steady electric current / flowing round the edge of an
infinitesimal area S, situated in the xy plane at the origin, is

=
SI cos B „
«•
SI sin e „ =
11 r
2yr3
' 4^3— '
/f„ 0. (108)

We product SI the area moment of the current loop. If

shall call the
Q is a charged particle revolving in a circular orbit with frequency /,
the areamoment is QfS. The external field of the current loop with
area moment IS is the same as that of the magnetic dipole (the solenoid)
with moment
*Z = pSI. (109)

Although the external fields of the solenoid and the loop are the same,
the internal fields are very different. The quantities that determine
the intensity of the field are also different. Thus, the magnetic flux
through a loop formed by an infinitely thin wire is infinite it depends ;

on the radius of the wire if the radius is finite. The external field
depends on the current and not on the flux. In an infinitely thin
solenoid the current is infinite and the flux is finite. If the cross section
of the solenoid is finite but small, the flux depends on the current, the
number of turns, and the area of the cross section; but it is the flux and
not the internal details that determines the external field. Of course,
if the solenoid is not infinitely thin, we have to make a correction for

the leakage of magnetic flux in the vicinity of the ends. This leakage
makes the solenoid effectively shorter, and it alters the field in the
immediate vicinity of the ends or "poles" of the solenoid. Similar
corrections have to be made in the case of a loop if the wire is not infi-
nitely thin. The effective area is reduced, and the field near the loop
is altered somewhat.
We should also mention toroidal coils. In the static case the field
is confined to the interior of the coil. In the a-c case there is an external
field. If the toroid is thin, its external field equals the external field of

a capacitor of area moment VS, where V is the magnetic current in

the toroid and S is the area of the loop.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 1.13 INFINITESIMAL ELECTRIC CIRCUITS 23

1.13 Infinitesimal electric circuits, small antennas, and trans-

mission lines
In the preceding section we have considered ideal circuit elements,
capacitors and inductors, from the point of view of Maxwell's theory.
They are obtained in the limit as the dimensions of the corresponding
physical structures become vanishingly small. Infinitesimal capacitors
and inductors (and resistors as well) possess at all frequencies the elec-
trical properties ascribed to them in ordinary circuit theory. A more
detailed study would show that actual physical structures approximate
the ideal structures when their dimensions are very small compared
with the wavelength A = c/f corresponding to the highest frequency /
under consideration.
Infinitesimal circuit elements may
be connected by infinitesimal
leads into circuits and networks which obey Kirchhoff's equations
provided the approach to the limit satisfies certain conditions. In
deriving Kirchhoff's equations from Maxwell's equations, we have to
make two assumptions: (1) The charge on each connecting lead is
negligible, and (2) the magnetic flux linked with each closed mesh is
negligible. Whether these requirements are satisfied or not depends on
the manner in which the dimensions of the leads approach zero. If
we let the radius of each lead approach zero while keeping the length
finite, the charge on each lead will approach zero, while the magnetic
flux linked with each mesh will increase indefinitely. If we keep the
radii finite and let the distances between the leads shrink, the magnetic
flux linked with each mesh will approach zero, but the charge on each
lead will increase indefinitely. We can realize intermediate cases by
imposing certain conditions on the ratios of various dimensions. In
order to satisfy both of the above-mentioned requirements, it is neces-
sary to maintain finite ratios between the radii of the leads and the
distances between the leads while letting all dimensions approach zero.
In this case the terminals of the various circuit elements and the leads
connecting them disappear from the equations, and Maxwell's equations
become Kirchhoff's circuit equations. Thus, if we pass to the limit
properly, the physical dimensions of generators, circuit elements,
and circuits themselves are eliminated, and we have to deal only with
the electrical properties: impressed voltages, currents in the various
branches, resistances, inductances, and capacitances.
At low frequencies most practical circuits satisfy the assumptions
implied in Kirchhoff's equations to a high degree of approximation;
but at high frequencies this is not necessarily the case. In theory,

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
24 SPHERICAL WAVES 1.13

where for the purposes of mathematical simplification some dimensions

are assumed to be vanishingly small, it is particularly necessary to
inquire whether the assumptions constitute a reasonable approximation
or a radical modification of the problem.
Small antennas are essentially small circuits in which the external
field of one element may dominate the field of the others. In a capac-
itor antenna the capacitance may be tuned out by an inductance whose
external field is negligible. The external field of such an antenna is

given by equation 96. To obtain the free oscillations of the capacitor

antenna, we replace ja by p — f + jv>,

*-^ + T? +
wW*- J
r) ,to '' »
At great distances from the antenna, Et = rjH^ By integrating the
Poynting vector over a large sphere, we express the radiated power in
terms of the moment A of the current in the capacitor. The damping
constant (— £) may be obtained by successive approximations. First
we obtain <o from the internal field of the antenna; then we calculate
the average loss of energy per second W on the assumption that £ = 0.
If 6 is the total stored energy, then £ = - W/2&.
Similarly, small loop antennas are loops tuned with capacitors
whose external fields are negligible.
Electric generators may
be located either in the immediate vicinity
of antenna terminals or at some distance from them. In the latter case
the energy is brought to the antennas by transmission lines. These
feeders and antennas form integrated radiating systems. After con-
sidering antennas separately from transmission lines, we must consider
the interaction between them. For this reason it is important to know
the ideal conditions under which the interaction between the local
circuits and transmission lines on the one hand and the antennas on the
other vanishes. We have already seen that infinitesimal circuits have
no external fields, and, therefore, do not interact with antennas. Simi-
larly, transmission lines with infinitesimal transverse dimensions do not
affect the behavior of antennas, and reciprocally are not affected by
antennas. First, let us consider a coaxial line formed by perfectly-
conducting coaxial cylinders. It is a limiting case of two coaxial cones,

Original from
UNIVERSITY OF MICHIGAN
1.13 INFINITESIMAL ELECTRIC CIRCUITS 25

with vanishingly small interna] angles. The characteristic impedance

for TEM waves is
K = £log?!> (HI)

where a and b >

a are the radii of the cylinders. This impedance
remains constant if a and b approach zero while b/a is constant. The-
oretically, it is possible to maintain any impedance; but, practically,
there are severe limitations. In the process of passing to the limit all
the higher modes disappear, and we have an ideal single-mode transmis-
sion line of the elementary transmission line theory. No energy can
pass through a perfect conductor; hence, the energy guided by the
coaxial line can escape it only through a hole, such as the hole at the
end of the line where the inner conductor is connected to one arm of the
antenna and the outer conductor to the other arm. The voltage across
the conductors of the coaxial line impressed on the antenna. The
is

same voltage is also impressed between the antenna arm connected to

the inner conductor and the outer surface of the outer conductor. How-
ever, in the limit there can be no current on the outer surface, assuming
of course that the energy delivered by the coaxial line to the antenna
is finite. Thus, if / is the current on the outer surface, the magnetic
intensity in the vicinity of the surface is = IJ2xp where p is the t

distance from the axis. The energy density is proportional to the

square of H 9 .Hence, the energy stored in the vicinity of the conductor
is proportional to log p and becomes infinite as p approaches zero unless I

also approaches zero. Thus, in the limit the coaxial pair impresses
a voltage on the antenna; but, otherwise, it has no effect on its per-
formance.
always some interaction between antennas and
In practice, there is

transmission lines feeding them, although it is customary to reduce

this interaction by means of specially designed "wave traps."
Another important type of transmission line is represented by a
pair of parallel wires. It is a limiting case of two diverging cones with
vanishingly small internal angles and vanishingly small angle between
the axes. In the case of equal wires, the characteristic impedance is

K = J-ooA-i^-l), (112)

where a is the radius and I the distance between the axes of the wires.
The impedance remains constant as long as the ratio l/a is kept constant
while a and I approach zero. In the limit the field at any finite distance

Original from
Digilizefl byGoOglc UNIVERSITY OF MICHIGAN
26 SPHERICAL WAVES 1.14

from the line vanishes. Such lines can be bent without affecting their
internal fields.
remains constant while a approaches zero, then
If I becomes K
infinite. Such a line qan serve as a source of infinite impedance supply-
ing a given current to a given structure. It is also possible to imagine
a parallel-wire transmission line in which both "wires" are cylindrical
cages consisting of an infinite number of infinitely thin wires, and choose
the infinities in such a way that the characteristic impedance of the
line is infinite and yet the fields external to the cages are equal to the
fields of smooth cylinders. The current in such lines is not affected
by the antennas which they may feed.

1.14 Biconical transmission linen

We have already noted the importance of transverse electromagnetic
waves for producing radiation. Without them an infinitesimal generator
cannot radiate, and a small generator can radiate very little energy
unless the internal fields are permitted to become very large. Antennas
guide the energy from the generator, or from the junction with the
feeder transmission In the vicinity of the generator
line, into free space.
(or the junction), TEM
waves are the only ones of importance. All
other waves are generated when the antenna wires connected to the
generator are suddenly discontinued at some distance from the gener-
ator. The waves on the voltage and current at the
effect of these
antenna terminals manifests itself through their effect on the TEM
waves. In the case of two (or more) infinitely long cones, the waves
must be progressive waves moving away from the generator. When
the cones are discontinued, there will appear a TEM wave moving
toward the generator, the wave "reflected" from the discontinuity.
As far as TEM waves are concerned, the discontinuity behaves as an
impedance discontinuity in transmission lines, and we can speak of the
"impedance (or admittance) of space as seen from the outer antenna
terminals."
Consider now TEM waves in a pair of cones diverging from the
origin. The current Io(r) in one of the cones and the transverse voltage
Fo(r) from this cone to the other obey equations 86, where the subscript
zero is used as a reference to TEM
waves. The general solution may
be expressed in terms of Fo and Jo at r = or at some other distance
r - I. Thus, it is easy to show that

V (r) = Fofl) cos fi(l - r) + jK I (l) sin fi(l - r),

(113)
/o(r) = /off) cos fi(l - r) + jK~* V (l) sin fi(l - r),

Original from
UNIVERSITY OF MICHIGAN
1.14 BICONICAL TRANSMISSION LINES 27

where K is the characteristic impedance, and is found from equations

90 and 96,
(L\X tj
u COS Ct cos — cos t?
Arr = , /3

\C J =
"77 ) -7T- COsh" 1
;
(114)
v '
2x sin or sin

If the outer ends of the cones happen to be at r = lt the admittance

y, = (H5)

is called the terminal admittance. It is the admittance of free space as

seen from the ends of the biconical transmission line. However, we can
always define an effective admittance (equation 115) at r = I, irrespec-
tive of the location of the discontinuity. Substituting Io(l) from
equation 115 in equation 113, we obtain

Kh(r) - V(l)[j sin 0(1 -r) + KYteoBfiQ - r)J,

V (r) = K(Z)[cos 0(1 - r) + jKY t sin 0(1 - r)].
The input admittance at r = is, then,

V Zo(0) y-i KY * cosj3/ + jsin g

'
K cos#+j7CK,sin# ' U1 ' J
(0)

It is also the input admittance r«(r<) at any other input sphere r = r*,

provided we interpret I as the difference I — The input impedance

is =
The electric be expressed in terms of
and magnetic intensity may
the voltage and current. If the cones are coaxial with the z axis, the
most direct method for obtaining the required expression is to use the
Ampere-Maxwell equation for a typical circle coaxial with the z axis
and the segment of the sphere with the center at the origin and passing
through this circle,

2rrmn6H 9 (r,e) = l'o(r), Hv = ^T> g (118)

Since Ee is the same function of $ as U9 , we assume Eg = A/sin 6 and

integrate along a typical meridian from one cone to the other to
it

obtain F(r). The quantity A is thus expressed in terms of V(r),

and we find

In the more general case of noncoaxial cones, the field is obtained

from the stream function (equation 83).

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
28 SPHERICAL WAVES 1.15

1.15 Quasibiconical transmission lines

In the case of two thin cones, equation 98 may be expressed in terms
of the radii a and b of the cones at distance r from the apex and the
distance d = 2r sin \$ between the corresponding elements on the axes
of the cones. Hence, we have the following expressions for the dis-
tributed parameters of the biconical transmission line

L-^Iog{d(r)[a(r)6(r)]-«}, C- (120)

Both L and C are independent of r, even though a, b, d are functions of r.

Let us now distort the cones into some other surfaces of revolution:
cylinders or spheroids, for instance. The magnetic lines will remain
circular,and the electric lines will approximately follow the meridians
except where r is comparable to a and 6. In this region we shall assume
correspondingly smaller deformations to keep the electric lines close
to the meridians. Then the field will vary with approximately in
accordance with equations 117 and 118, and, after appropriate integra- •

tion with respect to 0, we obtain the transmission equations 86 for our

"quasibiconical" line in which L and Care still given by equations 120.
Both L and C are now functions of r.
The approximate solutions may be obtained by the wave perturba-
tion method.* We shall summarize the results. There are two types
of approximation, depending on the rapidity with which L and C vary
with x = r — r it where r< is the radius of the input region. First let

fi(x) = w[L(x) C(*)]M K(x) - [Hz)!*

lC{x)]
(121)
*to =fQ 'f!(x)dz,
where V(t?), iC(t>) are V(x), I(x), K(x) expressed as functions
of # ; that is, V(t?) = V[x(&)], etc. Choosing v andi as the new depend-
ent variables and t? as the independent variable, we have

dv K' di
= . . K' . „„.
" -* " + Sic *•
(122)

If K'/2K is smaller, or at least not much larger, than unity, then,

v(») -*(#)+ viW+ *(*)+-,

»(*) = to(*) + (*) + *(*)
*'l
* " J

* Applied Mathematics, pp. 212-220.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 1.15 QUASIBICONICAL TRANSMISSION LINES 29

where
t>oW = A<T>* - Be**, i (d) = Ae~» + Be",

(124)

In particular, the functions,

io(0) = tr» v (9) = e-»

(125)
-
«"fw^*' •»« -
give one pair of approximate solutions i(#), i>(t?). The other pair is

given by t*(tf), —»*(£). If K'/2K is small compared with unity and

if is constant, we find the following approximate wave functions:

'*<*>-[•#§-]" <>+'•*<'*-''•

/-(x) = - ]* [1 - i* (126)
[^ff
= [K(p)K(xmi - jk{x)]r'>;
V~{x) = [K(0) K{x)\*[\ +jk(x)]e»:
The impedance Z(0) in tenns of Z(l) is

m - * (0)
m pm + j zd) p<m •
(127)

where
Pi (/JO = cos 01 + [*(/) + fc(0)] sin
P2 (#) = sin 01 + [k(l) - fc(0)] cos 01,
(128)
PsWO = cos 02 - [fcfl) + fc(0)] sin ft
PaW = sin 01 - [*(f) - A(0)J cosfl.

If A; (a:) is not small compared with unity, it is best to write equations

86 in the form,
dV
= -juLol - ML(x) - Lo]l,

(129)
«£ = _jw C F - j«[C(x) - Co]V,

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
30 SPHERICAL WAVES 1.15

where Lo, Co are some mean values of L(x) and C(x) in the interval
(0, 1) under consideration. Introducing new variables,

(130)

we have

—=- Jt - —= — — J f (»>,

(131)

If io(t?) = exp(— ji>), DoW = exp(— j'i)), then the second terms in
the series 123 are

iiW = - ije ȣ [fM + g(V )\ dv +

hje>* l/W - gMk-** <h,

(132)

viW = UM + gM\ cb -
hje»£ UM - gM)e-*« d*.

In particular, the mean values Lo, Co may be taken as the average

values L a C„ of L(x), C(x) in the interval (0,
, 1). If L{x) C(x) = const,
and the variations in L and C are relatively small, then f(v) g{<p) is +
a small quantity of the second order. In such a case, we have approxi-
mately

7m _ K Z(l) P (0l)+jK P
l a 2 (0l)
fm v

where

Pi(0l) = cos 01 - Ka- M(0l)

1
cos 01 + JCr 1
N(0l) sin ft
P 2 (#) = sin 01 - Ka ~ l
M(0l) sin 01 - K ~ a
l
N(0l) cos 01,
(134)
-K ~l
P 3 03Z) = cos 01 + tfo" 1 M(#) cos 01 a N(0l) sin 81,

Pt(0l) = sin 01 + K a ~i M(0l) sin 0l + /C a

_1
JV(#) cos#,

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 REFERENCES 31

where

M(fi) - p£ [K a - K(x)\ sin 2px dx, (135)

Niffl) = fljT [tf a - X (*)] cos dx.

In the case of thin antennas, K a is large, and Z{1) c^K a 2 /Z a ,

where Za = R a + jX a is a function depending on ffl. Neglecting the

M/K a N/K c R a/K a X a/K we obtain, from equation 133,
products of , , , a,

- (g. - jf) cos gZj

wK Ra sin & + j[(X a - N) sin gj
z m m ^"
, ,

[(K + Jlf sina + (X + JV) cos 01) -

) cos 02 a
} V

REFERENCES
1. Julius Adams Stratton, Electromagnetic Theory, McGraw-Hill, New York, 1941
2. S. A. Schelkunoff, Electromagnetic Waves, D. Van Nostrand, New York, 1943-

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 2

MODE THEORY OF ANTENNAS

2.1 General mathematical formulation of the antenna problem

The simplicity mathematical formulation of the antenna problem
of the
contrasts sharply with the complexity of its solutions. Figure 2.1 shows
an antenna; the dotted line indicates an input boundary Si which, in
the case of a transmitting antenna, is a somewhat arbitrary surface
separating the source of power from the rest of space. Outside this
boundary we have an electromag-
,Sle netic field satisfying Maxwell's
t
I~=D f
— J equations. Inside the boundary
we may have moving conductors,
Xa ' a tennft the " in P ut
clouds of interacting electrons,
F ?
boundary" Si encloses a source
'

power of
sources of heat, etc., and Maxwell's
in the case of a transmitting antenna and
a sink of power in the case of a receiv- equations have to be supplemented
ing antenna. by dynamical and thermodynami-
cal equations. In the case of a
receiving antenna, Si encloses a sink of power; this "load" may also be
complex. The electromagnetic fieldproblem concerns only the region
outside the input boundary ; the conditions inside Si are replaced by
the boundary conditions over Si.
The problem of the transmitting antenna in free space may be
formulated as follows:

Outside the input boundary the

1. field in the antenna and around
it must satisfy Maxwell's equations.
2. The components of the electric and magnetic intensity tangential
to the surface of the antenna must be continuous.
3. The field must be
everywhere except under certain limit
finite
conditions. If the input region is infinitesimal or if the antenna is per-
fectly conducting and its radius is zero, certain infinities are permissible
32

f~* Original from

D g,t.zed
( by ^OOgie UNIVERSITY OF MICHIGAN
2.1 THE ANTENNA PROBLEM 33

derived from the conditions existing under

the nature of these infinities is

normal circumstances by passing to the limit.

4. At must represent a wave moving away from
infinity the field
the antenna; that is, the field must vary as r~ f{ct — r), where r is the l

distance from some point around the antenna. In the monochromatic

case the field should vary as r
_1
exp — 0r).
5. At the input boundary the tangential component of the electric
intensity is given.

The second condition is necessary because, owing to the difference

inthe electromagnetic constants of the antenna and free space, Maxwell's
equations are discontinuous at the surface of the antenna. The equa-
tions have to be solved separately for the region occupied by the antenna
and for free space the solutions must then be connected. The second
;

boundary condition gives the connecting equations, and it may be de-

rived either from the integral form of Maxwell's equations or from the
differential form. In the latter case we should start with an assumption
of a rapid but continuous change in the medium in the "boundary
layer" separating the antenna from free space, and then pass to the limit
of a boundary layer of zero thickness.
In the case of the receiving antenna the fifth condition is replaced by

5. The ratio of the tangential components of electric and magnetic

intensity at the input boundary is given.

Sometimes assumed for mathematical convenience that the wave

it is

impressed on the antenna is a plane wave. In such a case the fourth

boundary condition should be replaced by

4. At infinity the difference between the total field and the primary
or impressed field must represent a wave moving away from the antenna.

That is, only that part of the total must satisfy the original bound-
field

ary condition at infinity which is produced by the currents and charges

in the receiving antenna.
A substantial simplification occurs in the idealized case in which the
antenna is assumed to be perfectly conducting. In this case the second
boundary condition becomes

2. The tangential component of the electric intensity must vanish

at the surface of the antenna.

From the practical point of view, this idealization is not objectionable.

Antennas are made from good conductors, and the effect of their finite
conductivity can be evaluated from the solution of the idealized case.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
34 MODE THEORY OF ANTENNAS 2.2

2.2 Two general methods of approach to the solution of the

antenna problem
The antenna problem is primarily a "boundary value problem." To
obtain the most general solution of Maxwell's equations is easy to find ;

that particular solution which satisfies the various supplementary con-

ditions is far more difficult. The problem is greatly simplified when the
surface of the antenna happens to coincide with a coordinate surface in
a coordinate system in which Maxwell's partial differential equations are
"separable" into ordinary differential equations. The "separation of
variables" is possible in only a few coordinate systems, and there is only
one such coordinate system, the spheroidal, in which there is a family of
coordinate surfaces with finite dimensions. For this reason electric
oscillations on spheroidal conductors have been widely studied as we ;

shall see in the next chapter, the major problem in this case is the evalua-
tion of the appropriate spheroidal functions.
The method of separation of variables is applicable to other than
spheroidal antennas if we relax the requirement that the antenna sur-
face coincide with an entire coordinate surface, and require only that it
can be subdivided into parts coinciding with parts of coordinate surfaces.
For example, conical antennas are bounded by portions of cones and
spheres. In this case we look for solutions which individually satisfy
only some of the boundary conditions, but which can be combined so
boundary conditions.
as to satisfy all the
Both the spheroidal and the conical boundaries may be "perturbed"
into other shapes, cylindrical for instance. The solution can be per-
turbed correspondingly, and the variety of shapes that can be treated
by the method of separation of variables is thus greatly increased.
The functions associated with conical boundaries are much simpler
than those associated with spheroids, and in their case the method of
perturbation is more practical.
We have already stated that the complete solution may be expressed
as the sum of solutions satisfying some, but not all, of the boundary
conditions. In spherical coordinates each term happens to represent
a relatively simple wave, a mode of propagation consistent with the
electromagnetic laws and certain boundaries. The entire analysis is
very similar to that used in the case of waveguides. Thus, we find a
mathematical justification of an intuitively evident connection between
antennas and transmission lines, for a waveguide differs from an ordinary
transmission line only in that it permits many modes of propagation
instead of just one mode. Waveguides, antennas, and even free space
may be described as multiple transmission lines. In the preceding

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 2.3 ANTENNA ANALYSIS IN SPHERICAL COORDINATES 35

chapter we have found that each mode of propagation may be described

by a simple conventional transmission line diagram. For this reason
the antenna theory that uses these concepts may conveniently be called
the mode theory of antennas. * On the other hand, in spheroidal coordi-
nates the various terms represent certain modes of forced oscillations
and exhibit an affinity between antennas and resonators. Thus, the
theory of spheroidal antennas, as developed in the next chapter, is an
example of the resonator theory of antennas.
Another approach to the solution of the antenna problem is based
on the fact that a differential equation together with the boundary
conditions may often be converted into an integral equation, that is,

an equation containing the unknown function under the integral sign.

In the case of antennas the unknown function represents the antenna
current, or more generally the density of the current. Some integral
equations for antennas are obvious generalizations of the Kirchhoff
equations for electric networks, and for this reason the antenna theory
based on them may conveniently be called the circuit theory of antennas.
In this theory our attention is concentratedon the current at various
points of the antenna, and the field around the antenna is kept in the
background, whereas in the mode theory the field is prominent, and the
antenna current is in the background. The mode theory of antennas
clarifies the relationship between antennas and ordinary transmission
lines the circuit theory of antennas clarifies the relationship between
;

antennas and ordinary electric networks. Physically both theories

present the same facts from different points of view. The circuit
theory of antennas as developed by Hallen will be presented in Chapter 4.

2.3 Antenna analysis in terms of spherical coordinates

If the dimensions of the source of power are small compared with X, the
waves excited by be spherical or substantially spherical. Hence,
it will

it is particularly convenient to formulate the antenna problem in terms

of spherical coordinates. Consider, for example, a symmetric antenna

(Fig. 2.2) with a source of power at the center. We surround this

In the first paper expounding this theory, the author called it " transmission
*

theory " to stress the clarity with which this theory explains transmission of energy
from the source along the antenna into free space. Subsequently, some authors
called it "transmission line theory." The inclusion of the word "line" altered the
intended meaning of the phrase and tended to create an impression that this theory
treated antennas from the point of view of ordinary transmission line theory. To
avoid this possible implication Aharoni in his book Antennae suggested the name
" wave guide theory of antennae " (p. 86), which is a good name for the present
theory. Our new name " mode theory of antennas " has the virtue of being some-
what shorter.

Original from
UNIVERSITY OF MICHIGAN
30 MODE THEORY OF ANTENNAS 2.3

source by a spherical input boundary St . We assume that we know

the electric intensity tangential to Si. Eventually we shall find that,

for sufficiently small sources, we need not know the precise distribution
of this intensity but only the voltageimpressed by the generator on the
input boundary. If X is large, say, of the order of 50 or 100 meters,
the source or generator region may be taken large enough to include the
tuning coils or capacitors and even
It)
more elaborate matching networks.
Since the dimensions of these net-
works are small compared with X,
(2)
/
on the antenna input
their effect
I
/
voltage may best be calculated
I
So
i from elementary quasistatic the-
ory. Later we shall see how these
methods fit into the framework of
electromagnetic theory.
\ Next we introduce the an-
tenna output boundary sphere So
passing through the outer ends of
FlG. 2.2 Subdivision of the space
the antenna arms. This boundary
around a symmetric antenna into (1)
sphere separates the free space re-
the free-space region, (2) the antenna
region, and (3) the generator region. gion from the antenna region 2
1

which is bounded in the interior by

the input boundary sphere Si. Our terminology implies that we shall
consider specifically transmitting antennas. In view of reciprocity the-
orems, the results will be general.
When the medium surrounding the antenna is divided into regions
as in Fig. 2.2, we automatically separate the lateral surface of the
antenna, which is in from the ends, which are
the antenna region 2,
parts of the output boundary surface So. Hence, the boundary condi-
tion at the surface of the antenna is now replaced by three separate
boundary conditions:

1. In the antenna region the tangential E must vanish at the

surface of the antenna.
2. The field in the antenna region must join continuously the
field in the free space region.
3. The field tangential to the ends of the antenna must vanish.
The second and third conditions are really parts of one condition at
the complete closed surface So. If the antenna is circularly symmetric
and is so energized that the currents flow along its length, the only non-
vanishing components of the field are Er ,
E$, H9 . When the transverse

„. .. C
Digit** by <^OOgie
^,-,,,1^, Original from
UNIVERSITY OF MICHIGAN
2.3 ANTENNA ANALYSIS IN SPHERICAL COORDINATES 37

dimensions of the antenna are small, these components are still the
most important ones, irrespective of antenna symmetry.
For a biconical antenna (Fig. 2.3) the boundary conditions may
be expressed as follows:

1. At infinity,

Bh H,*±—- (1)

2. In the antenna region,

BrW = 0, E r (T ~ H) = 0, (2)

where 6 — ¥ is the equation for the surface of the upper cone.*

/
/
i
i

Fig. 2.3 A symmetric biconical antenna.

3. At the output boundary So the field is continuous:

Ee(l + 0) = - 0),E,(l 4> < S <x- V,

H„(l + 0) = HS - 0),
(3)
Erd + 0) = E {l - 0), r * < 6 < IT ~ &
S«tf-f-0) = 0, 0<d<$, and jt - * < e < v.

The second and third of these equations are not independent, and we
may use either, as suits our convenience.
4. At the input boundary Ee is given.
5. In the free space region 1, the field must be finite.

The need for the fifth condition does not become obvious until we
actually study solutions of Maxwell's equations, and find that the
* If the lower cone angle ia from the upper, the second equation
different

Iwcomes Er(* ~ fa) ™ 0, with corresponding changes in conditions 3.

(~* Original from

,^
Digged by ^.OOglC UNIVERSITY OF MICHIGAN
38 MODE THEORY OF ANTENNAS 2.3

general solution in region 1 involves more arbitrary constants than

can be determined from the first four boundary conditions. The
analysis of these solutions shows that except those in a certain set
all

become infinite for one or more values of 6, and for that reason have to
be rejected.
Condition 4 implies that the transverse voltage in the input bound-
ary between the cones is given,

Vi =£"* E 9 rd6. (4)

This "input voltage" isnot sufficient to define completely the field

around the antenna. We must know its distribution over the input
boundary. The study of solutions brings out, however, the following
facts: (1) If the dimensions of the input boundary are very small
compared with the wavelength, the details of the voltage distribution
do not affect the antenna field significantly except near the input bound-
ary; (2) the effect always be associated with the generators and
may
networks in the input region; (3) because of the small dimensions of
the input region, this effect may be calculated by solving appropriate
electrostatic and magnetostatic problems; (4) under many practical
conditions, the effect is negligible.

No difficulties are encountered if the dimensions of the input region

are assumed to be infinitely small, in which case equation 4 represents
the fourth boundary condition completely. This enables us to separate
the antenna problem from the local circuit problem. This separation
is possible even if the system is such that the interaction between the
circuits and the antenna is substantial. It is only when the dimensions
of the input region become large that no useful purpose is served by the
separation between the antenna and the local circuits the interaction ;

becomes so great, the various parts of the system become so well inte-
grated, that a recognition of these facts can only simplify the problem.
The preceding formulation of the antenna problem applies specifi-
cally to forced oscillations. The boundary conditions for natural
oscillations differ from the foregoing conditions only in equation 4.
Thus, if the apices of the cones are short-circuited, this particular
boundary condition becomes

E,rde->0 as r->0. (5)

On the other hand, if the apices are insulated, then the current vanishes
at A, B, and
rH 9 {f)->0, as r — 0. (6)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.4 FIELD AROUND A BICONICAL ANTENNA 39

The main between thin biconical antennas and thin

difference
antennas of other shapes is expressed in the principal wave, which
constitutes only one term in the general expression for the field in the
antenna region. Inasmuch as we have already developed a theory
of principal waves (Chapter 1), our main problem in this chapter is
the development of a theory for biconical antennas. Having solved
this problem, we can readily generalize the results (Section 15 of this
chapter) for antennas of arbitrary shape.

2.4 General expressions for the field around a biconical antenna

When theindependent of the ip coordinate, Maxwell's equations,
field is

which normally involve six interdependent components of E and H,

break up into two separate sets of three equations each. One of these
sets involves E r Ee, H?, and the other contains only
, r H§, E,. In H ,

the first case the currents in the conical surfaces are strictly radial since
they must be perpendicular to H*\ for the same reason the currents in
the end caps flow along the meridians. In the second case the currents
circulate around the conical surfaces. We are interested in the first
case,which is a special case of the TM waves considered in Section 1.5.
Thus,

Er ~.
n(n + \) R(r) 0(0)
'

(7)
Ej$ — :
i dR do
3 -jT- i tl - — r de
j— >

jusr dr dB * r dd

where n is an arbitrary parameter and R(r), 0(0) are solutions of the

following differential equations,

sin $ + cos ~+ n(n + 1) sin 6 6 - 0, (8)

d*R
(9)

The general solution of equation 9 is

R(r) = A m Jn n (pr) + B n Nn n (fir), (10)

where Jn n (fir) and Nn n (fir) are the normalized Bessel functions of

order n defined by*
Jn n (fir) =(\*fir)*J +y2 {fir) t i
Nn n (&r) = (\wfir)* N H (0r). (11)

If n is not an integer, the general solution of equation 8 isf

0(0) = Cn P»(cos 0) +D n P n (-cos 0). (12)

* Applied Mathematics, pp. 396-397.
t Ibid., p. 418.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
•40 MODE THEORY OF ANTENNAS 2.4

The first solution reduces to unity when = and is logarithmically

infinite when = tt ; the second reduces to unity at = r and is loga-
rithmically infinite when = 0. If n is an integer, the two solutions
differ only by a constant factor ( — )" and are finite for all values of 0.
In this case the second independent solution is Q,(cos 0) ; this solution
is logarithmically infinite at $ = 0, x.
In the antenna region E r must vanish on the conical boundaries
6-01 and = 02- In view of equations 7, this condition is satisfied
either when
n = (13)
or when
e(*) = o, efc) = o. (14)

When n = 0, the radial component vanishes for all 6, and the waves
are transverse electromagnetic. This case has been fully considered
in Chapter 1.

From equations 12 and 14, we have

C n J\(coa 0i) + D n P„(-cos 0i) = 0,

(15)
Cn P«(cos 2) + D n P„(-cos 2) = 0.

Solving for the ratio D n /C n , we find

Z>n P,.(COS0,) Pn(COS0 2 )

(16)
Cn Pn(-COS0l) Pn(-COS02 )
'

hence, n must be a root of the following characteristic equation

P a (cosfli) P n (-cos02 = ) P»(ix»h)P«(-eoB*i). (17)

In the case of a single cone = 0i, we have to satisfy only the first
condition 14; the second condition is then replaced by the requirement
that the field must be finite. Thus, if the field is in the region 6 < 6\,

O{0) - C n P n (cos0), (18)

since P„(-cos0) is infinite at - 0. Similarly, if the field is in the

region > 0i,
e(0) = DnP n (-cos0). (19)

In the first case (equation 18) n must be a root of

P n (cos0,) = 0; (20)

and in the second case (equation 19) we have

P n (-cos0i) = 0.
(21)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.4 FIELD AROUND A BICONICAL ANTENNA 41

In free space the field must be finite at = 0, ir. This condition

is not satisfied by any solution when n is not an integer hence, n must ;

be an integer, and

99) = C„P„(cos0), n = 1,2,3, (22)

since Q n (0) is infinite at = 0, v.

In the case of two equal cones (Fig. 2.3), $i = and 02 = *-
Equation 17 becomes
[Pn(cos *)] 2 = [P n (-cos *)]»; (23)

therefore, we have either

P n (cos*) = P.(-cos#) (24)

or
Pn(cost^) = -P„(-cos^). (25)

In the last case D n = Cn , and

9(0) = L\' L n (cosd) t
(26)

where L n (cos 0) is an even Legendre function of order n defined by*

Men 6) - i[J\(cos 0) + P n (- cos *)]. (27)

In the first case Dn = — C„, and

e(«) = C.'M.fcasf), (28)

where JW„(cos0) is an odd Legendre function*

M n (cos0) = |[P.(eofl0) - Pn(-cos^)]. (29)

The derivatives,

-~ L B (cos0) = -§[P n '(cos0) - P„'(-cos0)] sin*,

(30)
M„(cos 0) = -£[P„'(cos 0) + P n '(-cos 0)] sin 0,

are, respectively, odd and even functions. Hence, if the tangential

electric intensity impressed at the input boundary is a symmetric (even)
function of 0, the field in the antenna region will be given by a series
of odd e functions (equation 28). The even functions (equation 27)
will appear in the case of an unbalanced or shunt type of feed.

*Applied Mathematics, p. 420. In the first paper on antennas by the author

{IRE Proc., September 1941), and in the paper by P. D. P. Smith {Jour. Appl
Phys. t January 1948), /-.(cos 9) was used to denote the odd function now called
jW„(cos 0). It is hoped that the subsequent change in notation will not cause mis-
understanding.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
42 MODE THEORY OF ANTENNAS 2.4

The balanced type of feed is the most important case from the
practical point of view, and
be used in the subsequent details of the
will
solution. The general method, however, is equally applicable to other
cases. In all cases we can write general solutions in various regions
bounded by concentric spheres as infinite series of solutions of the form
given in equations which the 9 functions are the characteristic
7, in

functions obtained from the boundary conditions. With each 9 func-

tion there will be associated an R function (equation 10). The addi-
tional arbitrary constants are determined from the boundary conditions
at the spherical boundaries separating the various regions. Thus, for
the symmetric biconical antenna (Fig. 2.3) so energized that the currents
in its arms are equal and similarly directed at points equidistant from
r = 0, the field in the antenna region is given by

r2E
'=-~2^^W) M ' {COSe)
'
(31>

rH < ~
2* sin 6 + 27 ? »(n + 1) XPT d* ^ (C0B $h (32)

rE'~
2^ sing + 'ftr"? n(n+l) S.W W^ (c°90) '

(33)

K7 (r) = V(l)[j sin 0(Z - r) + KY A cos - r% (34)

F(r) = F(/)[cos 0(1 - r) + jXK, sin 0(1 - r)] p

(35)

K - log cot £f,tfr) - + p„ JVn.OJr), (36)

where the summations are extended over all the zeros of iW n (cos ^).
Equations 34 and 35 are general expressions for the conduction current
in the cones and the transverse voltage (the integral of E» between the
cones along a typical meridian) associated with the TEM waves (Section
1.14). In anticipation of the forthcoming proof that the complementary
waves contribute nothing to the transverse voltage, the subscript
indicating the TEM
wave is not used in connection with this voltage.
The constants of integration in equations 34 and 35 are the voltage
V(l) at the output boundary So and the terminating admittance

y. = •
(37)

This is the admittance "seen by the TEM wave" at the output bound-
ary. The factors Sn (fil) in the denominators have been included for

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 2.5 Antenna current and transverse voltage 43

convenience in satisfying the boundary conditions at r = I. In the

general expression 10 for the radial wave function we have substituted
A n — a n and B n = p n a n in order to separate one set of constants a„,
which will be determined from the boundary conditions at the output
boundary sphere r = I, from the other set p n which will be determined
,

from the conditions at the input boundary. If the radius of the input
boundary is zero, we must have

Vn - 0; (38)

otherwise, not only the field intensities Ee would tend to infinity

as r approaches zero but also their integrals. The latter represent
currents and voltages and should be finite.

In the space outside the output boundary, we have a similar set

of equations, except that there are no terms corresponding to TEM
waves. Thus,

b
*raf p * (C08fl)
'
(39)

Bk<fir)
2t t-iXi.... MJe+l) RkW 4r
to
P*« o), (40)

*" i 1
.-5,... WTTT W- i " '<« <">

Ruiflr) = Jn k (0r) + qk Nnk {&r). (42)

When the medium outside the output boundary is homogeneous,

qk = -j, and

R k (0r) = Hn k (0r) = Jnk (fir) - j Nn k (pr). (43)

These radial functions become proportional to exp(—jfir) as r increases

indefinitely.

2.5 Antenna current and transverse voltage

The" current in the upper antenna arm is

I(r) = 2*-r sin * H v (r, (44)

We shall represent this current as the sum of the principal current J (r)
associated with the TEM waves and the complementary current I{r)
associated with all other waves,

I(r) - J (r) + J(r). (45)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
44 MODE THEORY OF ANTENNAS 2.6

The principal current is given by equation 34; the complementary

current is found from equation 32,

We define the transverse voltage V(r) as the line integral along a

typical meridian from the upper antenna arm to the lower,

V(r) =J*~*rE,dB. (47)

This integral vanishes for every term in equation 33 except the first.
Hence, only the TEM
waves contribute to the transverse voltage, and this
voltage is given by equation 35.
It should be noted that the transverse voltage as here defined is
not equal to the difference of the scalar electric potentials at the two
ends of the meridian; it includes a contribution from the dynamic
component of the electric intensity (that is, from the vector potential).
In the case of the biconical antenna, the transverse voltage is seen to
be sinusoidally distributed; hence, for other antennas, including the
cylindrical antenna, it is not distributed sinusoidally. On the other
hand, the scalar potential is distributed sinusoidally on the cylindrical
antenna but not on others.

2.6 Fundamental theorem

The input admittance of the antenna is

** }
Yi ~ T(rT " ~VW + -V(rJ {

Thus, we have a fundamental theorem: The input admittance of a sym-

metric biconical antenna equals the sum of two admittances. The first of
these admittances equals the admittance of a uniform transmission line
terminated into an appropriate admittance Y The second admittance t.

represents the local effect at the input terminals, and it vanishes as the
dimensions of the input region vanish.
From equations 34 and 35, we have
KY, cos 3{t - r.) + j sin - r.) I(r.)
KY (S{1
- r.) r *">
.

*
cos r.) + jKY t sin 0(1 F(r<)
" K

If r4 - 0, then,

KYi
_ KY, cos 01 + j sin (50)
cos 0l+jKY,sin&l
*

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.8 AUXILIARY FORMULAS FOR THE ADMITTANCE 45

Introducing in equation 50 the impedance Za seen at the distance

X/4 from the end of the line (this impedance is called the inverse radiation
impedance),
Z a = K*Y ti (51)

and taking the reciprocal, we find

K K cos fil + jZ a sin 01 .- Za - jK cos gj

sin 01 , .
7 K
'
Z„ cos 01+jK sin 01 Kmnfil— jZa cos fi ' }

2.7 General formula for the terminating admittance

When r = I, equation 32 becomes

''M-T^+iz-s^i"-*"* (53)

Integrating from 6 =$ to 6 = w — we find that all terms on the

right except the first disappear. Thus, we find

KY = t I B) dd. (54)
-yjijfj'*
Since #„ must be continuous at the output boundary sphere, it

is also given by equation 40. Substituting in equation 54 and integrat-

ing, we find

^^twmL. Turhr ^- p (55)

2.8 Auxiliary formulas for calculating the terminating ad-

mittance
The terminating admittance, normalized with respect to the character-
isticadmittance l/K, has thus been expressed in terms of an infinite
sequence of coefficients bi, 63, bs, "• The a„'s do not enter equation
-

55 directly, but in general they also have to be determined since the

t wo sets of constants are interdependent. Notable exceptions are two
extreme cases: tKn~ biconicaT antennas in which ^ is very small,
(1)
(2) spherical antennas in which $ is nearly equal to tt/2. In these
cases the limits approached by the 6's can be found rather easily. In
fact, in the case of thin antennas there are two methods by wliich we
can find Y even without obtaining the 6's. The solutions in the two
t

extreme cases may be taken as the first terms in sequences of successive

approximations to the solution of the problem for intermediate values
of the half cone angle ^. Thus, starting with the already determined
limits of the &*'s, we may use one of the boundary conditions at the

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
46 MODE THEORY OF ANTENNAS 2.8

output boundary sphere to evaluate the a n 's and then the other boundary
condition to re-evaluate the 5*'s; then the cycle is repeated. In order
to carry out these successive approximations, we need certain auxiliary
formulas which we shall now derive.
Substituting r - I in equation 32, we obtain

* < < X-
where n is a typical zero of 3/„(cos Similarly, from equation 40,

To simplify the form of the corresponding expressions for E*(l, 6),

we introduce the radial wave impedances at the output boundary for
the various modes of propagation,

(59)

In particular, when region 1 is homogeneous, q k = —j, and

(60)

Similarly, if the input region is mhnitesimal, p» = 0, and

z- - (61)

At the output boundary may thus be expressed either as

or as

< 3 < ir -
= 0, 0<K* or ir-*<Kf; (63)

and the two expressions must be equivalent.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.8 AUXILIARY FORMULAS FOR THE ADMITTANCE 47

Using equation 56 and the orthogonal properties of Legendre

functions and their derivatives,* we can express the a„'s in terms of
///„ and, therefore, in terms of the 6„'s. Thus,

2xn(n + 1) (///„) sin B JU B (cos B) do

an = — (64)

M n (cos B) sin $ dB
J
Substituting the expression for IH 9 in terms of the 6's, we have

an = Z u nkbk,

u nk =
n(n +
- —
sin 6
-f-
M.(eos 0) -jjp P*(C06 9) dd

(65)

k(k+
1)JJ
M n (cos0)j smB dd

Integrating the integral in the numerator of equation 65 by parts,

we have

£~* sin B (cos 0) P 4 (cos B) dB

= sin B M n (coa 0) -~ P*(cos 0)

£ *
M n (cos 0) -~ [sin B -~ P k (cos 0)] dB. (66)

The integrated term vanishes because M„(cos ^) and n (— cos^) M

vanish. The second term may be simplified since equation 8 is equiva-
lent to

sin = " n(n + l) sin * G (67)

•
4b ( °is) '

and P*(cos B) is its solution when n = h. Thus,

*~*
M
f sin -j^ n (cos 0) P*(eos 0) dB

= k(k + l)fj~* M R (cos B) P*(cos B) sin B dB. (68)

* Applied Mathematics, p. 431.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
48 MODE THEORY OF ANTENNAS 2.8

The value of the last integral may be found elsewhere,*

•
Jnr dP * r>
<1M»~\ r -*
sin

J*
M * (cos 0) P 4 (cos $) Bin Odd
n(n+l) - k(k+ 1)

-
.oi + „1h> + .) ^r"" m
Similarly, for the integral in the denominator, we find

f *

^
[4r M.(cos9)Tsin9<» = n(n+ 1)
J ^
f'~* [Jtf.(cos sin 6 dB,

(70)

2n + 1 a* an

If ^ is not too small, the last integral may

be evaluated more simply
by substituting an approximate expression for JW»(cos0). Thus, we
have evaluated the u coefficients in equation 65,

,
2n+ 1 P*(cos »)
(72)
nk
k(k+ 1) - n(n+ 1) dM a (coa*)/dn
y }

Using equation 62 and the orthogonal properties of the Legendre

functions, we can express the 6's in terms of lE t then, by substituting \

from equation 63, we can find the 6's in terms of the a's and V(l). Thus,

= r{2
bk
z£ f ' 8in 6
W Pk(cos 9) *' (73)

Substituting from equation 63, we have

fe(fe+l) 2 fc+l
n(n+l) *(*+l)-n(n+l) X u
V* n }

rfM w (cos ^)
sin ^ P*(cos ^)

Equations 65 and 74 are linear algebraic equations defining the

coefficients a„, 6* in terms of 7(0- If these equations are solved, the

* Applied Mathematics, p. 425, eq. 39.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.9 SYMMETRIC SPHERICAL ANTENNAS 49

6's can be substituted in the expression 55 for the terminating admit-

tance normalized with respect to K
V(l) cancels out. ;

The number of equations may be reduced in half by eliminating

the a's from equations 74 with the aid of equation 65,

{2h
6, = - V}\ V(l) P*(cos *) +E E |V «^u.J> a . (75)

2.9 Symmetric spherical antennas

The symmetric spherical antenna a
a biconical antenna
is special case of
with nearly equal to ar/2; it is
yfr
a spheroidal antenna of zero
also
eccentricity. As we shall presently see, the complementary waves
in the antenna region may be neglected when the length of the gap
between the hemispheres, s = l(r — 2^), is small compared with X/2.
Setting the a„'s equal to zero in equation 74, we have

K-- JgffiZffl P-W), (76)

where the radial wave impedances are given by equation 60. Sub-
stituting in equation 55 and then in equation 51, we have

Za = K2 Y t
=
J- t -,E,...WTT)
[P ' (cosW ^
The normalized values of the radial wave conductances and susceptances
for the first sevenmodes of propagation are shown in Figs. 2.4 and 2.5.
If the length of the gap between the hemispheres is small compared

with X/2, all the higher order waves in the antenna region are highly
attenuated because, even near the gap which is the widest part of the
region, the cutoff wavelengths for these modes are 2s, 2s/3, 2s/5, etc.,
and hence much smaller than X. Consequently, these waves are con-
centrated in the vicinity of the gap, the reactive energy associated with
them is and their effect on Y must be small.
small, t

This is an intuitive deduction from our knowledge of waveguide

theory. We may also use a more detailed mathematical analysis to
arrive at the same conclusion. When ^ is not too small and not too
near jt, we have the following approximate formulas for the Legendre
functions,*
* Applied Mathematics, p. 419.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
60 MODE THEORY OF ANTENNAS 2.9

'

_ _
2.0 •

i.e

1.6

M
kl f
1
I

1.0
J I
0.6

0.6
/ —^—
0.4

02

PHASE DISTANCE, /3r

Fio. 2.4 The product of the intrinsic impedance * and the radial conductance G m*
of progressive spherical waves in free space as a function of the phase radius 0r.

2.0

Fig. 2.5 The product of the intrinsic impedance ir and the radial susceptance B .+
of progressive spherical waves in free space as a function of the phase radius $r.

^ ^ sia '
«»*)
^psin*)H
(78)
1

Therefore,

*-^*> (J^sin*)* (79)

Original from
DigilizeO byGoOglc UNIVERSITY OF MICHIGAN
 2.9 SYMMETRIC SPHERICAL ANTENNAS 51

The roots of this equation are

q(\ir - i/<) = mx, m = 1, 2, 3,-

-

»
These roots approach infinity as ^ approaches x/2. For re we have

n=- i +V5^i--i+ g --i- 1^-.., (81)

Since x — 2^ = we can
express q and therefore
s/J, re in terms of the
length of the gap between the hemispheres,

For large re,

Jtt - W) ~ Jn " 030 = '

(n + 0.5)! '
(re + 0.5)! 2
Zn - * jS^/ST" » — (83)

n+ = 2mx +J
1
=
jtad joicS jtad

From these equations we conclude that the coefficients v kn in

equations 74 for the 6's decrease rapidly as re increases and s decreases
hence, the influence of the a's on the 6's decreases.
As the frequency approaches zero, approaches zero and
- 5 > !fc
Ntik(fU) = - -fcfiSL > Nnt '(fil) = (fc -

(84)

The terminating admittance will approach the admittance of the

capacitor formed by the hemispheres, and we may write

t, - y«c„

C
«-S^v ..WtV^(co ^. 8 (85)

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
52 MODE THEORY OF ANTENNAS 2.9

As ^ approaches r/2,

(86)
(f -*)•
Since Pi(cos tf-) also approaches zero, we need a finer approximation
than equation 78 in order to evaluate equation 85. To obtain this
approximation (for the case of general n) we shall use the exact values
of P„(cos \f/) and its derivative at ^ = t/2,

If ^ not too near

is or x, the approximate solution of Legendre's
equation is

P.(cos*)=
+
Vsin y>
f
= [(n + „, + mm
Hence,

Acos^ + *sin-2p«P.(0) f

(89)
-.4 sin-^ + tfcos-^ = - ± Fu >(Q).

Solving for A, B and substituting in equation 88, we find

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.9 SYMMETRIC SPHKIUCAL ANTENNAS 53

and equation 90 becomes

«»[(" + t)*-t]
P„(cos*) = (92)
Vfnx sin \J/

Let

t> = x- 2* = y •
(93)

Substituting from equation 92 in equation 85 and noting that k is an odd

integer, we have (for small t>)

(2fc+l)[l-cOB(*+))tfl
r
C
4e*
v
'-^w*L F^TT) (94)

Introducing a new variable of summation,

«-<*+»#. A/ = 2i>, (95)

we have

We shall now neglect At/4t in the parentheses and replace the sum
by an integral,

1 C0S
C = «JT" ~ 3
- <«. (97)

Evaluating,

~ cos ^ **
~ WL
sin
r
Ct - f9*1
1
+
4- - p; a,,!
2 3
(I*?) J

= 2e ^Iog-i+0.52^. (98)

To increase the accuracy we should evaluate the initial terms of the

series 94 and replace the remainder by an integral.
For small # the 94 converges slowly so does the imaginary
series ;

part of the series in equation 77 since, when k is large, the approximation

for Y k + is the same as for small w. Since we have evaluated C h we can
subtract juCt from Y t and thus increase the rapidity of convergence
of the remainder:

Y, = + £ JL -§±\j [/Moos ,)P [>V - ijT] -

(99)

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
54 MODE THEORY OF ANTENNAS 2.10

In fact, this series approaches a limit as t? approaches zero. Thus,

p»(«* «- p. [«.(}- 4)]

A;t (4>
(100)

and

(2*+l) [(!)!?
(101)
4-1.3,... r/A- - 1 \ "|2 * J
*(*+d[(V>]
2.10 Thin biconical antennas
As ^ approaches zero, the inverse radiation impedance, Za = K2 Y t,

approaches a limit which can be expressed either as a series,

Za = K*Y = t
f i_ j
E ™kX\) J**W)V*>W)
-jNntW)]

(102)
or in terms of the sine and cosine integrals,

Z a = R a + jX a ,

Ra=:
ib
Cin 2/W + (2 Cin 2 ^~ Cin 4/3/) 003 2 ^+

£ (Si 4^ - 2 Si 2/3/) sin2pZ,

(103)
X« = Si 2/3/ - J- (Cin 401 - 2 log 2) sin 30 -

Si 4/92 cos 2#.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.10 THIN BICONICAL ANTENNAS 55

These limits can be found in various ways. One method follows

from the fundamental theorem in Section 2.6, and from the earlier
theorem to the effect that, as the radius of the antenna approaches zero,
the current distribution approaches a sinusoidal form with the nodes
at the ends. Thus, as ^ — * 0,
7(r) jKr V(l) sin 0(1 -
1
r) 7 sin f(l - r). (104)

At the same time K approaches infinity, and from equation 52 we find

= -jK cot 01 + + f-J-V (105)

r=-l.

Fig. 2.6 Contour of integration (the broken line) used to obtain the complex
radiated power from a biconical antenna.

For the complex power input to the antenna we find

*= \ZJJi* = JZ*/o/o*sin2 fl

= ^-jK S m0lco80l+Z a +o(^^ (106)

This must equal the complex radiated power,

*= -lfE,I*(r)ds z=l

= - 4/o* f E. sin 0(1 - r) ds, (107)

where the integration is extended along the broken

line in Fig. 2.6 and E, is the tangential component Z=

of the electric intensity. This line is parallel to a

typical generator of the cone, with an infinitesi-
mal, semicircular indentation around the apex.
We are interested, of course, in the limit ap-
proached by ^ as \p approaches 0. Then the cur-
rent is distributed on the axis of the cono and
Fig. 2.7 Explaining
extends from z = —I to z = I. The field is*
equations 108.
(Fig. 2.7)

V 2 cos 01 _ e^*2 \
T2
)
* Electromagnetic Waves, p. 371.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 56 MODE THEORY OF ANTENNAS 2.10

P EP = ~-
V
4t
j/ (e" Wp» cos 0i + cos 2 - 2 cos 0$-** cos $),
(108)

P HP = (e- J "'i + - 2tr** cos

The spherical components of E are

Er = cos 6 + £?, sin 0, E 9 = - E, sin 5 + cos 0. (109)

On the surface of the infinitely thin cone 6 =

E = r E, + Erf
=
irL r r+r T^r-J- (110)
On the surface of the circle of infinitesimal radius r — centered at
e,

Because of symmetry, the integral in equation 107 can be evaluated

along one half of the contour shown in Fig. 2.6 and the result doubled,

¥ = - E r sin 0(1 - r) dr + Jo*^"* etf# sin # cfc

= -^o*^' S P sin 0(1 - r) dr - \jK sin # cos #I Io*. (112)

Comparing with equation 106, we find

Za = - sin /?(/ - r) rfr

+ z_ + x
"27Jo L ? r r=F? J
Wk0(l- r)dr. (113)
Therefore,

Evaluating, we obtain equation 103.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.10 THIN BICONICAL ANTENNAS 57

We might also use equation 54. From equation 108 we find, for r ~ I,

IH * -
#o(«- i2 " , " l> ' + e-w™» - 1 - €-*>*)
|
'
4rsin0 ^

From equation 104,

therefore,

Z = K 2 Y = ~—
fl t I iXi ^— — ?
(29. (116)

Taking account of the symmetry of the integrand with respect to

= 7r/2 and introducing new variables of integration x = sin and
y = cos iff, we again obtain equation 103.
Equations 102 and 103 represent the same function.* For finite
values of K the function Z a depends on all three parameters involved
in the problem, that is, on Z, X, and the maximum radius of the antenna
a. From dimensional considerations we conclude that the impedance
must involve ratios of these parameters l/a, a/\ that is, we may ;

use K, 01, and pa as three independent parameters. It is not difficult

to anticipate that, as K increases indefinitely, we can express Z a in
the following form:

MMK)~M^W+J^ei + J^pL+-. (117)

This may
be derived from the preceding expressions and from equations
37, 45, and 46. As $ approaches zero, the ratio a/X also approaches
zero; the order of magnitude of the total current 1(f) entering the
"cap" at the end of the upper antenna arm may thus be calculated
from the between the caps at the upper and
electrostatic capacitance
lower ends. The capacitance of a disk in free space is 8ea. Each cap
presents only one face to free space and its capacitance is 4ea. The two
caps are in series, and the capacitance between them is 2ea. Hence, the
corresponding admittance is 2jaca = Awja/Xij. Since K = 120 log(2i/o)
we have
a = 21 e-K ' iZ0 2jw>a - j
,
^- e~ K ' i2 °. —
(118)
n x
Hence, in effect, we have summed the series 102 of products of Rnftrtfil func-
*

tions. Following this clue Mr. S. O. Rice, a colleague of the author, undertook to
sum the series by direct transformation, and thus was able to sum other series of this
type. Subsequently his results were used in the theory of end-fed antennas. The
method of summation is described by Rice in his paper, Sums of series of the form

ZX/m-aU) J n+fi{Z), Phil. Mag. (7), 35, October 1944, pp. 686-693.
o

Original from
Digged by ^OOgie UNIVERSITY OF MICHIGAN
58 MODE THEORY OF ANTENNAS 2.10

Hence, the total current 1(1) flowing into the caps at the ends of the cones
vanishes exponentially as K increases; that is, it tends to zero more rapidly
than any power of \/K.
This means that, to obtain the coefficients of the various powers
of K in equation 117, we may assume

HI) = 0. (119)

From equations 37, 45, and 46, we then have

F
'=-w = -w?iR^
In the vicinity of = 0, we have
sin
^M " (cos
^ (l20)

<f/

p„(cos *) = i + om,
P.(-cos^) = | sin nr [log sin -| + *<n) - *(0)][1 + 0(#>)] +
[l + 0(^)] cob nx, (121)

1 ~ c° 8 nr -
M„(co3 *) =
2
[1 + 0(^)]

[log sin | + Hn) - *<0)][1 + OW)}

where ^(n) is the logarithmic derivative of the factorial of n. There-
fore,

d
& = - + ow, (122)

and, omitting the terms of order

r,
"Tw?rai' (123)

The index of summation n is a root of 3f n (cos^) = 0; ignoring the

terms of order >p 2 and introducing K
in the logarithmic term, we have
an approximate equation for the values of n:

l ~™ n* + sin n* [A + i!Sl ~ »<"> = 0,

]
(124)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.10 THIN BICONICAL ANTENNAS 59

As K increases, n approaches an odd integer ;

hence, we write

n = m+ S mt m= 1, 3, 5, (125)

Substituting in equation 124, we find

(126)

Hence, 8 m may be expanded in a series of reciprocals of K. The first

term is

4"
<0> = (127)
'

and

n = m + -7K
+0 (-W)- d28)

Substituting in equation 123, we find

F
--ar%^.55rTU' (129)

As K approaches infinity,

limJlf.(cos*) = P m (cos0), m= 1,3,5,* (130)

Since Er must be continuous &t r = I, we obtain, from equations 31

and 39,
£
M-1,3,...
a* P«(coa 5) - £
*-l,3....
J
&* / *(cos *). (131)

Therefore,
am = bm (132)
and

Y, = (133)
- inrm -5._. she + 1)
'

This is in agreement with equation 55, since in the present case

Pt(cos^) = 1.
Hence, to evaluate the limit of Z a = K 2 Y we need only the tt

limits of Kbk- These limits may be found in various ways. For

instance, we can expand some component of the field (equations 108)
for large r, and compare it with the corresponding expansion obtained
from equations 39, 40, 41. For instance, when r is large,

p = r sin d, r. = r + I cos 6, r2 = r - I cos $, (134)

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
60 MODE THEORY OF ANTENNAS 2.11

and, since Io'j V^/K,

r sin f = - [cos(# cos 6) - cos #] (135)

^ ^rW
From equations 1-14, we have

= {rsineH ^ (136)

therefore,

2wjt*x*E r = - 01 sin(# cos 0) «-'* (137)

From the known expansion,

sin cos*) = r (- )•*-»(»+ 1) J^r IT

/hhW^(cos«
*-I,3.... \ *pt

=
^ JJ (-)*<*"»(2* + 1) Jn*0«) P*(cos

(138)
we u
have

~ -Tp" €_J,,r 5 (-)*<*-,} (2Jfc + 1) Jn k {0l) P k (co80).

(139)

On the other hand, for large r the functions given by equation 43 become

Rk(fr) = jMe-M* = (_)K*+i>e-tfr (HO)

Hence, from equation 39, we have

(141)

Comparing equations 139 and 141, we find

Kb k = -V(l)(2k+\)Jn k (mWn k (fil)-jNn k ({il)l (142)

Substituting in equation 133, we obtain equation 102.

2.11 Wide-angle conical dipoles

If $ is neither very small nor near t/2, we have no alternative but to
solve equations 65 and 74 for the a*'s and the b k sf or the equivalent
equations 75 for the b k B. There are two methods that may be used in

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
2.11 WIDE-ANGLE CONICAL DIPOLES 61

solving an system of linear algebraic equations. If the series

infinite
for the field components are convergent, the high order coefficients must
be small enough to be relatively unimportant; hence, we can neglect
them and solve the remaining finite system of equations. The other
method is to start with approximate values of the coefficients and
obtain the higher order approximations by some iterative process.
To method, let us retain only the first internal
illustrate the first
and external complementary waves. From equations 65 and 74, we
have
8
a ni =u ntlb u -
]ff P
(coa»)+-|alt>i
1 a nj
B1 (143)

where the u and v coefficients are given by equations 72 and 74. Solving
for bi, we have

6, = - * V(lJ 7 * • (144)

Here we have
dM ni (cos
6(2ni + 1) sin ^ cos 2 ^ /
;

fln,Unil = (145)

ni(m -f l)[2 - m(m + l)]

2
dn

We can simplify this expression by noting that n and ^ are not inde-
pendent but connected by the boundary condition,
JW„(cos *) = 0. (14G)

Taking the total differential, we obtain

dM n dMJLdn =
d$
dT^ +'
dn
Q; (147)

therefore,

dMJdt dn
(148)
dMJdn #
Thus,
„
vi^mt
,.
6(2ni + 1) sin ^ cos 2 * dm .

» 1 (n I + l)(m + 2)»(ni-l)» d*

If ^ is not very small, we find, from equations 80 and 81,

"»-7^SF"
2x
2
1

— t -
IS
2^ nwn
(150)

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
S2 MODE THEORY OF ANTENNAS 2.U

Hence,
dni 4t 1
+
,
-£- + -••- USD

As ^ approaches ir/2,

3x /
8 V
In this case, we obtain the same value for bi as from equation 76 for the
spherical antenna. Furthermore, it is evident that ^ does not have
to be very near ir/2 for the vu product to be small compared with unity;
consequently, equation 77 should hold for wide-angle conical dipoles.
As ^ approaches zero, we have
1 dm _ 1
„
ni i1
+ log(2/*)
'
# *[log(2/*)] 2 '
(153)

vi ni u ni i = -1-

Substituting in equation 144,

="
61
KtfJ+Zi-) ^7T~ JrixWUndm-jNnm)]. (154)

This is in agreement with equation 142.

Next we could take the TEM
wave and one complementary wave
in the antenna region together with two external waves. Inasmuch
as the susceptance components of Y t are represented by slowly con-
verging series either when ^ is small or when ^ is near tt/2, it is to be
expected that the higher order waves are important ; but to take a
larger number waves would result in excessive algebraic com-
of these
plexity of the final formula. However, it is not difficult to solve linear
algebraic equations numerically, even if we take as many as ten higher
order waves. With this number of terms the results should be accurate
for most purposes even fewer terms may be adequate. In Section 9
;

we have seen that for wide angle cones the higher order internal waves
are not very important. Hence, we should be able to obtain good
approximations for such cones by neglecting the internal waves but
including all the external waves. On the other hand, for small angle
cones we need more internal waves. A graphic picture of their effect
on the terminal admittance is shown by C. T. Tai.*
The alternative is to solve our system of equations by successive
approximations in such a way that at each stage of approximation, we
• On
the theory of biconical antennas, Jour. Appl. Phys., 19, December 1948,
pp. 1155-1100.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
2.11 WIDE-ANGLE CONICAL DIPOLES 63

may take into account all the higher order waves. For this purpose
we shall write equations 75 as follows:
2k +
Z k +b k = -
1
q V(l) P*(cos - E Zk* b a , (155)

z ka - =
fe(fc +
l)(2fc+l)(2n+l)sin^^(cos^)iJ (cos A) tt 1 rfn _
*T n(n + l)(fc + n + l)(a + n + 1)(* - n)(« - ») #^ n
"
U00;

Transferring the principal diagonal coefficient to the left side, we obtain

(Z k + +Z kk ~)b k = - -^ti v V (i) Pk{cos _ Zfca6a( (157)

where the prime signifies the omission of the term corresponding to

a = k. We now write
bk = 6*<°> + &*«> + 6*< 2) + • •
(158)
where

V+ •
(160)

This iterative process based on the assumption that the dominant

is

coefficients are in the principal diagonal, so that the first approximation

(equation 159) may be obtained by neglecting all other terms. In
effect, instead of neglecting all the internal higher order waves (given
by the a*'s) in calculating a particular external wave, at first we take
wave and neglect only the
into consideration their direct effect on this
indirect effect through the remaining external waves. The largest
~
contributions to Z kk come from those terms in the series 156 in which
n is nearly equal to k.
Substituting from equations 159 and 160 in equation 55, we find

Z a = K 2 Y = Z a <0)
t + 2a (l)
+ Za (2) +" *, (161)
where

* *-ft.... k(k + l){Z k + +Z kk -)

'

*w --t^«£_^^^» (163)

The approximation reduces to the expressions we have already

first

obtained when ^ is small and when ^ is near jt/2 hence, we may antici- ;

pate that it will be a correspondingly good approximation for all angles.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
04 MODE THEORY OF ANTENNAS 2.11

For numerical computation we require the values of the Legendre

polynomials, and of the functions n(tf-), and dn/d$. If ^ is not too near
zero,

-<« - - 1
+
p%)* - ir •
-'•*»•••,
- 3 (l(H)
= 2mx J_ (t 2^)
x-2* 2 lorn* 1024m3 *-3
and
dn »W = 4my +
+ _L_ (165)
# - Srnx +
"

(x 2^)2

As $ becomes small, we have the following approximations to the

Legendre functions:

7\(cos0) = JoipO), p = Vn(n + 1),

P n (-coa0) = Ju(pO) coanx +N (p8) ainnir, (166)

M B (cos0) = i(l - cosnx) J (pe) - %8\nnir N (p6).

Hence, the zeros of M „(cos 0) may be obtained from

cot-pj- = , n = — cot
/. »
1
, • (lo/

If we tabulate solutions of this equation a function of p^, we for n as

have only to divide by p to obtain a table for n as a function of ^.
We have already seen that, as ^ approaches zero, n approaches an
odd integer 2m — 1, m = 1, 2, 3,- * •. From equation 164 we would
find 2m — 0.5. Hence, we anticipate that equation 164 fails only
for really small values of The solid curves in Fig. 2.8 represent n(\f/)
as given by equation 164 ; the dotted curves are taken from equation
167. In the immediate vicinity of $ = 0, the dotted curves approach
odd integers so sharply (on the present scale) that they appear to
coincide with the vertical coordinate axis. We can obtain a good idea
of the accuracy of equation 164 if we note that, when n is an odd integer,
JW„(cos0) coincides with P„(cos0); and, when n approaches an even
integer, M becomes more nearly proportional to Q«(cos0).
n (coed)

In either case it is easy to calculate the zeros of n (cos 6) as accurately M

as wo wish. The following table presents the exact and approximate
(from equation 164) values of ^ when m = 1

n - 2, 3, 4, 5;
(Exact) * - 0.355, 0.6847, 0.8776, 1.0022;
(Approx)* = 0.338, 0.6822, 0.8769, 1.0019.

/~* Original from

... ..
Digged by V^OOgK
<-» 1
UNIVERSITY OF MICHIGAN
2.12 ANTENNAS IN DISSIPATIVE MEDIA 05

For m= 2, we have
5, 6;
(Exact) * - 0.198, 0.4366, 0.6078;
(Approx)V = 0.183, 0.4331, 0.6070.

12

—
10

o
II

!'

O
<o 5

7\
0.1 0.2 0.3 0.4 0.5 0.6 0.7 o.a
ANGLE, lp, IN RADIANS

Fig. 2.8 Roote of the characteristic equation M „ (cob — 0. The solid line repre-

sents equation 164; the dotted line represents the results obtained from equation 167.

2.12 Antennas in dissipative media

Although we have assumed that the medium surrounding the antenna
is nondissipative, the expressions we have obtained may be generalized

immediately. We need only replace jfl by a = Vjajifg + j<ae) (and

jwe by g + iwe, where necessary). For example, the principal current
and transverse voltage in a dissipative medium are obtained from

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 MODE THEORY OF ANTENNAS 2.13

equations 34 and 35,

7 (r) = K~ V(/)[sinh a (I - r) + KY, cosh a(l -

l r)],

V(r) = 7(i)[cosh *{l - r) + KKi sinh <x(2 - r)], (168)

Equation 103 for thin biconical antennas may be written in the

form

Z a = K*Y = t + j Si 20) + j 2 log 2 sin 201 +

[2(Cin 201

e-2 ^ (Cin 201 - j Si 2$) + e*w (Cin

1
+ j Si 2#) - 2/32

e»W (Cin4/H+iSi4/M)]. (169)

Replacing j0 by <r, we have

Za = -2- [2 Ein 2<r2 + 2 log 2 sinh 2aJ + e" 2* 1 Ein (-2*2) +

e
2* 1 Ein 2*2 - e
2" Ein Ml, (170)
where

-
+f~ ^-
1
Ein w
t
dl = C + logw dt. (171)

As re(ffZ) increases,

Z ^^:(C + log2<r2). (172)

2.13 Extension of known solutions

In the preceding example the known solution was obviously general

except in superficial details; even the transformation from equation 169
to equation 170 was made only for convenience. We might equally
well have left Z a in the original form 103 then, in numerical calculations
;

we would simply use the formula = —j<r. In the following example

the changes that should be made in the known solution are slightly
more extensive.
Consider a biconical antenna in a medium the constants of which
change abruptly at the spherical surface r = l\ (Fig. 2.9). Suppose
that the antenna is thin and that we already know the field in the
absence of the discontinuity; that is, the bk's as given by equation 142.
We can take this field as the primary field impressed on the surface of
discontinuity r = h. From the continuity of E H9
tl at this boundary
we calculate the toward the antenna. Since the medium
field reflected

is continuous across the antenna output boundary So and since for thin
biconical antennas P*(cos 6) equals M*(cos 0), there will be no rereflec-

Original from
Digiiized by VjOOQ IC UNIVERSITY OF MICHIGAN
2.13 EXTENSION OF KNOWN SOLUTIONS 67

tion at this boundary. The primary field is given by equations 40 and

41 with Rki&r) defined by equation 43. For the reflected field R t (0r)
= Jn k (0r) and for the transmitted field R k — Hn k {$'r). Hence, for the

/ [J
MV\
1
I ^ \\ \ \

Fio. 2.9 A biconical antenna in a medium the constants of which change abruptly
across the sphere r fi.

fcth order components we have (omitting the Legendre functions)

2vk(k + l)rHp . n = bk -gj-ggy + ck

-j^y , r<h,
Hn k (&'r)
ak
Hn k (0'li)
' r>h,
(173)

~ » ik HnMIO ' r > h.

From the continuity conditions at r = h,

. HnuWx) Jn k (0h) _
bk
Hnuifii)
+
, .
* /n»W0
~~ d *'

(174)
gn/(gi) Wgji) j' Hnk'(fi'h)
°*
Hn k (fil) * Ck
Jnuifl) v Hn k (fi'h)
Solving for c k /b k ,

y' Hn k '($'h) Hn k (0h) Hn k '(Qh)

Ck v Hntifi'h) Hn k ((3l) Hnk (fil)
(175)
bk Hn k '(0'h) Jn k (0h) _ Jn k 'Wi)
V Hnufflx) Jn k (J3l) Jn k (pl)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
68 MODE THEORY OF ANTENNAS 2.14

Now at r = I,

2rk{k + l)lH p , n - 6* + C„ - b k
f\
+ -JlJ ; (176)

that is, the old 6*'s should be multiplied by 1 + (c k/b k ). This is the
factor we have to insert in equation 102.
If, in particular, h = I, then,

1 + -^- -j|Jn 4 W)ll»*W)] [— "

^(W"J
*

(177)

In this case the first bracketed factor cancels the normalized Bessel
functions in equation 102, and the second bracketed factor appears
in their stead.
One reason for retaining the radial impedances in equation 162 and
the associated equations. is that it makes the expressions concise. An
even more important reason is that in this form the expressions are very
general. Our antenna may be imbedded in a medium consisting of
concentric spherical strata, and the same formula will still apply as
long as we substitute proper wave impedances. The determination
of these wave impedances is a simple one-dimensional problem in
reflection.*

2.14 Input regions

The general form of the radial wave function in the antenna region is

given by equation 36. At r the Nn n functions become infinite

= 0,
for all n > 0; and, if the voltages are to remain finite as r approaches
zero, the p B's must be equal to zero. If the radius r< of the input bound-
ary is different from zero, these coefficients do not necessarily vanish,
and the voltage impressed on the antenna does not define uniquely
the field around the antenna —
we need the complete specification of
E$ over the input boundary. Expanding i?»(r<) in a series of the
type 33 appropriate to the antenna region we have

d
r, Et{u) sin B-^ M n (co& 6) d$ (178)
X"-*
Em =
sin 6 ]^-~ M tl (vos tf)J d$

Electromagnet ie Waves, pp. 22G-227.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
2.14 INPUT REGIONS 69

Comparing with equation 33, we have

jv a n Sn'(frt)
En . (179)
27rn(n+ 1 )£»(#)

This represents another infinite system of linear equations which must

be added to the systems involving a„'s and b n '$ directly.
The handling of these equations can be simplified if we take advan-
tage of the fact that our equations are linear. // we solve our antenna
problem for each term in equation 178, we have only to add our solutions to
obtain the solution for an arbitrary input field. The solution of each
component problem involves only a doubly infinite system of coefficients,
a„'s and 6 n 's, the case already treated in the preceding sections, for all
the E n's except one will be equal to zero in each case. We shall presently
see that this method simplifies the solution to an even greater extent
than might be anticipated.
Let us consider the first term and assume that all E n'a vanish.
From equations 179 and 36 we have

+p = pn = -
f
Jn n (0n) NnJtfrt) (180)
0,
^/ff
•
n

Hence,

S n (f$r) = Jnn (fir) - Ar n (0r), (181)

jffiff
and our problem is not different from that in which the radius of the
input region is zero; we simply have another function 5 n (/3r) in place
of Jn n (fir). All formal algebraic expressions remain the same, and
the only difference is in the numerical values we have to substitute
to obtain the answers under specific conditions. Furthermore, we shall
usually find that is small compared with unity then, ;

- u?w;
Nnn'ifc) 2 2 »+ I n(n + J)!(n J)!
'

This quantity is small, particularly for large n. The expressions for

Za involve S n {fil); since the amplitudes of Nn n {f3l) are comparable
to those of Jn n (fil), the second term in equation 181 is negligible.

Next let us consider any higher order mode in equation 178. If

the cones are infinitely long, the radial wave function is given by p n = —
and the corresponding field by

r2E = - 4*-Hn n (0r)

r M n (cosd), Hn n (0r) = Jn n {fir) - jNn n ($r),

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
70 MODE THEORY OF ANTENNAS 2.14

(183)

At the input boundary,

B+?M-Mm A, =
+ (184)
n(n+l) ;

hence,

For cones of wide angle, n is large and the field at r = 2 will be only a
very small fraction of its value at r = n hence it makes
; little difference

whether the cones are continued beyond r = I or not. The same is true
for thin cones in the case of any but the lowest mode ; in addition there
is little difference between the field configurations in the antenna region
and on these modes is small
in free space, so that the effect of the cones
and their termination will cause little reflection. Consequently equa-
tions 185 represent the required solution to a high degree of approxima-
tion. The increase in the input current due to the departure of the
field distribution at the input boundary from that corresponding to
the TEM waves is

*-jr?*rfegH*W*- ~» ( (186)

The ratio Ali/V(Ti) represents an admittance in parallel with the

antenna; it is mainly a capacitive admittance, and it is associated
primarily with the input region,* since the length of the antenna is not
involved. The admittance is small because the En a r
are normally
small and the ratio of the radial wave functions in equation 186 is of

the order of 0r<. If the antenna is fed by a pair of parallel wires, for
instance, the distribution of the impressed field is similar to that of the
TEM wave in the antenna region and the E n 'a represent only the small
difference between the distributions corresponding to TEM waves
along parallel wires and along divergent cones.
• Thisnot the total admittance of the input region (which can be made
is

arbitrarily large), but that part ot it which represents " fringing " or electric energy
stored outside the input region due to a possibly irregular field inside the input region.

Digitized by V^OOgK UNIVERSITY OF MICHIGAN

2.15 THIN ANTENNAS OF ARBITRARY SHAPE 71

The above adequate for practical applications, but it is

results are
instructive to evaluate the effect of the discontinuity at r = I on the
field given by equation 185. In accordance with the previously outlined
method, we assume that E n is zero for all values of n except n = m,
let us say. For these values the radial wave function is given by
equation 181- For the TEM
wave V(r ) = 0, and, instead of equations {

34 and 35, we have

Io(r) = 7o(n) cos fi(r - r,) f V(r) = -jK / (r.) sin fi(r - r,).

(187)

Hence, the field in the antenna region may be written as follows:

„ an S n (0r)
*
jf \
rH r = {rH,) r +
\
h{r)
+ ^ £ -^-^ -gjg- w M
1 d ,
n(cos
m.
e),

rE > = (rE
^ + 2^ib + < 188 >

v a» d
~mr m M
V 8m' (fir)
J ' (C09 6) '
-2? ? ir^r+ry
where (rH P ) p and (rE 9 ) p are the "primary fields" given by equations
185 with n = m. If we choose the S m (fir) in the form 181, then the
corresponding term in equation 188 contributes nothing to r<E#(r<),
and at r = r,- the function rE$ reduces to Em . The free space field will
be of the form given by equations 40, 41, and 43. The equations for
the a„'s and 6*'s are then obtained as in Section 2.7; there will be
only slight alterations in some of the terms.

2.15 Thin antennas of arbitrary shape

In theory the general pattern of analysis in the preceding sections is
applicable to nonconical boundaries. In practice we are limited by
our ability to determine the proper wave functions consistent with
the more general boundaries. If the diameter of the antenna is small
compared with its length, these wave functions may be calculated from
those associated with conical boundaries by a perturbation method.
In fact, the general formula for the input impedance, to the same order
of accuracy as that obtained in Section 2.10 for thin biconical antennas,
can be obtained by evaluating the perturbed wave functions for the
TEM waves alone. This follows at once from equations 34 and 35.
Since Yt is of order 1 /K 2 , the second terms in the brackets are of order
\/K. If we perturb the boundary into some other shape, the
conical
first correction term to be added to Y t will be of order X/K
3 and its
,

2
effect on the current, voltage, and input impedance will be of order 1/K .

f~* Original from

_,. ..
Digitized by V^OOglL
I
UNIVERSITY OF MICHIGAN
2.16 ASYMMETRIC SPHERICAL ANTENNAS 73

highly attenuated and store a negligible amount of energy in the vicinity

of the gap. Hence, at the surface of the sphere of radius I, we have
E$ (l,$) = 0, if 0<$ <$i;
b

=
mL- if

if 02 < < tt;

<
189 >

0,
where

and V(l) is the transverse voltage between the two sections of the
spherical antenna. The expressions for the field outside the sphere
differ from equations 39, 40, 41, and 43 only in that the index of summa-
tion A; now assumes all positive integral values. If

E$(l, 0) = 5 A k Pk i(cos0), 7V(cos0) =

d P* ( ™ d)
. (190)

then, in view of the orthogonal properties of the associated Legendre

functions,

At =
2^TTrX'^ y,(?)Pfcl(C03f?)siIlfi ^ (191)

In particular, when E$(l, 0) is given by equation 189,

Ak
y(2k + 1) V(l)[Pt(cos0 2 ) -P t (cosgx)]
(192)
uk(i +
\)Ki •

The propagation factors for various modes from r = I to r = r are

the ratios Hn k '(fir)/Hn k '{fil); therefore,

The radial wave admittances at r = i for the various modes are the
reciprocals of the wave impedances (equation 60),

Hn k (0l) Jn k (0l) - j Nn k (fil)

Yk + = - jNn k '(pi)]
,

V**)
jr, Hn k '(fil)
'
j v [Jn k 'ifil)

and the propagation factors for the various components of H 9 are the
ratios Hnk (flr)/Hnk (0iy t hence,

B,(r, = XA k Yk + iVfrosfl). (195)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
74 MODE THEORY OF ANTENNAS 2.16

From equation 54 we now find the inverse radiation impedance,

K*+ tP*(C08tf2 )- P*(eos*,)]*

Z a = K*Y = t
£^ £ k(k + \) (196)

The admittance F ( and the space outside it, as seen across

of the sphere
the gap, is thus a parallel combination of an infinite number of admit-
tances of networks of the type shown in Fig. 6.2a. The multipliers
of Y k + are merely transformer ratios either raising or lowering the
general admittance level of each network.
Green's function for the spherical antenna is the response to a unit
voltage across an infinitesimal gap. When the response is the current
in the sphere, this function is called the transfer admittance. From
the above equations we find the following expression for the transfer
admittance between 6\ = 62 = ^ and 6 = 6:

y(g;^) = 2 y sing 3 in^i: Yk + Pt l

(cos 0) Pk l
(cos (197)

The wave admittances Y k + may be expanded in partial fractions

representing natural oscillations of the sphere with the gap conductively
closed. Because of radiation the natural oscillation constants are
complex, and for their determination it is more convenient to replace
jo) by p and write the wave admittance in the following form

„ . Kn k (w) . /—

(198)

v
' i^o m\ (k - m)\ (2w) m
For natural oscillations,

oo, Kn k '(w) = 0. (199)

Let the nth root of this equation be w = wk , n. The residue of Yk +

at the corresponding pole is

~*nk = -
B k ,n = Bm(W - w k n) ^\ lim

_ Kn k {w kl% ) .

**
(
u Kn k "{w k n )
}
,

But the normalized Bessel function Kn k (w) satisfies the differential

equation
d2Kn
dw'2

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.16 ASYMMETRIC SPHERICAL ANTENNAS 75

hence, the residue is

«m = -f 1
[i + •

From equation 198 we see that, since the exponential factor may be
canceled, the radial admittance is a rational fraction. The degree of
the numerator equals that of the denominator; this we find either
by observing
directly or that, as w approaches infinity, Kn approaches
exp(— w) and Y k + the ratio Hence, the expansion in partial
fractions is

Y k+ = —+^
1
L . Bk ,
n

= t203)
7 f
1
" ? lw k .n
2
+ k(k +!)](»- "

Thus, on the real frequency axis,

Y k+ = ~
1
1 - E ^ =4 > (204)

where the natural oscillation constants are given by

"
Ptin = ^. (205)

Substituting in equation 196, we find an expression for Y t in terms

of the natural oscillation constants. Similarly, substituting in equation
197, we obtain another expression for the transfer admittance.
For any gap extending from 6 = B\ to $ = 62 we can express
finite

the current 1(6) in terms of the meridian electric intensity in the gap,

1(d) = ijT* Y(6; *) E9 (l, *) #. (206)

given by equation 189, 1(6)

If £,(/, 0) is = 2wl sin be
//,(£, 0) may
obtained more directly from equation 195. Substituting from equation
204 and noting that outside the gap

ZA k P k *( C ose) - 0, (207)

we have
2TM^,^P^(cosg)sinf? _
KB) = -T.T. f {m)

where A* is given by equation 192.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
76 MODE THEORY OF ANTENNAS 2.17

2.17 Asymmetric conical di poles

We now assume that the arms of a corneal dipole are of equal length
(along the generators of the cones) but that the cone angles are different.
The conical dipole is similar to the spherical antenna (Fig. 2.10) except
that the angles 8i, 02 are arbitrary. The differences between the sym-
metric and asymmetric cases are minor, although the equations are
somewhat more complicated, principally in the antenna region. In
free space, the field is given by equations 39, 40, and 41 with k assuming
all In the antenna region the values of the
positive integral values.
index of summation n are the roots of equation 17. The corresponding
meridian wave functions become

0(0) = P n (cosO) ~ 209 )

p^-Zsl) (

instead of

6(0) = M n (cos0) = |[P B (cos0) - Pn(-costf)). (210)

In the symmetric case it was natural to introduce the factor J in the

expression for 0(6), but in the general case nothing could be gained by
introducing The a n 's absorb whatever factor we choose to include.
it.

For convenience, we shall rewrite the expressions for H ¥ and E$ in the

antenna region,

rH * '
2*sin0
+ 27 V n(n + 1) IM G * Wi
(21 1

tE%
- 2rKsm0 + 3 27 ? n(n + 1) ~SWT W "

If the radius of the input region is zero,

S n (0r) = Jn»(fir). (212)

The characteristic impedance ia

K = -^ log(cot \ tan
-f-)
(213)

The normalized terminal admittance is represented by a formula

analogous to equation 54,

KYt =
inj) JT H *1 {1
> •> *• < 214 )

Digitized by V^OOglL UNIVERSITY OF MICHIGAN

2.17 ASYMMETRIC CONICAL DIPOLES 77

Substituting from

2wfr k(k+l) dS
(215)
7 . ' bk Rt'(fir) d
rE' j
2r £ k{ 1) AV.^i <tt
P*(cos0),

we find

[P*(cos() 2 ) -P*(C06ffi)l (216)

As in Section 2.8 we can express the a n 's in terms of the b k s, and

vice versa. Thus,

a„ = E u nk b k ,

(217)

where

k(k + 1)^'' [9» W sin 9 di)

(2n+ 1) sin *P* (cosfi)

^ (218)

(k -n)(n + +1) sin O^f^f

ft

fc(A)+l)(2fc+l)
2 „ (B +, ) (n _ fc
) (fc + B + TJ |fr
sin • >'*<co S 9)
^ (219)

Eliminating the a„'s from equations 217, we have

(2fc+ 1)9 F«)
6* v_ IP* (cos fc) - P*(cos*i)] + EE v knu nab a. (220)

As in Section 2.10 we shall write these equations as follows:

(2k V[l)
Z k +b k = [P 4 (cos*2 ) - P*(cob*i)] - E 2* -6 a
,
(221)

(222)

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
78 MODE THEORY OF ANTENNAS 2.17

Transferring the principal diagonal coefficient to the left side, we obtain

(Z* + + zk*-)bk
2k + 1
- - T,'Z ka -b a
v V(l)[P k (cos02 ) P*(cos0i)] , (223)
2K
where the prime after Y. signifies the omission of the term corresponding
to a = k. This equation is of the same form as equation 157, and
we may write b k in the form 158 with
,
o
(2fc+l)iy[P»(coBfe) k -P (me x )]
V(l). (224)
2K(Z k + Z«-) +
Substituting from equations 218 and 219, we find

Z k k~ = —Z k + Z v knu nk

h(k + l)(2n + l)(2k + \)Z n ~ [sin 6

^
P*(cos 6)

2n (n + l)(n - fc)
2 (n
+ + ft l) 2 sm0§; ^ f
(225)

In evaluating the various terms of the series we note that a constant

multiplier associated with 9 does not affect
the results and we may write

Bn ($) = P.(-cos*i)P.(eostf) -
P.(cosfc)P»(-cos«) (226)

instead of equation 209. When — $t 6

t

vanishes for all n ;

hence,

= 0. (227)
dn
We also have'
2 sin nir
Fir. 2.11 A conical antenna (228)
above a finite hemispherical
60 It-*! 7T sin 0i

ground. Since 2 and n are connected by the equation

e„(*2 ) = 0, (229)
we have
d9 n (02 )
0>
a8^__«^i*_. (230)
d02 cm
The largest contributions to Z kk ~ come from values of n nearest to k.
* Applied Mathematics, p. 433.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.18 END-FED ANTENNAS 79

If $2 = jt/2, we have the case of an antenna above a finite hemi-

spherical ground (Fig. 2.11).

2.18 End-fed antennas

It is possible to connect one terminal of a generator to a wire and leave
the other terminal floating (Fig. 2.12a). The antenna will thus be
operated against the capacitance of the generator. An idealized struc-

fa) (b)

Fig. 2.12 End-fed antennas.

ture representing all the essential features of an end-fed antenna is

shown impressed between a cone and

in Fig. 2.126, where the voltage is

a small sphere of radius 6. Thin cones may be deformed into cylindrical

wires and wires of other shapes and the results obtained for cones may

Fio. 2.13 An end-fed antenna.

thus be extended by the perturbation method explained in Section 2.15.

A practical method of feeding an antenna from one of its ends is shown
in Fig. 2.13.
The impedance antenna shown in Fig. 2.126 as seen at the
of the
gap equals the impedance of the antenna proper in series with the
capacitive admittance of the sphere. There is also a parallel capacitance
shunting the cone and the sphere. Our model contains all the essential
features of an actual antenna, and the various effects are clearly sepa-
rated, so that we can make proper allowances in applying theoretical
results to various practical situations.

Original from
... ..
Digitized by V^OOgK
<-»,-»,-* I

UNIVERSITY OF MICHIGAN
80 MODE THEORY OF ANTENNAS 2.18

The impedance of an infinitely long cone of small angle 2$ is found

to be

K = 60 log -| - 30. (231)

The first term equals one half of the impedance of two such cones going
in opposite directions. The capacitance of the small sphere of radius
b in series with this cone is

c» -
i - Sw, •
M
The numerator is the capacitance of the sphere in free space; the
denominator expresses the effect of the cone. Equation 232 is of
practical interest only insofar as it illustrates our ideas about the imped-
ance of the end-fed antenna; the actual value of the capacitance is
of little interest since it varies with the structure of practical generators
and practical input regions. Similarly, if t? and ^ are both small,
and yet t? is large compared with then the capacitance in shunt
between the cone and the sphere is

Tea

C.h = (233)

This is small compared with the capacitance of the sphere. It will

bo large if t? small compared with ^.
is

The input impedance of a thin cone of length I is

_ Z a sm0l~ j(K + 60) cos 01

7
Zi ~ K (K + 60)sin#-jZ a cos# '
(234)

where
Z a = K 2 Y = 30 Ein 201 - 15(1 - e" 2 "
t
1
),

R a = 30 Cin 201 - 15(1 - cos 201), (235)

X a = 30 Si 201 - 15 sin 201
As might have been expected, Ra
represents the radiation resistance
with reference to the current antinode, when the current is sinusoidally
distributed and vanishes at the end of the antenna opposite to that
connected to the generator.*

* Note that Ra and X a for an end-fed wire are the functions Ru and Xn, re-
spectively, tabulated in Appendix II.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.18 END-FED ANTENNAS 81

For an end-fed antenna of any shape,

a
R a sin gj + j[(X a - N) sin gj - (K* + 60 - M ) cos fl]
'
s
[(K a + 60 + M) sin # + (X a + N) cos #] - jR a cos #
(236)

Here K a is the average value of K; the M

and N functions are the
corresponding functions for the balanced two-arm antenna divided
by two (see Section 1.15).
These results may be obtained by any one of the methods already
explained in connection with thin biconical antennas. Likewise the
solution for cones of large angles is analogous to that of large angle
dipoles. The proper functions for the antenna region are P n ( — cos0),
and the values of n are the roots of
P n (-cos^) = 0. (237)
For very small ^,

11 = "' + ' «= 0,1,2, ". (238)

2 log(2/*)

Otherwise we use equation 78 and obtain

n= -^— +w/tt)
1
*
2 8m + 6
(239)

For m =
and ^ = we find from this formula n = A» while the
exact value from equation 238 is zero. Thus, it is only for very small
values of ^ that equation 239 fails.

In the case of a single cone, TEM

waves do not exist ;
however,
if is small, the waves corresponding to the smallest root,

(240)
re
=2T5iW'
closely resemble TEM waves in that the current associated with them
is sinusoidally distributed except in the immediate vicinity of the apex.
Thus, one of the radial wave functions is

Kn n (j(Sr) = e->* - n(n + l)e'+ [Ci 2(3r + j(J* - Si 20r)], (241)

and the other is its conjugate. For larger cone angles the resemblance
disappears.
Figures 2.14 and 2.15 show the input impedance of a cylindrical
monopole for K
a = 320 ohms. The characteristic impedance of the
dipole made by two such monopoles 700 ohms. The maximum is

resistance of the monopole is 1380 ohms; computed on the same basis,

the maximum resistance of the dipole is 1820 ohms. Hence, the maxi-
mum resistance of the monopole is substantially greater than one half
of the maximum resistance of the dipole. As the characteristic imped-

Original from
Digttized byGoOglc UNIVERSITY OF MICHIGAN
 0.1 0.2 0.3 0.4 0.5 0.6 0.7 O.fl 0.9

LENGTH, 4siN WAVELENGTHS

Fio. 2.14 The input resistance of a cylindrical end-fed antenna in free space.

Fia. 2.15 The input reactance of a cylindrical end-fed antenna in free space.
82

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
2.19 CURRENT ELEMENT ABOVE A DISCOID GROUND 83

ance increases, the stored energies per unit length in the monopole and
dipole tend to become equal but the radiated power is different. The
;

difference is caused by the mutual radiation of the two arms. At

antiresonance the monopole is just a half-wave antenna, and its radiation
resistance with reference to the current antinode is 73 ohms, so that
the antiresonant resistance is K
a (K a — 146)/73; the corresponding
radiation resistance of the full dipole is 199 ohms, and that of one half
of it Thus, the ultimate ratio of the maximum resistance
100 ohms.
of the monopole to half that of the dipole is 100 to 73. In the preceding
example the ratio is 100 to 66.

2.19 Current element above a discoid ground

As another illustration of mode analysis, let us consider a current
element of moment Is just above a perfectly conducting disk of radius /
(Fig. 2.16). Here we might write general
expressions for the fields in the three sepa-
rate regions into which the entire space is

subdivided by the disk and a hypothetical

sphere S of radius I centered at the center of
the disk. In this formulation we would have
three sets of coefficients in the expressions \v t

for the fields. The problem is simplified if \ /

we note that the field of our current element
must be the sum of the fields associated with
pairs of current elements, each of moment °\ 2' 1 A curren
^f?
m f
alK„
?.

ir j
\Is, and arranged as shown
»-T^- ,
rt
m
j
Figs. 2.17a and
^ ment >ove a
ground.
discoid

2.176. In the sum the elements below the

plane cancel and leave that region source free. In the arrangement
shown in Fig. 2.176, there are no currents in the disk; hence the disk
can be removed without disturbing the field, and we are left with a
single current element of moment Is. In the arrangement shown in
Fig 21?a
-
'
E r (0) = E t {* - 8). (242)

This condition enables us to express the field in regions 1 and 2 in

terms of a single set of coefficients.
Since E T must be finite on the axis, = 0, (except at the element)
the proper angular wave function is

0(0) = P n (cos*). (243)

Since E T should vanish at the surface of the disk where = t/2, the
proper values of n are roots of
Pn(0) = 0. (244)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
84 MODE THEORY OF ANTENNAS 2.19

These roots are

n= 1,3,5,---. (245)

The corresponding radial wave functions are

Sn(fr) = Jn n (fir) +p n Nn H (fir). (246)

The expressions for the field are analogous to those given by equations
31, 32, and 33 except that the TEM wave is absent. As r -> 0, Nn n
approaches infinity as 1/r"; hence, the corresponding term in E r

*\ o) t&l*'*" (3)

(a) fb)

Fig. 2.17 Two arrangements of current elements in the presence of a discoid ground.

approaches infinity as l/r n+2 On the other hand, the radial electric
.

intensity of the given current element approaches infinity only as 1/r3 ;

hence, for all n > 1, the coefficient p n should be made zero, and

5„03r) = Jn n (0r). (247)

For n = 1 it is convenient to split the radial wave function into two

components: (1) the "primary component" representing the field of
the current element above a disk extended to infinity so that there
is no discontinuity at r = I, and (2) the component representing the

wave reflected from the discontinuity. The second component must

be finite at r = 0, and the corresponding wave function is

Siffir) = Jmtfr). (248)

The first component represents the fieldequal to the free-space field

of the current element of moment Is. In the present notation this is

r*E r > = - Hm(fir) Pi (cos $),

rHs = M!-Hni(fir) Pi* (cos*), (249)

rEt p = jrt
ML Hm'ifir) Pi' (cos B).

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.19 CURRENT ELEMENT ABOVE A DISCOID GROUND 85

In writing the general expression for the field above the disk
(0 < x/2) after the pattern given by equations 31, 32, and 33, we
shall now find it convenient to absorb the factor l/2xn(n 1) in the +
coefficient a„. Thus,

ja«*Ef = jw*B r> - E n(n + l)a. =jNg| P n (cos 6),

rH. = rHs + Ia n P»*(C0fl 5), (250)

rfl, = r*» + jn £ a. Pn ' (cos 5).

j^'ff
For odd integral values of n,

P„[cos(t - $)] = P«(-cos 0) = -P a (cos 6), (251)

hence, to satisfy equation 242 the expressions for the field below the
disk are the same as equations 250 with the sign reversed.
In region 3 denned by r > I the field must be finite when = 0,
t ;
hence, the proper angular function is

6(0) = P*(cos0), (252)

with the added condition that k is an integer. From equation 242 we

find
P*[cos (r - 6)] = P*(-cos 6) = P*(cos 0) ; (253)

that is, the wave function must be an even function of cos 0. Hence,
k must be an even integer,

k = 2, 4, 6,- -. (254)

Thus, in region 3 (Fig. 2.17a) we have

rH * =lt h S£y Ptl (cos e) '

(255 >

At r = I the field must be continuous. In accordance with Max-

well's equations the continuity of E T implies the continuity of Hr, and
vice versa ;
we can match either the pair E T E$ or H Pt Et. We
hence, ,

shall take the latter pair. The coefficients 6* may be expressed either
in terms of H ¥ (l) or in terms of E$(l) by using the orthogonal properties

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
86 MODE THEORY OF ANTENNAS 2.19

of Legendre functions and the symmetry of the field,*

bk = P*' (C0S 8) 8in 9 *'

k{k + 1) Jo '

/2
z t +h = z B< (Z) p *' (cos fl) sin 9 (256)
fc ^ X i) X'

From equation 250 we have

1 = Hmifil) Pi 1 (cos 9) + _E
n
o. P. 1 (cos <?), e < |x,

2 = Hni W)Zi + Pi 1
(cos 9) - T. Zn~a n P„' (cos <?), (257)

Substituting in equations 256, we obtain

b„ = Au kl + 5 il = ffmOSl), (258)
»— 1,3,...

Z k +b„ = AZi+uh - £ Z.-m*„o„ fc = 2, 4, 6, • • •

, (259)
n
where
2fc + 1 Z^"2

= (2fc + P 4 (cos 0) P»(cos 6) sin <fc

2&+ 1
*n ' X
(*-«)(* + » + !)

/2
[V*(cos ff) P n '(cos 0) - P n (cos $) P* l (cos
5)J*
2/ !

P*(0)P n i(0)
(fc-»)(fc + n + l)
_ 2(-)*<*+»+*>(2fc+l) ttfc-j)l(jn)! ,
9fin
.

n)(A; + n + 1) (in- *)!(!*)! ^ ;

* Incidentally J/ f vanishes for r > I in the plane of the disk. Hence, this part
of the plane may be considered as an effective magnetic conductor which, together
with the disk, separates the upper half of space from the lower. See S. A. Schel-
kunoff, Generalized boundary conditions in electromagnetic theory, Proc, Nat. Elec.
Con/., 2, 1946, pp. 317-322.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.19 CURRENT ELEMENT ABOVE A DISCOID GROUND 87

Similarly, we can express the a«'s in terms of either IH 9 (1) or

lEt(l); thus,

ai = -A + J I H¥ (l) Pi 1 (cos 0) sin 6 d$t

~ °
(261)
2n 4- 1 T" 2

or

(262)
2n"a n = - I E9 (l) P n ' (cos 6) sin 9 <0.
w(w | 1} Jq
From equations 255,

I Hp(l) = E bt PmHvm 6), I E,(l) = E Z* + 6 4 (cos 0). (263)

Substituting in equations 261 and 262, we find

ai = -A + E t>u&*, an - Ef n *&*, n > 1, (264)

Zi-ai = - E 2*+tfi*6*, Z n -a B = - EZ k +v nk bk , (265)

where

_
fc(fc+l)(2n+l)
Vn Wt
*-n(n+l)(2fc + l)

2(- )**+»+» (2n---

;

/,•:/•
I
-f 1 ^ j)! (jn)!
(266)
x (fc_ «)(*+*+ 1) n(n+ 1) (in - *)! (ft*)!

Equations 258, 259, 264, and 266 are not independent. Equations
258 and 264 are obtained by expanding the same function I H v (l) into
two different series of Legendre functions; they simply express one
set of coefficients in terms of the other. If we substitute from equation

264 in equation 258, we should have an identity. Hence, from

bk = Au k i - u kl A + EE u knv nab a , (267)

we conclude that
E UknVna = 0, « k,
(268)
= 1, a = k.

Similarly, by substituting from equation 258 in equation 264, we find

E v nk u ka = 0, n a,
* (269)
= 1, n = a.

_,. .. Original from

Digitized by V^OOglL UNIVERSITY OF MICHIGAN
88 MODE THEORY OF ANTENNAS 2.19

Likewise equations 259 are equivalent to equations 265. To find the

coefficients a„ and b kt we may select any pair of nonequivalent sets of
equations from 258, 259, 264, and 265. For example, we may choose
equations 258 and 259 then, if we multiply equation 258 by Z k + and
;

subtract equation 259, we obtain

E (Zk+ + Z n ')u kn a n (Z k + - Zi + )u ki A. (270)

n— 1,3....

We may also make the type of substitution we used in the case of bicon-
ical antennas. The actual method of handling such systems of equa-
tions should be selected aftera study of the orders of magnitude of the
coefficients. We may also rely on physical intuition to guide us in
adopting a particular method of successive approximations. There
is no general rule which we could follow successfully on all occasions.

If the function Jn n (0l) vanishes some particular value of n,

for
the corresponding terms in equations 250 become indeterminate. This
may or may not cause a difficulty in solving the equations. The
difficulty can be removed by writing

a- = nn Jn n (ft), Z n ~a n = -jv Jn n '(0l) fl„ (271)

where fl n is a new coefficient.

If the disk is large, we expect that the coefficients of reflection
an be small when the order of the wave is small. Thus, we may
will
neglect these coefficients in equations 258 and 259 and determine the
coefficients of transmission b k For small k and large I the radial imped-
.

+
ance Zk is nearly equal to rj hence, each set of equations gives approxi-
;

mately the same result. It is to be expected, however, that a better

approximation might be obtained by using both equations. Let us
multiply equation 258 by Z k + and add equation 259; using the sub-
stitution 271, we have

(272)
2Z^ 2Zi^

Neglecting the a's, we have

1+
= Zt+
2 +f
Amu. (273)

Next we turn to equations 264 and 265 and introduce equation 271,

<h Jm(fi) = -A + L vitbk, nn Jn n (0l) = £ v„ b t k ,

(274)

-Mi Jni'tfl) = Zi+A - L Z +v k lk b kt

k

-jrfln Jn n '(0l) = - ZZ k +v mkb h .

(275)
t

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.19 CURRENT ELEMENT ABOVE A DISCOID GROUND 89

Multiplying the nth equation of each set, respectively, by jij Nn n '(0l)

and Nn n (&l) and adding, we find

= lZi
+ Nm(pl) - jn Nm'(0l)]A +
E tin MH'tfQ - Z k + Vnt(0)M* (276)
k

Mn = E [ji? ATn n '(#) - Z*+ Arn„0M)K*6*. (277)

Substituting from equation 273, we have

Mi = [Zx+NnM -jvNm'WM +
Z*+ 1+
AE Mn Nm'ifi) - Z k+ tfmCM)]
2
j"/ (278)

Zk 1+
iv5n = AE ft JVn.'OM) - Z k+ Nn n (0l)) *t? v nk u kl . (279)

Using equation 269, we have an identity

A = AEntUki. (280)
k

Introducing this in the first term of equation 278 and combining the
result with the second term, we obtain

a, = ay.
*
Wl* - »t+) L M + i^m z* J
Vuun . (28D

Similarly, for n ^ 1,
= E VnkUki. (282)
ft

We can multiply this by jrtNn n '(0l) and subtract from equation 279;
we can also multiply the same equation by Nn n (f5l) and add to equation
279 ; in this manner we find

a» = AE i(Zi + - zk+) l~l^T + 1, (283)

z*+ J^*
Thus we have obtained approximate expressions for waves of
various orders reflected from the edge of the disk and for waves trans-
mitted beyond the edge. From the latter we can obtain the effect of
the disk on the radiation pattern of the current element; from the
former the effect of the edge on the element itself. At the element
r = 0, and there is no contribution from waves of any order but the
first. For small values of /3r, Jniifir) = %@2 r 2 and from equation 250 ,

we find that along the element

E - E
r r <» + #»mSi. ( 284 )

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
90 MODE THEORY OF ANTENNAS 2.20

Let us recall that when the element of moment Is is placed above

the disk, the field is the sum of the fields corresponding to the arrange-
ments shown in Fig. 2.17. We have seen that, for the arrangement b,

the field equals the free-space field of the current element of moment Is.

For the arrangement a, we have represented the total field in region 1

ae the sum of the "primary field," obtained on the assumption that
the radius of the disk is infinite, and the reflected field. The former
equals the free-space field of a current element of moment Is. Hence,
the total field of the element of moment Is above the disk (Fig. 2.16)
equals the sum of the free-space field of an element of moment 21s
and the reflected field obtained above. This means that the second
term in equation 284 represents the effect of the finite radius of the
disk as compared to the infinite radius. The corresponding increment
in the impedance of the element is

AZi = -If= _^!L, (285)

where fli is given by equation 281 and A by equation 258.

2.20 Treatment of reflection phenomena by successive approxi-

mations
It isimportant to remember that a sequence of "successive approxima-
tions" to an unknown quantity may converge either rapidly or slowly;
it may even diverge. The success of the method depends on the proper
choice of recurrence relations, and this requires a study of the problem
and the equations involved. We shall stress the point by a simple
example in which we know the exact answer, and compare it with
various sequences of successive approximations. Consider a transmis-
sion line of length I and characteristic impedance let Z be the imped- K ,

ance across the end of the line. The current and voltage in the line
may be expressed as follows:

I(x) Ae-*K*-» + ae-*"'-*\

(286)
v ;
V(x) - KAe-"<*-» - ffoe-'Wi-".

We assume that the amplitude A of the wave moving toward Z is given

Let 1(1) = b ; then the boundary conditions are

A +a = b, KA - Ka = Zb. (287)

We may solve the first equation for b and the second for a,

b = A + a, a = A --j^-b. (288)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.20 SUCCESSIVE APPROXIMATIONS 91

Suppose that Z is nearly equal to K ; then the reflection coefficient

should be small. we may start by neglecting it in calculating b,
Hence,
thus obtaining b = A then we may substitute this value in the second
;

equation to obtain a. Hence, in the first approximation we have

b = A, a = A (l - (289)

We now use the new value of a in the first equation 288 and substitute
the result in the second equation. Thus, in the second approximation,

b = A+A(l-f), a = A(l--|-)(l-4). (290)

Repeating the process, we obtain the third and fourth approximations:

»-*.m(i-4)(i-#-).
o = 4-4)0-4+*)-
(291)
6 =

The nth approximation can be obtained by induction.

Now the exact values are

As we inspect our sequence of successive approximations, we find that

it converges if the absolute value of Z is smaller than K ; otherwise it

diverges. We
have apparently predicated our iterative process on
the assumption that Z is nearly equal to and yet, when the sequenceK ;

converges, it The sequence

converges very slowly for such values.
converges rapidly, not when Z is nearly equal to K and when a is small
as we assumed, but when Z is small compared to K and when a is not
small. While the error in 6 given by the first approximation 289 is

small, only the absolute error in a is small ; the relative error in a is

nearly 100 per cent (when Z K). The second approximation to a is

even worse, although the third is better.

If we start our sequence of approximations by neglecting a in the

second equation of the set 287 and obtain a from the first, the sequence

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
92 MODE THEORY OF ANTENNAS 2.21

will converge only when Z > K. It will converge rapidly when Z> K
and slowly when Z~K .

On the other hand, we may solve both equations 288 for a and take
the average,

a =

as the defining equation for a, while retaining

^ b - '
(293 >

t = A+o (294)

as the equation for b. Then our sequence of approximations becomes,

U
b = A
A, a = AA K-Z ;
2K

'-4i+ -^£+ (^)' + -]'

These series converge rapidly when Z ~ K. We

could also obtain
expansions in powers of (K — Z)/2Z. Neither of these expansions
converges as rapidly as one of the previous ones when Z <SC or Z K. K »
Each expansion has its own region of rapid convergence so that they are
supplementary to each other.
The primary wave may be chosen in different ways. We can
choose a progressive wave or a wave totally reflected at the discontinuity.
The totally reflected wave may be of a kind in which H reduces to zero
at the discontinuity or of a kind in which E vanishes there.
2.21 Miscellaneous problems
There are many problems which can be treated by spherical mode
analysis. In this section we shall discuss some of them briefly. The
case of a conical moaopole above a hemispherical ground of the same
radius as the length of the monopole (Fig. 2.11) ia just a special case of
a dipole with arms of the same length but different spread (Fig. 2.10).
If the radius of the ground from the length of the monopole
is different
(Fig. 2.18a), we have a special case of the general asymmetric dipole
(Fig. 2.186). In this case we have three regions to consider, with
different boundary conditions. In region 1 the angular wave function

_,.
Denized by
„
^OOgK I .-, Original from
UNIVERSITY OF MICHIGAN
2.21 MISCELLANEOUS PROBLEMS 93

is of the form
0(0) = A x P„(cos0) + B! P n (-cm0), (296)

and the proper values of n are determined by the boundary conditions

6(0i) = 9(02 ) = 0. In region 2

G(0) = A 2 P n (cos0), (297)

and the proper values of n are the zeros of P„(cos 02). In region 3, 6 is

also of the form 297 but with n required to assume only integral values.

Fia. 2.18 (a) A monopole above a hemispherical ground and (6) a general asym-
metric dipole.

In region 1 we shall have a TEM wave and waves, with the re- TM
quirement that their intensities be finite at r = thus, the radial wave ;

functions are a„ Jn n (&r). In region 2, both solutions of Bessel's equation

(a). (b)

Fig. 2.19 Conical antennas on two sides of a discoid shield.

are finite, and the radial wave function is b n Jn„((3r) + c„ Nn n (&r).

In region 3, c„ =
—JbH Hence, we have four sets of coefficients in
.

the general series representing the field around the dipole. They can
be calculated from the four boundary conditions expressing the conti-
nuity of Et and H T .

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
94 MODE THEORY OF ANTENNAS 2.21

Amonopole above a discoid ground can be treated more con-

veniently by considering first two symmetric arrangements (Fig. 2.19).
In the arrangement a there are no currents in the disk, the disk may be

Fio. 2.20 Conical antennas with a voltage source above a icting disk.

removed, and we have a symmetric dipole in free space. In the ar-

rangement b the general form of the field in region 1 is exactly the same
as in the arrangement a; but in region 2 the algebraic signs of the

Fia. 2.21 Conical Fig. 2.22 Hollow conical

antennas .energized antennas.
off center.

coefficients should be reversed. In the free-space region surrounding

the system b the field is given by series of the types 39, 40, and 41, but
with k assuming even values.
In Fig. 2.20a we have a monopole connected to ground and fed at
some point above the apex; in Fig. 2.206 we have the same monopole
disconnected from ground. For an infinite ground both of these ar-
rangements are equivalent to a symmetric dipole fed symmetrically at
two points (Fig. 2.21). In the case a the apices A and B should be

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 2.21 MISCELLANEOUS PROBLEMS 95

short-circuited and F(0) vanishes; in the case b the boundary condition

is 7(0) = 0.

If the cones are hollow (Fig. 2.22), we have two interior regions,
3 and 4, to consider ;
but, on account of symmetry, only one set of
additional coefficients. Since E r is an odd function of cos 0, the field
in region 3 will be of the form

juer*Er = E c„ Jnn (fir) P n (cos 0), (298)

and, in region 4,

juzr*E r = -E c n Jn n (f3r) P n (-cos $). (299)

Proper values of n are roots of

Pn(cos^) = 0. (300) iv (2)

In the case of a single hollow cone

(Fig. 2.23) we have two possibilities. The
impressed voltage may be near the apex
in region 3 then we have an end-fed an-
;

tenna. If the source is in region 1, we have

a circular horn. For the dominant wave
in the horn the angular wave function is Fig. 2.23 A circular horn.

(301)

where n is the first root of

-^P. l
(coe*) = 0. (302)

In this equation 2^ is the interior angle of the horn. At r = I the horn

is terminated, and a fraction of the dominant wave is reflected. In
addition, other interior modes are generated, with angular wave func-
tions of the form 301 where n is a root of equation 302. In region 3
the angular wave function is

T(9 lV>) = PJl-coad) cos*, (303)

where m is a root of

JL-P m >(-cos*)=0. (304)

Besides TE modes there will be TM modes for which n and m are the
zeros of P n (cos ^) and P., (— cos f).
l 1

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
96 MODE THEORY OF ANTENNAS 2.21

Figure 2.22 also shows two horns arranged back to back hence, ;

we have a way of calculating the " cross-talk " between two horns.
A nearly rectangular horn is bounded by the half-planes, tp =
and <p = <po, by the plane B = w/2, and by the cone = 0o (Fig. 2.24).

\ \1

Fia. 2.24 A nearly rectangular horn.

If the radius of the sphere in Fig. 2.126 is large, we have an antenna

above a spherical ground. Figure 2.25 shows an antenna radiating into
a horn. In Fig. 2.26 we have a spherical resonator radiating through

Fia. 2.25 An antenna in a horn. Fia. 2.26 A spheri-

cal resonator with a
hole.

a hole.In this case we may represent E$ and E v in the hole by a series

of suitable Legendre functions. For example, a circularly symmetric
field with E 9 = in a hole of angular diameter 2f may be expressed in
the form
W,0) = Za«Pn l
(cos6), (305)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.22 V ANTENNAS, FANS, PORCUPINES 97

where n a zero of P„(cos

is This function may then be expanded
in series of Legendre functions appropriate to the interior and the
exterior of the resonator, that is, in a series of the form

E*{1, 0) = E&m P.. 1 (cos*), m = 1, 2, 3, (306)

where bm a linear function of the a„'s. From this we obtain the

is

transmitted magnetic intensity just outside the sphere,

Hr '{l, B) = Em Zm
/V(cos9), (307)

and the reflected magnetic intensity just

inside,

H,'(l,0)- -Z-hrPmHaxO). (308)

In addition we shall have the primary field

generated by the given source when the
resonator is completely closed; for ex-
ample, we may have
E$ *(l, 0) = 0, Fig. 2.27 A capacitively
(309)
H 9 *(l, 0)-APiHco*6), loaded antenna.

where A is given in terms of the strength of the source. In passing

through the hole the magnetic field should remain continuous; hence,

Z {^T + zir)
bmPm ^ C0Sd ^ ~ A *V(coe*) =0, < *
(310)

Multiplying this equation by Pn l

(cos 6) sin 6, where n
a typical zero is

of P n (cos and integrating, we obtain linear equations for the 6 m 's and
therefore for the a„'s.
Capacitively loaded dipoles (Fig. 2.27) may be treated in the same
way.

2.22 V antennas, fans, porcupines

In the case of symmetric radiating systems, such as a V antenna
less

(Fig. 2.28a), a " fan antenna," (Fig. 2.286), or an antenna with a

counterpoise (Fig. 2.28c), which is a special case of a " porcupine
arrangement " of conductors, the boundary value problem becomes
much more complicated unless the conductors are thin. To illustrate

what is involved, consider the V antenna with two arms of length I.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
98 MODE THEORY OF ANTENNAS 2.22

The " antenna region " is bounded by the sphere of radius I centered
at the common apex of the antenna arms. For this region we must
obtain a set of angular wave functions which vanish at the surface of
each arm. These angular wave functions must be finite everywhere
except in the interiors of the conductors, which are excluded from the

Fig. 2.28 (a) AV antenna, (6) a "fan" antenna, and (c) a "porcupine" antenna.

antenna region. If the cone angles are small compared with the angle
# between the axes of the cones, the current is distributed nearly uni-
formly around each cone, and the angular wave function is

= P n (-cosO') - P n (-cos0"), (311)

where 6' and B" are, respectively, the angles made by a typical direction
with the axes of the cones. The negative sign reflects the fact that,
while in one cone the current flows away from the apex, in the other it
flows toward the apex. Proper values of n are roots of

P„(-cos^) - P„(-cost>) = 0. (312)

cone angles are not small, the current is not uniformly dis-
If the
tributed around each cone, and the wave function is of a more general
form,

6 = £ (A m co8m<p' + B m sin?V)Pn m (-cos0') -

m-0.1,2,...

E
m~0.1.2....
(A m cos trup" -f Bm sin mp") P n "(-cos 0"), (313)

where <p' and <p" are the azimuth angles associated with the cones. If
<p' and <p" are measured from the plane passing through the axes of the

cones, then B m = 0; the remaining coefficients as well as the proper

values of n have to be determined from the condition that 9 is zero on
the surface of each cone. A typical direction on the surface of one cone

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.23 VARIATIONAL METHODS 99

is ¥>') ; the axis of the cone is given by (#, 0) ;

hence, the cosine of
the angle d" between these directions is

cos 6" = cos # cos + sin t? sin ^ cos f>'. (314)

This enables us to express P„( — cos $") in terms of Next we have

m
to express P n ( — cos 9") cos rtup" in terms of tp' On substitution in .

equation 313, we shall then have a function of <p' alone which should
vanish for all values of */. Hence, if this function is expanded in a
Fourier series, all coefficients must vanish. In this way we can obtain
a system of linear equations for the A„s. It will be a homogeneous
system hence, it will have nontrivial solutions only when the deter-
;

minant of all the coefficients vanishes. This condition gives the equa-
tion for the proper values of n.
Thus, even the calculation of proper wave functions constitutes a
complicated problem. When this is solved, we still have to match the
fields at the boundary between the antenna region and the free space
region. In the case of cones of small angles, however, we need not
obtain proper wave functions nor do we have to match the fields. We
can use the first method described in Section 2.10 for thin biconical
antennas; then we can generalize the result as in Section 2.15 and
obtain a formula for the impedance of V antennas made with cylindrical
wires (or wires of any shape). The general formula 1-136 is the same,
irrespective of the angle t? between the arms of the antenna; the M
and N
functions are also independent of t? and the formula for Z a is ;

Z a = K 2 Y = R a + jX t a} k = sin

Ra = 60 Cin 2k0l +
30[2 Cin 201 - Cin 2(1 - k)fi -Cin 2(1 + k)ffl cos2# +
30[-2 Si 201 + Si 2(1 - k)0l + Si 2(1 + k)fil] sin 201, { }

X a - 60Si2fcfl + 30[Si2(l - k)0l - Si 2(1 + cos 201 +

30[Cin 2(1 - k)(&- Cin 2(1 + k)0l + 2 log(l + k)] sin 201.

Similar expressions in closed form can be obtained for the "fan" and
"porcupine" arrangements of radiating wires.

2.23 Variational methods

An important contribution to the theory of biconical antennas was
recently made by Tai.* Following Schwinger's variational method
for waveguide problems, he derived the following expression for the
terminal admittance in terms of the aperture field (the aperture is the
* C. T. Tai, Application of a variational principle to biconical antennas, Jour.
Appl Phys., 20, November 1949, pp. 107&-1084.

Original from
UNIVERSITY OF MICHIGAN
100 MODE THEORY OF ANTENNAS 2.23

part of the boundary sphere where ^ < 8 < r - <p):

2*yAJ'~* E.(9) M „(cos 6) sin B d»T

Yt -Z
n(n+
—
1)nA£
^
*E.(fi)def
+ .

2* Yt+
f* E W p
\
° {9) * (cos $) sin e
*T
? "
TTT =5 (316)

k(k + 1)N (6) del

where JV»* and N t' are the normalizing factors,

*
N" = [Af " (C0S 9) ' 2 S ' n 9 d6 N "' = [P * (C0S 9)12 8in 9 *
X' '
fo'
(317)

The summation in equation 316 is extended over the zeros of

first

M n(cos and the second summation over the odd integers. The ter-
minal admittance is thus expressed as the sum of the wave admit-
tances Yn~, Yk + of the various normal modes (as seen from the boun-
dary sphere in both directions) with appropriate weighting factors.
In the aperture we have

Ee[e) = "
ltd + ^ An iM »(cosf? )' < 318 )

where the A's are so far unknown coefficients. The first coefficient Ao,
however, be made equal to unity since Y is independent of the
may t

scale of the aperture field. Substituting in equation 316, we have

Y =Y,0+
t 2Y,*nA n + Y,0nA + IE W^.,
n
2
( 319)

where
2ft +1

i M n (cos d) P*(cos $) sin 6 dd.

2
n
0„ = n(n+ 1)AT,< Y.~

Original from
Digitized by VjOOO IC UNIVERSITY OF MICHIGAN
2.23 VARIATIONAL METHODS 101

Tim Tmi

The terminating admittance Y is stationary with respect to the varia- t

tion of E a (0). Hence, the unknown coefficients A n may be determined

from the condition dY /dA„ =
t 0. Thus,

ctn + n An + Ewt m = 0, n = m, (320)

Multiplying by A n and summing, we have

= o. (321)

Subtracting from equation 319, we obtain

Yt= Y<o + T,<*nA n . (322)

The first term on the right is Y t given by equation 77 where all

higher-order internal waves were neglected. As wo have anticipated
150

2 RESISTANCE = 39.23°
if
0125 j
1
1
1

// J
£100
<
\\
\\

a
75 Lh "
\\
iV '

ii \\
il \>
s h !
\
ID 50
u 9 il
§ ii Vv REACTANCE
il \\
V >^
in 25 S 1 ..

0.5 1.0 1.5 2.0 2.5 3.0 35 4.0 4.5 5.0

Tai, CourUrj Jour. AppL Phyt.

Fig. 2.29 Inverse radiation impedance of a biconical antenna, ^ = 39.23°; the

dotted curve is a plot of equation 77; the solid curve includes a correction term ob-
tained by Tai.

in Section 2.9, the correction term represented by the sum in equation

322 turns out to be small unless the cone angle is small. The solid curve
in Fig. 2.29 includes the first correction while the dotted curve is a plot
of Y t Q. The characteristic impedance of the antenna whose cone angle
is $ - 39.23° is about 124 ohms hence, this antenna is well matched
;

to free space > when ffl 2.

Equation 316 may be obtained from equations 56, 57, 62, and 63.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
102 MODE THEORY OF ANTENNAS 2.24

First we use thetwo equations and determine both sets of coeffi-

last
cients a„, bt in terms of the aperture field E a (9). Then we substitute
these coefficients in the first two equations which represent the magnetic
intensity in the aperture in different forms. Equating these two forms,
we obtain an integral equation for E a (B). Following this outline, we
obtain

a. = - -^j Y»-f^* -35- M"(cos 9) sin B de,

2* *
h
7 Yk+
£ lEa{9)
W" Pk<fi0se)
sin 6 M '

7(1) =£"*lE a (e)dS, (323)

y.-^Jif „(«»«)
Y' r ,-> E
S ; / a (B) dB-Z 7— —TTTT-j
;
—

YS-^-P^cosB)
= ? w+m? I EM *>
p ' (cos e) sin 6 *

Multiplying the last equation by Z? a (0)sin0 and integrating with

respect to B between $ = $ and B = tt — we obtain equation 316.

2.24 Spherical waves on infinitely thin wires

The formation of spherical waves on infinitely thin wires can be studied
with the aid of a simple solution of Maxwell's equations discovered by
Manneback.* He gives the following expressions for waves originating
at one end of an infinitely thin wire extending to infinity (Fig. 2.30),

Hp = —(-7) 4irr
*

;in
7* B = § (324)

where I(t) ia the current flowing from the end. He shows by direct
substitution that these equations satisfy Maxwell's equations when r is

large and when terms of order are 1/r 2 neglected. This solution is
regular everywhere except on the radius 8 = 0. Since the magneto-
* C. Manneback, Radiation from transmission lines, A1EE Jour., 42, February
1923, pp. 95-105.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
2.24 SPHERICAL WAVES ON INFINITELY THIN WIRES 103

motive force 2wpH p — 2rr sin 6H V approaches /

K) as 6 ap-

proaches zero, the current along this radius indeed /

is
(-7)
Manneback was interested primarily in the problem of radiation
from transmission lines in which the apparently approximate character
of equations no importance. However, these equations are
324 is of

really exact. They appeared to him to be approximate only because

in their verification the radial component of E was neglected.
The existence of this component we deduce from the fact that the
current 7(0 flowing out of the origin implies a point charge

9(0
"XI mdt> (325)

and, hence, the radial field

dE r
E r
dt
(326)
4«r2 4wr»

When this equation is combined with equation 324, we find that Max-
well's equations are satisfied exactly. It appears, however, that the
boundary condition along the wire,
where E r must vanish, is not satis-
fied. Actually Later
it is satisfied.

in this section we shall obtain the

exact expressions for a cone of
finite angle ^ and prove that the
limit of Er as the cone angle ap-
proaches zero is given by equation
326 as long as 0*0; but the
limit of E r (yp) is zero since ET ($)
is identically zero. All that hap-
pens is that the function ^) E r (r, 6 ;

, , Fig. 2.30 The wavefront of ..a wave

.......
for the finite cone approaches the
8tarting ftt the end of a BemHnfinite

limit (equation 326) nonuniformly. thin wi re .

The above equations represent

the spherical wave generated by an arbitrary current source at 0. If

1(f) is a step
function, the wave may be illustrated graphically by an
expanding spherical wavefront as in Fig. 2.30. A general time function
may be considered as a succession of step functions and thus repre-
sented by a succession of wavefronts.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
104 MODE THEORY OF ANTENNAS 2.24

Suppose that the wire is terminated at A (Fig. 2.31). First let us

consider the case in which the energy reaching A is absorbed. To
obtain the corresponding solution we superimpose on the field given
by equations 324 and 326 the following field:

'
BE r
dt
4«(r') 2
(327)

\ « c ) 1 + cos B'
E.' =
sin B'

where I is the length of the wire and r', 6' are spherical coordinates with
respect to A. The magnetic intensity H 9
'
vanishes on the radius

(a) (b)

Fio. 2.31 The wavefronts of waves originating at the ends of a finite thin wire.

B' = 7r (Fig. 2.31a) ;

hence, the original current starting at continues
to flow. On the radius 6' = the total magnetic intensity H 9 + H+
vanishes.* Hence, the continuation of OA is no longer a singular line
of the field. Equations 327 represent a spherical wave originating at A.
As this wave expands, it slides off the wire at without causing a new
wave since no current is associated with it along 8' = v. When the
current is given by a step function, the wavefronts of the waves originat-
ing at and A divide the space into three regions (Fig. 2.316) : region 1

which is field^free, region 2 where the field has reached its steady state,
and the intermediate region 3. This intermediate region contains the
energy lost by radiation.
* Note that r sin 9 = r sin 6' = p, where p is the distance from the wire.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
2.24 SPHERICAL WAVES ON INFINITELY THIN WIRES 105

Suppose now that there is no absorption of power at A and that

the wire is merely discontinued at this point. In this case the current
must vanish at A at all times. This condition is satisfied if we assume
that A is the source of current toward equal to the current arriving
from 0. This source generates the following wave:

BE/'
= _
(-1-4)
_______
dt
4*e(r') 2
(328)

I
\1
ti v
11 1 COS—
al
sinfl'
~ '
jp — rr //
-

\ttt'

The total field is the sum of equations 324, 326, 327, and 328. The
point charge at A disappears, and E/ +E" r = 0.

Fig. 2.32 Waves on a bent wire.

A bend in the wire (Fig. 2.32) becomes a source of current in the

new direction as well as the source of opposite current in the former
direction. The field will be the sum of equations 324, 326, 327, and

lit - -c - —
dE r " \ c )
dt
4*e(r') 2
(329)

1
('
H* ^
c c ) 1

sliT^— '
E$ m " H* '

f~* -^r-v,- Original from

Digitized by V^OOglLrt UNIVERSITY OF MICHIGAN
106 MODE THEORY OF ANTENNAS 2.24

where 0" is the angle made by a typical radius from A with the new
direction AB of the wire. The electric intensity due to the current
segment OA is AB
but the induced current is vanishingly
impressed on ;

small since the wire is infinitely thin. The field of the induced current
is also vanishingly small everywhere except on AB where it is just

sufficient to balance out the impressed electric intensity. For wires of

finite radius the induced current and its field are small but not vanish-
ingly small.
The point charge at the origin occurs only because we have
assumed that a charge of some kind is being driven away from it.

When the charge is merely transferred from one wire to another as at

(a) (b)

Fig. 2.33 The wavefronts when the source of waves is between the ends of a wire.

point be no accumulation of charge at the

in Fig. 2.33, there will
source. In addition to the field (equations 324 and 326) associated
with the positive, let us say, charge flowing out of to the right, we have
the following field of negative charge flowing out of to the left:

c )
dE/ \
at
(330)

_L i-c«
c
„ . _ _^ cos 6
sin 9
4*r

The total radial field is seen to vanish. waves in this Both spherical
case have the same wavefront. When one of the. component waves
reaches an open end (Fig. 2.336), a new wave is generated. New
waves are produced at the other open end as the waves from O and A
reach it.

Digitized by V^OOgK UNIVERSITY OF MICHIGAN

2.24 SPHERICAL WAVES ON INFINITELY THIN WIRES 107

Consider now a pair of parallel wires (Fig. 2.34a). Let us assume

that equal and opposite charges are driven from the ends of these wires.
The corresponding field is given by the sum of equations 324, 326, and
327, provided that in the latter set we let I = 0. The coordinates
(r, 0) and (r'r 0') are relative to the wires. The wavefronts subdivide
the space into three regions: Region 1 is external to both fronts and is
field free, region 2 is common to both fields with the steady state already
established, and the intermediate region 3 contains energy lost by
radiation in starting the current. When and 0' are small, the meridian
components of E are perpendicular to the wires, cos 6 and cos 0' are sub-

Fig. 2.34 Waves on parallel wires.

stantially equal to unity,and the denominators 4tt sin and 4tit sin 0'
are equal to 4trp and 4^', where p and p' are the distances from the
wires. Hence, for small and 0' the field in region 2 represents sub-
stantially a plane wave guided by the wires. The region of the plane
wave increases with the distance from the ends of the wires since it
depends on the smallness of and 0' and does not depend directly on the
distances from the wires. If the wires are open at A, A' as in Fig. 2.346,
we shall lose additional energy. The loss will be there even if we short-
circuit the wires or insert a resistor to absorb the incoming energy. In
the practical case of transmission of energy from one place to another,
we have a complete conductive circuit (Fig. 2.35). Here spherical
waves originate at the generator G, at each bend, and at the load L.
Thus Manneback's equations give clear pictures of radiation and
of formation of plane waves guided by parallel wires. We should
remember, of course, that these equations are exact only when the wires
are infinitely thin. However, the main difference between infinitely

Original from
UNIVERSITY OF MICHIGAN
108 MODE THEORY OF ANTENNAS 2.24

thin wires and thin wires is in the energy carried in the vicinity of the

wires. In the former case this energy is infinite in the latter case it is ;

finite but large. In both cases the radiated energy is the same (assum-
ing that the currents are the same).
We shall now connect Manneback's equations with the mode
theory of antennas. Instead of the infinitely thin wire, we shall con-
sider a thin cone coaxial with it. The field around this cone can be
expressed in terms of spherical waves as explained in Section 2.4. We
are interested in the particular field that
corresponds to a charge driven from the
apex. An equal and opposite charge is
L kept at the apex. From equation 326 we
J
see that, in the case of an infinitely thin
cone, the field of this point charge the
is

Fia. 2.35 Waves on same as in free space.

parallel In the case of a thin
wires near a point source at G. cone we expect that this field will be rela-
tively unaffected by the cone, except near
its surface. From equations 7 we find that the inverse square law is
obtained only if R(r) is finite at r - 0. When n = 0, we have

Jn (fir) = sin fir, NnoQfr) - - cos fir,

R(r) = A sin fir - B cos fir. (331)

In this case, the inverse square law is exact. From equation 9 we find
that, if n is small, the JE-function is the same as above except when r
is small ;
but, when r is sufficiently small, then so that we can neglect fi
2
,

H(r) behaves as r~\ Thus, for small values of n we have very nearly
the square law. Hence, we shall study those solutions for which n is
small. The function must be finite at B = v since along this radius
we have no singularity in the field. Therefore, 6(0) = i\(-cos0).
For small n, we have
Pn(-vosd) = 1 + 2n log sin §0. (332)
Hence,

(333)

For waves traveling from the apex B n = -jA n If we approximate .

Rn(r) by exp(-jfir), then from equations 7 we obtain

>sr*AV = -Ae + 2n log sin J*),

*,-^*l±S-i. (334)

Original from
Digitized byVjOOOlC UNIVERSITY OF MICHIGAN
 HKFKRKNCES 109

On the surface of the cone 6 = the radial component of E must

vanish; hence,

*-- »k,ifa» -
(335)

The product 2irr sin ^ H*(r, yp) must equal the current in the cone. This
product is 2xA(l + C0B.^)exp(— jfir), or approximately 4jr4exp( — j&r).
If Io is the current at the apex, 4tj4 = /o and A = 7o/4x. We have
thus determined the required field for waves of any frequency.
The exponential factor indicates that the waves travel with the
speed of light. Interpreting jta as differentiation with respect to time,
we have the following equations for any wave on a thin cone:

dE,
dt
K) A 4«r» \
_ log sin \6 \
log sin \$)
f
1

(336)

H) 4xr
1 +
sine
cosfl

E 9 = VH 9 .

As ^ approaches zero, the radial component approaches the expression

given in equation 326 for any B greater than zero. Since E r (r, ^) =
for every ^, its limit as $ approaches can picture the
zero is zero. We
radial field as a uniform field with a deep well around the cone. As the
cone becomes thinner, the well becomes narrower without ever dis-
appearing altogether.

REFERENCES
1. Theory of antennas of arbitrary size and shape, IRE Proc., 29,
S. A. Schelkunoff,
September 1941, pp. 493-521.
2. J. A. Stratton and L. J. Chu, Forced oscillations of a conducting sphere, Jour.
Appl. Phys., 12, March 1941, pp. 236-240.
3. S. A. Schelkunoff, Spherical antennas, U. S. Patent 2,235,506, March 18, 1941.
4. S. A. Schelkunoff, Principal and complementary waves in antennas, IRE Proc,
34, January 1946, pp. 23P-32P.
5. P. D. P. Smith, The conical dipole of wide angle, Jour. Appl. Phys., 19, January
1948, pp. 11-23.
6. C. T. Tai, On the theory of biconical antennas, Jour. Appl. Phys., 19, December
1948, pp. 1155-1160.
7. C. T. Tai, Application of a variational principle to biconical antennas, Jour.
Appl. Phys., 20, November 1949, pp. 1076-1084.
8. J. Aharoni, Antennae —
An Introduction to Their Theory, Clarendon Press,
Oxford, 1946.

Digged by V^OOglL UNIVERSITY OF MICHIGAN

110 MODE THKORY OF ANTENNAS
9. E. Roubine, Les recentes theories de l'anlenne, Rev. Tech., Thomson-Houston,
Paris, 1947.
10. J. R. Whinnery, The input configuration on antenna impedance. Jour.
effect of
Appl. Phys., 21, October 1950, pp. 94^-956.
11. C. H. Papas and Ronold King, Input impedance of wide-angle conical antennas
fed by a coaxial line, IRE Proc., 37, November 1949, pp. 1269-1271.
12. J. T. Rolljahn, Antennas near conducting sheets of finite size. University of
California Department of Engineering Report 162, December 1949.
13. C. H. Papas and Ronold King, Radiation from wide-angle conical antennas fed
by a coaxial lice, IRE Proc., 39, January 1951, pp. 49-51.

Original from
UNIVERSITY OF MICHIGAN
 3

SPHEROIDAL ANTENNAS

3.1 Prolate spheroidal antennas

A prolate spheroidal antenna is shown in Fig. 3.1 where the voltage is

applied symmetrically between two halves of the spheroid by means of
a biconical transmission In this case the field is independent of
line.

the <p coordinate, the current flows along the meridians of the spheroid,
the magnetic lines are circles coaxial with the spheroid, and the electric
lines lie in meridian planes. Let a and b be the semimajor and semi-
minor axes of the spheroid and I the semifocal distance. This spheroid
is one of a family of confocal spheroids given by the following para-
metric equations:*

Z = luv, p = l(u* - 1)M(1 - p2)«

u ~ cosh £, v = cos t9,
^
when u (or £) is constant. If V is constant, the surface is a hyperboloid.
The major axis of a typical spheroid is lu, so that our particular spheroid
is defined by

uo = -j = °( fl2 " b2 )~*- ( 2)

For a thin spheroid uq is slightly greater than unity. If ua is large, the

spheroid approximates a sphere of radius luo ; then the angle t? is the
polar angle 0. The parameters u, v or £, # are called spheroidal coor-
dinates.
If the field is symmetric about the major axis, Maxwell's
circularly
equations in spheroidal coordinates 1-10 consist of two independent
sets, one involving only E u E v and the other only u
, , v Ev The H H , ,
.

first case, in which the magnetic lines are circles and the currents flow in

* Applied Mathematics, p. 151.

Ill

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
112 SPHEROIDAL ANTENNAS 3.1

axial planes, is the one we are interested in. The metric form is

(3)

Hence,

(4)

e3 = p = l(u 2 - 1)M(1 -y2 )*.

Fig. 3.1 A prolate spheroidal antenna fed by a biconical transmission line.

Maxwell's equations are

1 d(pH.)
e2 Eu = -7
dv jwep du
(5)
»(*&)
du
B(eiEu )
dv
—= — jw/ieieaffr

The field intensities may thus be expressed in terms of the auxiliary

wave function,
4 = (6)
Thus

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
3.1 PROLATE SPHEROIDAL ANTENNAS 113

n, - t(u» - i)(i - *)Y*a,

-J-

where A satisfies the following differential equation,

{u2 - i)
S- + (i " v2)
^ + <* w " v2)A = °- (8)

We
note that 2tA is the magnetomotive force round a typical
magnetic line. Hence, the current in the antenna is
f(t)- 2wA(uo, v). (9)
The substitution,
A = U(u) V(v), (10)

separates the variables in equation 8, and the differential equations for

U and V are
{v? - 1) + (0W - k)U = 0, (11)

(1 - t>
2
) + (* - (PW)V = 0. (12)

Both U and V satisfy the same equation.

If we substitute

U(u) = (u 2 - 1)» U(u), V(v) = (1 - 0« F(t>). (13)

then,

- J
> + 2"4!r - (* + ^i) 17 + Wf7 °- < 14 >

"
o - "2)
4? 2"
4r + ( fc
- - = °- < 15 >

Another substitution,

U(u) = (u 2 - 1) U(u), U(u) = (u 2 - 1)« 0(u),

« (16)
F(t>) = (1 - t>
2
) FW, F(t») - (1 - V 2 )-* ?(v),

gives

(u2 - 1) + 4u-^- + (0W - A + 2)6 = 0, (17)

(I - t») ^ - 4*-^- + (ft - 2 - 0W)f = 0. (18)

_..
.. Original from
D«i!.zedby^OOglL UNIVERSITY OF MICHIGAN
114 SPHEROIDAL ANTENNAS 3.1

The form of the equation for spheroidal wave functions is used

last
by Stratton* and by Morse*. Some of the earlier writersf have dealt
with the form 15 which becomes an associated Legendre equation
when pi = 0. Page and Adams use equations 12 and 15. Abraham
and Ryder employ the form 12. The transformation equations 13 and
16 enable us to pass from one form to another.
At the upper and lower ends (i> = ±1) of the spheroid the current
vanishes; hence,
F(±l) = 0. (19)

This condition also follows from the fact that, outside the spheroid, E v
should vanish on the z axis. For most values of k, equation 12 has no
solutions satisfying the above condition ; thus, equation 19 defines the
proper values 'of k. Let Vk be the set of proper functions. At large
distances from the spheroid u is large, and equation 11 becomes

+ (PPU = o. (20)

At these distances we also have

r = Vz 2 + p
2 = lu. (21)

At such distances the field must be proportional to exp(— jfir) —

exp(— jfilu). Hence, the proper U functions are those solutions of
equation 11 that satisfy the following condition:

Uk ->e-W» as u-»oo. (22)

Thus, we have a general solution satisfying the requirements on the z

axis and at infinity,
A- £ at Ut(u) Vt(v).
*
(23)

From equations 7 we obtain the electric intensity tangential to a typical

spheroid confocal with the given spheroid,

E,{u t [(1 - - v*))-x La k Ut'(u) V k (v). (24)

* J. A. Stratton, Spheroidal functions, Nat. Acad. Sci. Proc., 21, January 1935,
pp. 51-56; Philip M. Morse, Addition formulae lor spheroidal functions, ibid., pp.
56-62. L. J. Chu and J. A. Stratton, Elliptic and spheroidal wave functions, Jour.
Math. Pkys., 20, August 1941, pp. 250-309.
t R. C. Maclaurin, On the solutions of the equation (V
4
k*)^ « in elliptic +
coordinates and their physical applications, Camb. Phil. Soc. Trans., 17, 1898,
pp. 41-108.
E. T. Hanson, Ellipsoidal functions and their applications to some wave
problems, Roy. Soc. Lond. Phil. Trans., Series A, 232, June 1933, pp. 223-283.
M. J. O. Strutt, Lamtsche- Malhieusche- und verwandle Funktionen in Physik
und Technik, Springer, 1932.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
3.1 PROLATE SPHEROIDAL ANTENNAS 115

At the perfectly conducting surface of the given spheroid u — wo the

total electric intensity should vanish,

£,(«o, «) + v) =0; (25)

therefore,
9

» 2 [(1 - **)(«o
2 - i>
2 )]"«
Ea h tV(«o) F»(p) = -tf.«(«o, ) (26)

To obtain the coefficients we shall use the orthogonality property of

the V functions which we can prove as follows. Let V k and F„ be two
solutions of equation 12 which satisfy equation 19; then,

( 27 )
(l-^)^i+(n- W)F. = 0.
Multiplying tho first equation by Vn and the second by Vk , and sub-
tracting, we have

Dividing by
(1 - v2)
4r (
(1 — v
2
v*

)
t - F'
and integrating from v
+ ( *

= —1 tov =
- n) VtV " =

1,
°-

we have
(28 >

= 0. (29)
-l
If k ^ n,

/_;<i- <r*F n rft; - 0. (30)

Multiplying equation 26 by ifil2 /jr))(u 2 - v2)

H (l - v
2 )~*
V»(v),
integrating, and using equation 30, we find

a- = *,'(«.) £ *** v)
* (31)

where is the normalizing factor

Nn - (1 - t-
2 )" 1
[F.(t)P A. (32)

A voltage Vo impressed between v = t>o and t> = t>i is the line

integral of Ev * ds v , where ds* is an element of length along the ellipse
u = wo; hence,

r.-- r**i*23-*.
-
(33)
t/po v1 v2

Original from
Digitized by VjOOO IC UNIVERSITY OF MICHIGAN
116 SPHEROIDAL ANTENNAS 3.1

If the difference vi - vo is infinitesimal,

F.-- "gzJf
VI - J
fE.<dv.
t'
2 "9
(34)

In this case,
a- Vo
- w: utm -
(35)

Substituting this value in equation 23 and then in equation 9, we obtain

the antenna current,

/fe «o) - -jVo— L ^W. (36)

If we divide the current by Vo, we shall have the transfer admittance

between y = fo and = vt

W
t>

K(„; ,„) = -,2£ E ! ^jag (37)

Aa approaches vo, the transfer admittance increases logarithmically as*

v

— log|s — so|, where s is the distance along the spheroid. This is

natural, since, in the immediate vicinity of the infinitely narrow gap
across which we have applied our voltage, the two sections of the
spheroid are substantially plane and thus form a 180° wedge. This
is true no matter how small the radius of the spheroid as long as s — so
is correspondingly small.
Suppose now that the voltage is applied by means of a biconical
transmission line across a small gap of length s in the equatorial plane of
the spheroid. The differential element of length along the meridians is

At the equator v =
ds , = lV

0; hence, near the equator

VI —^ v2
2
dv. (38)

ds, = luo dv, (39)

and the gap extends between v = -s/2luo and v = s/2luo. If the

voltage is distributed uniformly, then the current at v = s/21uq may be
obtained by integration from equation 36. Dividing this current by
Vo, we obtain the input admittance

(40)
„ j2*0l U k r»(«/g«o) /••*«-
(uo)
Vi{v)dv
n JV»^'(«o)(./Ztto)J-. s „ -

* L. Infeld, The influence of the width of the gap upon the theory of antennas,
Quart. Appl Math., 5, July 1947, pp. 113-132.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
3.1 PROLATE SPHEROIDAL ANTENNAS 117

O.OU
(b)
d/b =7070 a/b =707
0.012

0.010
q. 1 Qi
0.008

0-006
i

—TT
0.004

0.002
bi n h. '
'
'li

-/—
(

/
1

—-H j i b3 N y i

|
"
HI
-0.002 — i

i
*b 5

< -0.004 1
i

g- 0.006
* / v /
0.014

a/b =70.7 a/b =7.05

0-012

g
f
0.010

\9i t 9.
O 0.006
/
1

'T1 1

/
/
1
'
S 0.006 '
I
J i
•
/ 1

V
i

b,/ I

0.004 K i
i
I

0.002

-0.002 _
HH 1
I
\
\
\
s.
-0-004
1

-0.006
0-2 0.3 0.4 0.5 0.6 0-7 0.8 0.2 0.3 0.4 5 0.6 0-7 0.6
21 A 2l/A
Stratum and CKu. Courts* Jour. Appl. PK U :

FiQ. 3.2 The conductance gn and eusceptance b H of the nth spheroidal mode of
propagation as seen at the equator of the spheroidal antenna.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
118 SPHEROIDAL ANTENNAS 3.1

400

"°/
300 #-
u 705
2

a. 100
(a)
D
a
z ,

0.5 0.6

SfrolWn and Cnu, Courts Jour, ^ppl. Phya.

Fig. 3.3 The input impedances of spheroidal antennas of various eccentricities.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
3.1 PROLATE SPHEROIDAL ANTENNAS 119

The input admittance is thus expressed as the admittance of an infinite

number of admittances in parallel. Each admittance component
represents the reaction of the corresponding spheroidal wave on the
source at the gap. Stratton and Chu give the conductance gi and
susceptance 61 of the first mode for various ratios 2a/2b of the major and
minor axes of the spheroid (Fig. 3.2). The figure shows also the sus-
ceptances of the first few higher modes the conductances are too small
;

to be shown (except when 2a/\ is greater than unity). The figure

shows clearly that the admittance of the first mode behaves as the
admittance of a resonant circuit. The same is true of the higher modes,
but their resonances occur at higher frequencies than those shown in the
figure. The input impedance obtained by Stratton and Chu is shown
in Fig. 3.3. The conductance components obtained from these curves
should be exact, but the susceptance components are uncertain.*
It is easy to calculate the admittance of each mode at resonance.
Equation 12 has a simple solution when

k = an*, (4i)
since it becomes
d2 V
dv 2
+ 0W-O. (42)

The solution is

V = p cos 0b -f q sin 0b. (43)

Since this function should be even and should vanish at v = ± 1, we must

have g = and

cos fi = 0, 01 = (2n - 1) |- i n= 1, 2,
•
(44)

The corresponding V function is

„ =
Vn cos
(2n —
^ 1)jtu
(45)

The equation for U is the same as equation 42, but the proper function
should represent a divergent wave ;
hence,

rr T (271 • - 1)TU "I

bn = exp[-3^ (46)
2"^J
For a gap of zero length, we find, from equation 40,

(2n - I)*- 2 U n (u . .
v .

3
)

vU n '(uo)N H '

ibid.

Digitized by V^OOgK UNIVERSITY OF MICHIGAN

120 SPHEROIDAL ANTENNAS 3.1

Substituting from equation 46, we have

F
--5r (48)

The normalizing factor is obtained from equations 32 and 45 ;

thus,

Hm _
r + cos(2n - l)rv
1
dv _ aa 2(2n _ m
Jo I — XT
Hence,

Z„ = Cm 2(2n - l)x = 30 Cin 2(2n - l)ir. (50)

For the first mode we have

Zi = 30Cin2T = 73.13. (51)
In a spheroid,
a2 -P + b 2
; (52)

hence, for thin spheroids for which b <C a, I and a are substantially
equal. Therefore equation 44 becomes

thus, the resonances of the various modes occur when the length of the
spheroid is nearly equal to an odd multiple of X/2.
As approaches zero, equation 15 approaches an associated
31

Legendre equation- Hence, for low frequencies the proper values of k

are
fc£*n(n+ 1), n = 1,2,3/--, (54)
and
F.W = PnHv), V n (v) = VT^PJiv). (55)

The corresponding V functions are

n (u) = Q.ifcO, U n {u) = y/&*=lQJ(u). (56)

The normalizing factor becomes

W.-jriW W p*-*£+li. ( 57 )

Therefore,

y --^ J£tl) ^ggg^ X.. (58)

Original from
Digitized by GoOglc UNIVERSITY OF MICHIGAN
3.1 PROLATE SPHEROIDAL ANTENNAS 121

where the factor x» depends on the gap and is approximately equal to

2lu<
Xn ° Pn Pnl (59)
(2lu ) {2lu )

For the first admittance component we have

(60)

For thin spheroids this becomes

(61)

•Kf-0
The next admittance component for thin spheroids is

(62)

144 (log

In general, if we neglect the terms of order b 2 /l2 but retain those of

order (b 2/l2 ) log(2I/6), we have

Y ~Y 3<**d(2n+\)
X
— 2
(n+ l) 2

1 --g n(n+ 1)-^ log -y-

(63)
.

log-*-
21
" t
1
2 3
l + ^(»-l)(„ + 2)-g-log-f-
The xn factor is (for small a/21)

2(n/2)f

= 2n
Xn as n oo (04)

However Xn is independent of wo only if n is not too large.

From the above expressions we find that at low frequencies the
terms in the series 40 for the admittance diminish as (1/n) 2 provided ,

n is not too large. An estimate of the upper bound for n may be ob-

Original from
Digtlized by Google UNIVERSITY OF MICHIGAN
122 SPHEROIDAL ANTENNAS 3.1

tained by inspecting equation Suppose that log(2t/&) = 5; then

63.
the second term in brackets of the numerator in equation 63 will equal
approximately one tenth if n = 15. As n becomes greater, the con-
vergence of the series becomes slower; in fact, unless we keep a finite
gap, the series diverges. To prove this we shall obtain an approxi-
mation for the Un function as n increases indefinitely. For large
fc = n(n + 1), equation 11 becomes

_ «(n+l).
du 2 2
u — 1

Liouville's approximation to the solution of this equation is

Un^A-Vu2 - lexp[± Vn{n + 1) cosh" 1 «]. (66)

The proper function in our case should decrease as n increases j

hence,
we take the lower sign. Therefore,

rrT" * (67)
U»'(uo) Vn(n + 1) Vn(n + 1)/

Instead of equation 58, we now have

n(n + l)Vn(n + 1)

In the case of a vanishingly small gap x« is proportional to n (see equa-

tion 64) then Y n will be proportional to 1/n, and the series will diverge.
;

On the other hand, if we retain the finite gap, Xn becomes an oscillating

function,and the series converges. If we replace the Legendre functions
occurring in Xn by their Liouville approximations, it can be shown that

f Y n - -2j«d Ci (—) (69)

For details of the derivation the reader is referred to the paper by

Infeld already cited. If b is much smaller
than a and if s is not much
smaller than 6, then the effect of s on the admittance is negligible for ;

practical dimensions of the conductors this will usually be the case.

Nevertheless, on account of the relatively slow convergence of the
the susceptance component in equation 40, a rather large
series for
number of terms should be taken.
The major problem in connection with spheroidal antennas consists
in calculating the spheroidal functions. From the physical point of
view they are not very suitable for the treatment of forced oscillations,
except in thin spheroids near resonance but they are forced upon us by
;

the nature of the spheroidal system of coordinates. The main difficulty

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
3.1 PROLATE SPHEROIDAL ANTENXAS 123

is the complex dependence of spheroidal functions on the frequency and

the eccentricity of the spheroid. This is not surprising since the same
functions are expected to describe waves on very thin spheroids and on
spheres. Figure 3.4 shows Ryder's sketch of the behavior of the ninth

Ryder, Courtesy Jour. Appt Pkya.

Fig. 3.4 Qualitative behavior of the ninth harmonic V*9(y): (a) when 01 = 0,
(6) when 01 = 9x/2; (c) and (d) when 01 > tor/2.

harmonic Vq{v). The curve a is for pi = 0: that is, either at zero fre-
quency for any spheroid or for a sphere at any frequency. The curve b
represents the case $1 = Qtt/2 that is, the case of spheroids in which the
:

distance between the foci equals nine half wavelengths. In this case the
length of the spheroid varies from nine half wavelengths for thin spher-
oids to near infinity for near spheres. The curves c and d represent two

Original from
Digitized by VjOOQ IC UNIVERSITY OF MICHIGAN
124 SPHEROIDAL ANTENNAS

cases for which 01 > 9jt/2. Spheroidal functions have to behave in a

complex manner to enable them to represent so many widely different
physical cases. Of course, it is possible that further studies may dis-

close simple approximations. It is also likely that in some cases at least

different sets of proper functions will give simpler results. Thus we may
replace the boundary condition 19 by
V(l) = or 7(-l) - 0. (70)

Equation 12 always possesses a solution which vanishes at v = 1 and

another solution which vanishes at v = — 1, irrespective of the value of
k. The proper values of k are then obtained from the boundary con-
dition
U k (uo) = 0. (71)

The roots of this equation will be complex. The difference between the
two methods of expansion may
be explained by analogy with strings
under tension with their ends fixed. One set of proper wave functions is
formed by sinusoids these correspond to the V functions defined by the
;

boundary condition 19. These functions are very convenient for the
treatment of free oscillations. In the case of forced oscillations we cal-
culate the interaction between the proper modes and the source. But
we can also use progressive waves emerging from the source; these
waves are reflected from the ends and react on the source, and the in-
tensity of the total wave is calculated from the conditions at the source.
In the case of the string this second solution is preferable because it gives
the result in simple closed form instead of an infinite series. In the case
of waves on spheroids however, we shall have an infinite number of
modes of propagation consistent with equation 71 but some will be ;

more highly attenuated than others, and it is likely that under some
conditions the final results will be simpler. On very long, thin spheroids,
for example, we may expect waves to travel substantially with the
velocity of light and to be gradually attenuated. Hence, equation 71
will probably have a root ^/k ~
01 —jd where t? will depend on 01 and on
log(uo — 1), the latter quantity being proportional to the average
characteristic impedance in the mode theory of antennas. Studies of
this kind would be of immense value in bringing mathematics into line
with intuitive physical ideas concerning the behavior of antennas. It
is a mistake to assume that, if a particular mathematical method of

analysis of a given problem seemingly conflicts with intuitive physical

ideas about the nature of its solution, then the intuitive ideas are wrong
and that,the numerical results support the intuitive ideas in spite of
if

this apparent conflict, then the agreement is only a coincidence. Ac-

Original from
UNIVERSITY OF MICHIGAN
 3.2 OBLATE SPHEROIDAL ANTENNAS 125

tually there are infinitely many methods of mathematical analysis of

any given problem, of which we choose one or more that we happen to
know and these methods may not always be the best and most straight-
;

forward. These methods will often suffer from limitations imposed, not
by the physical nature of the problem, but by their mathematical char-
acteristics. The solution of a physical problem, for example, may be
well behaved in a given range of a real independent variable; but the
representation of this solution by a power series may be impossible on
account of a singularity in the complex plane which has nothing to do
with the physical conditions. In such a case it may happen that a
representation of the solution by a series of Legendre polynomials will
converge rapidly even though, when the power series and Legendre
;

series are terminated, we have polynomials which superficially do not

look alike. The difference is in the method of obtaining the final
coefficients of the polynomial approximation.

Fig. 3.5 An oblate spheroidal antenna fed by a biconical transmission line.

3.2 Oblate spheroidal antennas

Oblate spheroidal coordinates are defined by the following parametric
expressions for the cylindrical coordinates:

z — luv
(72)
ti ~ sinh £, v = sin < £ < « ,
__<#<_.
A typical oblate spheroid (Fig. 3.5) corresponds to a fixed value of £ or
u in the above equations. If £ = or u = 0, we have a disk of radius I ;

if u we have an approximate sphere. Just as in the case of

is large,

prolate spheroids, we can express the field in terms of an auxiliary wave

Original from
UNIVERSITY OF MICHIGAN
126 SPHEROIDAL ANTENNAS 3.2

function A defined by equation 6 ; thus,

"H
E* =
i «l - *
2
> (w2
+ y2)1
'
(73)

where A satisfies

(U2 + 1) ^~ + (1 - *
2
) ^ = + 2 M.
t> (74)

By substituting
A = U(u) V(v) t (75)
we obtain

+ W)f/ = 0, (76)

(1 - ^) iLl + (fc + 0W)7 = 0. (77)

If we write
u = j% (78)

we find that the equations for U and V become identical.

The substitutions,

U(u) = (u* + 1)« U(u), V(v) = (1 - 2

v )* V(t>), (79)

lead to

(80)

a-^)5f-2^ + (*- + swr = o. (81)

On the other hand, the substitutions,

U(u) = (u 2 + 1) U(u), U(u) - (17* + r/ (w)i (82)

F(t-) = (1 - v*) V(v), V(v) = (1 - v 2 )-X V(v), (83)

give

(w2 + !) -0- + 4u 4ir + ° 3W " *

+ 2)G = °' (84)

(1_t,2) -4v-^-+(^-2 + /3W)7 =0. (85)

"S-

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 REFERENCES 127

When 01 = an associated Legendre function of v and U of ju.

0, V is

The V functions are those solutions that remain finite at v - 1 for ± ;

large values of u, the proper U functions should vary as exp(— jfilu).

In this manner we form a solution

A = £a
*
k Uu(u) 7*0;),

E* = -jjp- [(1
- 2
f )("
2
+ ak U k '(u) F*(»),

where the proper values of k are obtained from the condition F*(±l)
= 0. Then we determine the coefficients as we did in the case of prolate
spheroids. Finally we obtain the input impedance in exactly the same
form 40 ;
only the functions are different.
Oblate spheroidal coordinates may be used to evaluate the effect of
a discoid ground on the performance of a thin monopole. We should
find the field of the monopole in free space and expand the component
tangential to the disk in spheroidal coordinates this will give the field ;

reflected by the disk. The solution of this problem requires " addition
formulas " derived by Morse, and also formulas for the calculation of
spheroidal functions obtained by Stratton and Chu. Circular plate
reflectors parallel to the antennas can also be treated in this manner.

REFERENCES
1. L. Page and N. I. Adams, The electrical oscillations of a prolate spheroid, I,

Phys. Rev., 53, May 15, 1938, pp. 819-831.

2. Chu and J. A. Stratton, Forced oscillations of a prolate spheroid, Jour.
L. J.
Appl. Phys., 12, March 1941, pp. 241-248.
3. R. M. Ryder, The electrical oscillations of a perfectly conducting prolate spher-
oid, Jour. Appl. Phys., 13, May 1942, pp. 327-343.
4. L. Page, The electrical oscillations of a prolate spheroid, II and III, Phys. Rev.,
65, February 1 and 15, 1944, pp. 98-117.
5. E. Roubine, Les recentes theories de l'antenne, Rev. Tech, Thomson-Houston,
Paris, 1947.
6. A. Leitner and R. D. Spence, Effect of a circular ground-plane on antenna ra-
diation, Jour. Appl. Phys. 21, October 1950, pp. 1001-1006.
7. W.
S. Lucke, Electric dipoles in the presence of elliptic and circular cylinders,
Jour. Appl. Phys., 22, January 1951, pp. 14-19.
8. Jj. Brillouin, Antennae for ultra-high frequencies —
wide-band antennae, Elect.
Comm., 21, 1944, pp. 257-281; and 22, 1945, pp. 11-39.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 -

INTEGRAL EQUATIONS

4.1 Nonhomogeneous Maxwell's equations

In the preceding chapters the sources of electromagnetic fields were
represented by boundary conditions. This enabled us to work with
solutions of homogeneous Maxwell's equations (that is, equations ad-
mitting the trivial solution E— H— 0) for source-free regions. We
can, however, alter our point of view and express the field intensities in
terms of the conditions existing inside the true and the virtual sources of
the field. A true source is a region that contributes energy t o the field ,

and a virtual so urce is a region that acts on energy borrowed from the
t rue source . In regions including sources Maxwell's equations are
nonhomogeneous, . y

curl E = -jwfiH - M, curl H = ju*B 4- J, (1)

where J and M
are the densities of electric and magnetic currents other
than those given by the first terms on the right-hand side of the equa-
tions. In this form of Maxwell's equations the electronic currents in
vacuum tubes (true sources) and the conduction currents in antennas
.(virtual sources) are all included in «/. If there are any dielectric media,
the polarization currents are included in J. If there are any magnet-
izable media, the magnetization currents are included in Loop cur- M .

rents and double current sheets may be represented either by J or more

conveniently by M.
The field intensities may be expressed in terms of J and M as
follows*
E = —jwpA — grad V — curl F,
(2)
H = curl A - grad U - joxF,
* Electromagnetic Waves, pp. 126-132.
128

Digitized by V^OOgK UNIVERSITY OF MICHIGAN

4.2 SPECIAL SOLUTIONS ^ 129

where

where q v m v are the volume densities of electric and magnetic charge, and
,

r is the distance between a typical element of current and charge and a

typical point of the field. The charge densities are not independent of
the current densities since the current passing out of a given volume must
equal the rate of decrease of the charge in the volume. Hence, the
scalar potentials V, U must depend on the vector potentials A, F. In
fact,
div A = -jueV, div F = -jupU. (4)

If the currents are confined to surfaces, we should replace the volume

integrals (equations 3) by surface integrals with the corresponding
changes in the moments J dv, M
dv of the current elements and in the
elements q v dv, m v dv of electric and magnetic charge. Thus, the moment
of a typical electric current element will be J dv = C dS, where C is the
linear density of current. If currents are confined to lines, then J dv
= I dpi where / is the current.
Equations 1 are useful for media that are almost everywhere npn-
dissipative. While they are perfectly general, they are useless in the
case of antennas in sea water, since in that case the conduction currents
exist everywhere,and the integrals 3 must be taken throughout the entire
medium. In dissipative media we should subtract the conduction cur-
rent density gE from J, where g is the conductivity of the medium.
Then in the antenna we shall have only that part of the conduction
current density which is in excess of gE.

4.2 Special solutions

If all electric currents are parallel to the z axis and if there are no mag-
netic currents, the magnetic vector potential A is parallel to the z axis,
and the electric potential F vanishes. From equations 4 we have

__ = -joxV. (5)

From equations 2 we have

Original from
Digitized by GoOglc UNIVERSITY OF MICHIGAN
130 INTEGRAL EQUATIONS 4.3

In particular, for a current element J dv, we have

where

(8)

Hence, for any distribution of currents parallel to the z axis,

(9)

In the case of a current filament extending from z — z\ to 2 = Z2,

the moment J equals /(£) df. Since

r = [(*-*) 2 + 2
P ]* (10)

where p the distance from the axis of the filament, fofr/dz = — d\fr/d$
is

and dH/dz 2 = dH/df. Integrating the first term in the integral 9

by parts, we find

" il ' ) *"

*-Tgsr C(-S- +,w)* <
(f

In the course of this integration we must assume that /(£) and its de-
rivative are continuous in the interval {z\, z%).
If the current is distributed uniformly round a cylinder of radius a,

J dv = /({) d(.d^/2r t
and

A. =
±f* G(t - z, P) di, (12)

E
<=^r& + * 2G
)
md s> <i3)

where

r
--iX
= [ft
1

- *)
V*
P 2 * e-*&

+
2
p
2 - 2pa cos *> + a 2]*.
(14)

There is also an equation similar to equation 11 except that G replaces

4.3 Equations for surface currents

Suppose that we have one or more surfaces capable of supporting electric

currents. Let u, v be the coordinates of a typical point on a typical sur-

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 4.4 EXTERNAL AND INTERNAL OSCILLATIONS 131

face. Let C u (u, v) and C v (u, v) be the components of the linear current
density. From equations 2 and 3 we can obtain Eu (u, v) and E v (u, v) at
the surface. If the components of the impressed electric intensity are
EJiu, v) and £V'(w, v) and if the surface impedance is Z(u, v), then,

Eu + Ej = ZC ut E + Ev = ZC
v * 9 . (15)

In particular, if the surfaces are perfect conductors, Z = 0, and

E% = ~EU \ E v = -E 9 \ (16)

These are integral or integrodifFerential equations for C u and C,.

Consider, for example, a perfectly conducting hollow cylinder coaxial
with the z axis, and assume that the impressed field Ej{z) is distributed
uniformly round the cylinder. From equation 13 we have the integral
equation of the first kind,

f* ("U + P 2G) *= - 4 "** E ' i{z) >

(17)

and from equations 6 and 12 the integrodifferential equation,

frJT G( * z ' a) m d* + 02
f
H
G( *
"*
a) Ia) *
= - 4xjcoe E*{z). (18)

Equations type 17 were used by Pocklington* to show that on thin

of the
wires (straight or bent) electric waves travel approximately with the
velocity of light, and that the current distributions are approximately sin-
usoidal. These have been extremely important in the develop-
results
ment He also obtained the damping constants for
of radio engineering.
free oscillations in circular loops. Hallen's theory of antennas is based
on the integrodifferential equation 18, which can readily be reduced to an
integral equation of the first kind with some arbitrary constants of in-
tegration.

4.4 External and internal oscillations on closed surfaces of

revolution
Consider a perfectly conducting surface of revolution (Fig. 4.1). Cir-
cularly symmetric excitation can be obtained in several ways. One or
more toroidal solenoids, coaxial with such an antenna, can be used to
generate an impressed field which will drive electric currents along the
meridians of the antenna. Since the field cannot penetrate a perfectly
conducting surface, we must have either internal or external oscillations,

* H. C. Pocklington, Electrical oscillations in wires, Camb. Phil. Soc. Ptoc,

9, October 25, 1897, pp. 324-332.

Digged by V^OOgK UNIVERSITY OF MICHIGAN

132 INTEGRAL EQUATIONS 4.4

depending on whether the driving solenoids are inside or outside the

antenna. We can also cut a circular ribbon out of the antenna, be-
tween two circles of parallel, and imagine that, in the gap previously
occupied by the ribbon, there exist electric forces which
transfer electric charge back and forth between two sec-
tions of the antenna, without any opposition other than
that of the charges and currents in the antenna. That
is, we assume that the internal impedance of the gap
is zero. If the internal impedance of the gap is finite,

it can later be added in series with the input impedance

of the antenna. In this method of driving, the external
and internal regions are connected through the gap, the
total current in the antenna equals the sum of the
external and internal currents, the external and inter-
nal regions are electrically in parallel, and the input
Fia. 4.1 An an-
tenna of revolu- admittance equals the sum of the external and internal
tion. input admittances. Thus, if we assume that the im-
pressed field exists only in the gap and derive equations
of the types represented by equations 17 and 18, we shall obtain equa-
tions for combined external and internal oscillations.
In the case of spheroids, external and internal oscillations are easily
distinguished by the type of characteristic functions. For the external
oscillations these functions must obey certain conditions at infinity, and
for the internal oscillations the characteristic functions must be regular on
the axis. It was shown by Synge* that a similar separation between ex-
ternal and internal oscillations can be made in integral equations. In
addition his method leads to simpler general equations. Thus, for a
circularly symmetric antenna in a circularly symmetric field Synge's
equation is

JT(A -•»)'»*
where
n L J \<*f GOT***
is the tangential component of the impressed electric
09)

in-
tensity, p{z) is the radius of the antenna, and

*--*-T-> rm {({-*)* + [p(f)]»}K (20)

* G. E. Albert and J. L. Synge, The general problem of antenna radiation and

the fundamental integral equation with application to an antenna of revolution:

Part I, Quart. Appl. Math., 6, July 1948, pp. 117-132; Part II, by J. L. Synge. pp.
133-156.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
4.4 EXTERNAL AND INTERNAL OSCILLATIONS 133

Equation 19 gives external oscillations if it is satisfied in the interval

(zi, 22), and internal oscillations if it is satisfied outside the interval

(21,22)-
The simplest derivation of this equation is based on the equivalence
theorem.* Suppose that we know the field produced by an externally
impressed voltage. By the equivalence theorem this field may be cal-
culated from certain electric and magnetic current sheets assumed to
coincide with the surface of the antenna. This theorem also states that
the field of these sheets is identically equal to zero in the interior of the
antenna. The foregoing equation expresses this condition on the axis of
the antenna. Had we actually known the field at the surface of the
antenna, this equation would be merely a check on the correctness of the
field but, since we know only the electric field (and therefore the equiv-
;

alent magnetic current sheet), we have instead an integral equation for

the electric current.
Let the tangential component of the impressed electric intensity at
z = £ be the intensity produced by the antenna current will then
;

equal — 2£ In accordance with the equivalence theorem the density

4
(f).
of the equivalent circulating magnetic current is 9
= The M
field of this current sheet must be added to that of the electric current

/(£) in the antenna. By symmetry the radial component of the electric

field vanishes on the axis, and we need to evaluate only the axial com-
ponent.
The vector potential due to the electric current sheet is

A = '
~^f HO f ft z * ) (21)

For the scalar potential we have

v= ~hSm H(i * z) (22)

where q is the density of charge per unit length along the axis. Since
dl/d( = -jwq,

7 --T3=r/#** (23)

Integrating by parts and noting that the current vanishes at the ends of
the antenna, we find

* The derivation by Albert and Syngc simultaneously includes a proof of the

equivalence theorem.

Original from
Digitized by VjOOO IC UNIVERSITY OF MICHIGAN
 INTEGRAL EQUATIONS 4.5

Hence, the total electric intensity due to the electric current sheet is

Dting the formula for the field of a magnetic current element, we

find that on the axis of the antenna the field due to the magnetic current
aheet of density 3/, = is

Equating to zero the sum of equations 25 and 26, we obtain equation 19.

4.5 Integral equations and Kirchhoflfs network equations

The resemblance between integral equations of the type 17 and Kirch-
hofTs network equations,

if Zm Jn = Vm\ m- 1,2, 3,—, N, (27)

n-I

is evident. The kernel jG/^voje (if we divide equation 17 by — 4ir;aje)

corresponds to the transfer impedance Z mn in the network equations.
The impressed electric intensity E t
{
(z) in equation 17 corresponds to the
voltage Vm
impressed on a typical mesh of the network. If we ap-
*

proximate equation 17 by a finite sum and multiply both sides by Az n we ,

obtain equations in the exact form 27. This analogy is not restricted to
Htraight conductors. We can have any system of conductors, some
wound into coils andsome'forming "capacitors," and still bo able to write
for them a network type of integral equation. But the equation will be
very complex.
If, however, the dimensions of the conductors are small compared
with X, then the integral equations assume automatically the algebraic
form 27. What happens is that the entire system of conductors can be
subdivided into sections in each of which the current is substantially the
same at all and then 7(f) can be taken outside the integral sign.
points,
If we integrate the original integral equation over each such section, the
integral of the impressed intensity becomes the voltage impressed on the
section. The integrals remain only as coefficients Zm n independent of ,

the currents. For further details concerning the relationship between

integral equations and Kirchhoff's equations, the reader is referred else-
*
where.

*J. Aliaroni, Antennae — an Introduction to Their Theory, Clarendon Press,

Oxford. 1040.

_,. .. Pr»r»rtI/> Original from

Digitized by V^OOglL UNIVERSITY OF MICHIGAN
 4.6 KING SOURCE, GREEN'S FUNCTION 135

4.6 King source, Green's function, and the input impedance

Since our equations are linear, we can consider any distribution of the
impressed electric intensity as due to the superposition of "delta dis-
tributions." In equations 17 and 18, for example, we may assume that
Es(z) vanishes everywhere except in the interval ({* — $s, { §s), +
where s is infinitesimal. This represents a ring source. If we assume
that

lim r+ *'
EM *(z) dz = 1, s -> 0, (28)

the corresponding solution of equations 17 and 18 is the Green's func-

tion I(z, f) of our problem. The solution for any given distribution of
impressed electric intensity is then

lb)* f*I(z,[)E,*(t)di. (29)

The Green's function is symmetric, of course, I(z, f) = /(f, z). At

z = it has a logarithmic singularity ;
thus,

/«-,*)* iogir-*i, (30)

as £ — z approaches zero. Hence, for a ring source the input admit-

tance is infinite and the input impedance vanishes. For a ribbon source
of finite width the current (equation 29) and the admittance are infinite.
It is to be noted that, since the vector potential is finite, its con-
tribution to the integral 28 is zero in the limit and condition 28 is equiv-
alent to a unit discontinuity in the scalar potential,

F(r + 0) - V(S-0) = 1. (31)

Throughout this section we have tacitly assumed that the radius

does not vanish anywhere along the antenna. In Chapter 2 we have
seen that, at a point where the radius is zero, the Green's function is

finite, and, hence, the input impedance is finite. In this case the ring
source becomes a point source, and it is the only case in which the dimen-
sions of the source do not enter the equations. It may be noted that it
is easy to ascertain the nature of the fields in the immediate vicinity of
a point source and a ring source in terms of appropriate solutions of Max-
well's differential equations; but to do the same by using integral equa-
tions is very difficult. In equations 17 and 18 the kernels and the ex-
pressions on the right-hand side are singular at the ring source at z =£;
here at least we have a warning of the unusual conditions. But in equa-
tion 19 the kernel is regular, the right-hand side is regular, and there is

Original from
Digitized by VoOO^lC UNIVERSITY OF MICHIGAN
136 INTEGRAL EQUATIONS 4.8

nothing in the equation to indicate a logarithmic singularity in its solu-

tion I(z, for a ring source at z = f

4.7 Solutions in terms of orthogonal functions

Some integral equationsbe solved with the aid of orthogonal ex-

may
pansions, Fourier series, for example.* Consider equation 17 for the
hollow cylindrical antenna, and let the antenna extend from z = to
z = 22. Let

I(z) - f J„ sin ^
n-l
, EKz) = tE
n-1
n sin ^ **
(32)

The sine series for 7(z) is chosen in order to satisfy the end conditions

7(0) = 7(20 = 0. (33)

Substituting from equations 32 in equation 17, multiplying by

s\n(mvz/2l), and integrating from z = to z = 21, we obtain

t Z mn I n = -4irj»dE m , m = 1, 2, 3, •

—i (34)

' mn
Jo (^ + ^) sm ^r 8in
-2r'i?£fe -

Hence, the integral equation has been reduced to an infinite system of

linear algebraic equations.
The evaluation of Z mn is difficult. If the impressed electric in-
tensity is concentrated in a small interval as it is in a transmitting—
antenna — a very large number of terms must be taken in the Fourier
a satisfactpry approximation. In a re-
series for E*(z) in order to obtain
ceiving antenna the load is highly concentrated, and we must again take
a very large number of terms.

4.8 Variational theorems

Schwinger found that the impedance of certain waveguide discontinu-
ities possesses the extremal property. This property is shared by more
general functions. First, let us consider the finite network equations
27 and form the following function:

^=11 Z n J m I
77i n
n . (35)

• F. H. Murray, On the numerical calculation of the current

in an antenna,
Amer. Jour. Math., 53, 1931, pp. 873-890.
F. B. Pidduck, Currents in Aerials and High Frequency Networks, Clarendon
Press, Oxford, 1946.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 4.8 VARIATIONAL THEOREMS 137

Multiplying equations 27 by I m and summing over m, we find that, when

these equations are satisfied,

* = Em FV/.. (36)

We also find that, if we keep the F's fixed,

*(* - 2£ Vm*Im ) - 0. (37)
tn

To prove this we take the variation of 4> from equation 35 and note that
Z mn = Z nm .

If all the Vs except V\ vanish, the impedance to V\ is

(38)

If we keep I\ fixed,
SZi = 0. (39)

In the case of the hollow cylindrical antenna, we form the following

integral analogous to the series in equation 35,

* - iirX' (S? + **)

!

£ m 1{z)
* *• (40)

where G(( — z, a) is obtained from equation 14. Since G is an even

function of £ —
z, it is a symmetric function of £ and z. Multiplying
equation 17 Ijy I(z) and integrating over the antenna, we find that, if

I(z) satisfies equation 17,

*- f&fy) l{z) dz. (41)

We also find that, if l(z) is the solution of equation 17, then,

a[*-2^tfK*)/(s)<fe] = 0, (42)

where in obtaining the variation we keep E*(z) fixed.

Suppose now that E*(z) vanishes everywhere except in the input
interval (f — Ja, f + £s) and is such that, in this interval,

1(g) = /. = constant. (43)

In this case the input impedance,

is stationary with respect to variations in I(z), subject to the condition

that the input current is constant.

Original from
Digitized by VjOOOIC UNIVERSITY OF MICHIGAN
138 INTEGRAL EQUATIONS 4.9

If we substitute from equation 32 in equation 42, we obtain equa-

tion 34.
Equation 44 had been used long before its extremal property be-
came known. It was first obtained by the induced emf method, using
the reciprocity theorem.* There is a similar expression based on the
induced emf method and the energy theorem,

*=
i^?r r + * 2g
)
/(?)
™* *' (45)

which has also been used frequently but which does not possess the ex-
tremal property. When the simple sinusoidal approximation to the
antenna current is used in equations 44 and 45, the results are the same.
Equation 44 is very useful when a reasonably good approximation
to the antenna current is known. One should be careful, however, not
to draw unwarranted conclusions from it. For instance, if the influence
of the length of the input interval (the antenna gap) on the impedance
were not known beforehand, it would be difficult to deduce this effect
from equation 44. We must be careful not to conclude that the effect is
absent if we neglected it in using the equation.

4,9 Asymptotic solutions

Hallen has obtained important asymptotic solutions of the integrodiffer-
ential equation 18. He solves the differential equation and then notes
that, as the radius a of the cylinder approaches zero, the integral tends to
become proportional to I(z). The coefficient of proportionality tends to
infinity and hence contains an indeterminate constant. The choice of
this constant affects the usefulness of the associated asymptotic expan-
sion of I(z) for practical values of the antenna radius.
In obtaining Halten's asymptotic solutions it is necessary to assume
that the antenna current is finite everywhere. Hence ring sources must
be explicitly excluded, and the applied electric intensity must be dis-
tributed over a finite interval. The principal effect of the length of this
interval on the antenna current and on the antenna impedance is ex-
cluded automatically by the nature of the transformations involved. In
the limit this effect disappears ; but, for antennas of finite radius, the
asymptotic expressions for the antenna impedance should be used only
when the length of the input interval is not much smaller than the radius
and when it is known from other considerations that the effect of this
length is sufficiently small. As far as the antenna current is concerned,
the effect of the length of the input interval is confined to an interval
• P. S. Carter, Circuit relations in radiating systems and applications to antenna
problems. IRE Proc.. 20, June 1932, pp. 1004-1041.

Original from
UNIVERSITY OF MICHIGAN
 REFERENCES 130

comparable to the antenna radius in the immediate vicinity of the input.

The ambiguity associated with neglecting the length of the input inter-
val affects the input susceptance but not the conductance.
The essential details of HalleVs method are given in the next
chapter.

REFERENCES
1. H. C. Pocklington, Electrical oscillations on wires, Comb. Phil. Soc. Proc, 9,
October 25, 1897, pp. 324-332.
2. E. Hallen, (Jber die elektrischen Schwingungen in drahtformigen Leitern,
Uppsala Universitets Arsskrift, 1930, no. 1.
pt
3. F. H. Murray, Conductors in an electromagnetic field (E°e ,
H°e F '), Amer.
Jour. Math., 53, April 1931, pp. 275-288.
4. F. H. Murray, On the numerical calculation of the current in an antenna,
Amer. Jour. Math., 53, October 1931, pp. 873-890.
5. L. V. King, On the radiation field of a perfectly conducting base insulated antenna
over a perfectly conducting plane earth, and the calculation of radiation re-
sistance and reactance, Phil. Trans. Roy. Soc. (Lond.), Ser A, 236, November 2,
1937, pp. 381-422.
6. E. Hallen, Theoretical investigations into the transmitting and receiving quali-
ties of antennas, Nova Acta (Uppsala), 11, 1938, no. 4.
7. J. F. Carlson and A. E. Heins, The reflection of an electromagnetic plane wave
by an infinite set of plates, I and II, Quart. Appl. Math., 4, January 1947,
pp. 313-329, and 5, April 1947, pp. 82-88.
8. A. E. Heins, The radiation and transmission properties of a pair of parallel
plates, I and II, Quart. Appl. Math., 5, July 1948, pp. 157-166, and October

1948, pp. 215-220.

9. J. W. Miles, The diffraction of a plane wave through a grating, Quart. Appl.
7, April 1949, pp. 45-64.
Math.,
10. J. W. Miles, On certain integral equations in diffraction theory, Jour. Math.
and Phys., 28, January 1950, pp. 223 226.
11. H. Eine strenge Berechnung der Dipolantennen mit rohrformigem
Zuhrt,
Querschnitt, Frequenz, 4, June 1950, pp. 135-141.
12. A. E. Heins, The reflection of an electromagnetic plane wave by an infinite set
of plates III, Quart. Appl. Math., 8, October 1950, pp. 281-291.
13. H. Levine and C. H. Papas, Theory of the circular diffraction antenna, Jour.
Appl Phys., 31, January 1951, pp. 29-53.
14. J. E. Storer, The impedance of an antenna over a large circular screen, Jour.
Appl Phys., 22, August 1951, pp. 1058-1066.
15. J- Aharoni, Antennae — An Introduction to Their Theory, Clarendon Press,
Oxford, 1946.
16. F. B. Pidduck, Currents in Aerials and High Frequency Networks, Clarendon
Press, Oxford, 1946.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 5

CYLINDRICAL ANTENNAS

5.1 Fundamental equations

In this chapter we shall present HalleVs method of obtaining asymptotic
solutions of the integrodifferential equation 4-18 for hollow cylindrical
antennas. Solving the differential equation for the integral, we have

Ae-if + Be»' + ^ sin /J<{ - z) <*«, (1)

where z = zq is a point on the antenna, and

G« - z) = ±-£'r-i e -'e' d*, r - [ft

- *)* + 4a* sin* (2)

The arbitrary constants of integration A, B can be determined from the

boundary conditions
I(zx) = 0, I(z 2 ) = 0, (3)

after the general solution of equation 1 has been found. For hollow
cylindrical antennas these conditions are exact. In the immediate
vicinity of either end
antenna the charge per unit length varies
of the
inversely as the square root of the distance from the end hence, the ;

charging current must vary directly as the square root of this distance.
The integral on the left in equation 1 equals the vector potential
multiplied by 4t. Using equations 1 and 4-5, we obtain

V= ZOAe-V' - 30Be">>
+£ cos p($ - z) d$. (4)

140

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
5.2 PROPERTIES OF THE KERNEL 141

If E*(() = everywhere except at $ = f,

H
f G(i -z) /(f) df = Ae-«' + 7 = 304*-"* - 30Be"'f
2,<2<f '
(5)
= Ce-"" + De'*', = 30Ce-"* - MZto"*,
f < 2 < 22.

The vector potential is continuous. Hence, for a unit discontinuity in

the scalar potential we have the following conditions across the ring
source at z =
Ae -m + BeP* = Ce~M + &/*,
Ce-m - DJn - Ar* + Bet* = ~ (6)

In particular, if z\ — —l f
Z2 = I, f = 0,

- - P cos 0t - ±
xJ (?({ «) /«) <$ sin (7)

If the antenna and if the energy is supplied to it by a trans-

is thin
mission line consisting of two wires perpendicular to it, the impressed
intensity arises solely from the charges on the transmission line (since
the vector potential due to the line currents is perpendicular to the
antenna). Therefore,

^. (8)

where V\ is the potential due to the charges on the line.

5.2 Properties of the kernel

The kernel (equation 2) may be expressed as follows:

~ = ~l " ~ " 6~"r)

z)
if?*
r
Ho* r 1(1 (9)

If the radius a of the antenna is small compared with its length, then
r c~ |j — z\, except in the vicinity of £ = z. If 0a <^C 1, then, in this
vicinity the last integrand in equation 9 equals jft except for small
quantities of order (0a) 2
Hence, in this term we can approximate r by
.

|£ —
z\. The first term on the right in equation 9 may be transformed
into a complete elliptic integral of the first kind. Thus, we find
" *> =
F(l*> « - - zl-'U le - exp(-#|{ - *|)] + O03V), (10)

_.. .. /~* <-».-»,-* 1 Original from

Digitized by l^OOglC UNIVERSITY OF MICHIGAN
142 CYLINDRICAL ANTENNAS 5.3

where
k = 2o[(f - 2)
2
+ 4a )-« 2
(11)

is the modulus of the elliptic integral F. As £ approaches z, the modulus

approaches unity, and

F(iT, k) —* log 4(1 - k 2 )~X

= log 4[(f - z) 2 + 4a 2 )H - log |£
- z\. (12)

Finally, in the immediate vicinity of £ = z, F(%t }

k) tends to log 8a —
Iog|£ - z|, and

Hence, the kernel is logarithmically infinite at £ = z for all values of a,

and approaches infinity almost as l/a when a approaches zero.

If |£ - z\ » a, then,

<?(£ - *) - -L |{ - + 0(a 2 ). (14)

Therefore, as a approaches zero, the kernel approaches values inde-

pendent of a everywhere except at distances from { — z comparable to a.

5.3 Mullen's method of iteration

Hallen's method based on the
of iteration for solving equation 1 is

observation that, as the antenna radius approaches zero, the value of

the integral on the left side is determined increasingly by the values of
/(£) in the vicinity of £ = z. Hcnco, if /(£) does not vary too rapidly
in this vicinity, we can separate the integral into two components one
of which constitutes a " principal part of the integral,"

= <?(£ - *) d£ - JT" <?(£ - z)[I(z) - /(£)] d£. (15)

The coefficient of I[z) in the principal component increases as a de-

creases. The second component is substantially independent of a. The
decomposition of the integral into components behaving in this manner
is not unique since the kernel also may be split into two components

one of which is substantially independent of a. Hence, we can write

equation 1 in the following form,

(16)

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 5-3 HALLfiN'S METHOD OF ITERATION 143

where S is the following linear functional of I(z):

.S[/(2)l = W(z) /(*)

''
G(i - z) /({) rf{, (17)
-f
and/(z) is the right-hand side of equation 1. The function W(z) is so
chosen that S[I(z)] is substantially independent of a. In particular,
it is possible to choose

W(z) = 21og(^^U A, (18)

where A a rather arbitrary constant.

is

Let us choose a sequence of functions Iq(z), h (z), • •

; defined by the
following equations:
W(z)I (z)=f(z),
W(z)I l (z) = S[Io(z)) t

W(z)I 2 (z) = SUxiz)),

Adding, we find that the series

I(z) = h{z) + h{z) + h(z) +•• (20)

is a formal solution of equation 16. In particular, if W is a constant,

then,

J(s) = W~if(z) + W-2 S\f(z)] + W-* S 2 \f(z)] + -

-, (21)

where S n denotes the n times repeated operation 17.

If
W= Wi + A, (22)
then,

w- = Wr {i+AWr
x l l )- l
= Wr*-AWr2 +A 2 Wr* ,

w- 2 =Wr 2 (i+AWr w -i
l )- 2 =Wr -2AW 2
l
-3
+ZA 2
l ,

(23)

W-*-Wr*(l+AWr 1 )-*

= Wr»-nAWr n - l
+ Mn-i)A 2 Wr»-2 -
Substituting in equation 21, we find the functionals associated with the
new choice of the expansion parameter i, W
&1/M1 = mz)} - Aft*),

SxV(z)\ = S 2 [f(z)\ - 2AS{f(z)\ + A 2f(z) t (24)

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
144 CYLINDRICAL ANTENNAS 5.4

5.4 Special asymptotic expansions

It has been noted that the representation of the integral equation 1 in

the form 16 is not unique. The part of the integral on the left side of
equation 1 which depends primarily on a is expressed by (z), and W
S[I(z)] approaches a bounded function as a approaches zero. Any
bounded part of W
(z) I(z) may be transferred to S. An obvious choice
of W and S is suggested by equation 15. Thus,

Wi(z) -JT" G(( - f) «- Q(z) + *(*),

"w-iirjr^** (25)

Except for quantities of the order a 2 we find ,

B(«) = log([(2 - *,)* + a^]H +(z- zi)) +

logl Kit - *)
2
+a 2
]* + (z2 - z) I
- 2 log a,

*(*) = - Cin/3(z - ti) - Cin/3(22 - z) -jSifi(z - z%) -

j&i0(z2 -z).
If « is not too close to either z = 2i or z = 22,

fl( z )^log «(»-«j(«-'»> .

(27)

At either end,

Q(zi) = Qfe) = log

2fe ~ 2l)
- (28)

In the last term in equation 15 we can approximate r by |f

— «| ; thus,

&[/«] -J^Gtt - *) I'M - /(f)I *

aejT'UW - /(*)]!• - fl"
1
exp(-#|« - f|) (29)

To examine the behavior of Q(z) let us take zi = — I, za = I; then,

except in the vicinity of the ends,

0(*)-Q + Iog[l (30)

Original from
Digitized by VoOCVMC UNIVERSITY OF MICHIGAN
5.4 SPECIAL ASYMPTOTIC EXPANSIONS 145

where

o-aiog-2L '

(31)

At each end U(z) = £fi + log 2. Hence, 0(z) varies slowly over the
greater part of the interval (— I, I). Likewise ¥(z) is a slowly varying
function; thus,

¥(0) = -2Cinfl - 2jSifl

= -2C - 2 log # + 2 Ci 0Z - 2j Si fit,

(32)
= = -Cin2# - jSi2#
- -C-log2#-jSi2#,
where C= 0.577 ••• is Euler's constant. The difference [Wifc) —
Wi(0)] I(z) is numerically less significant than S[I(z)]. Since the
evaluation of S[I{z)] is much simpler (analytically at least) when W(z)
is constant, it is only natural to include the difference in S and choose

W 2 (z) = W :(0) =
r
2 log - 2 Cin /SI - 2j Si A (33)

&[!(*)] = ft[T«] + [Fi(0) - TFi (*)]/(*).

In his second paper Hallen expressed the second term in the series
20 and the corresponding input impedance in terms of sine and cosine
integrals on the assumption that W(z) = fl. King and Blake 16 * pre-
sented tables and curves based on these expressions. Bouwkamp 4
evaluated numerically the effect of the third term in the series 20 on the
input impedance and discovered that it was large for = 10, 15, 20.
It should be noted that 12 = 20 corresponds to an extremely thin antenna
(a = 2J/22026). A comparison with the mode theory and experimental
evidence6,8 shows that Hallen's formula gives excessively high values
for the first antiresonant impedance even if three terms of the series 20
are used. Gray 7 observed that choosing

W3 (z) = re Wi(0) - O + re ¥(0) (34)

instead of W(z) = was HalleVs formula for the

sufficient to bring
impedance into substantial agreement with the mode theory and with
experimental results, even if only the first two terms of the series 20
were used. The still existing differences were reasonable in view of the
fact that the theoretical approximations were obtained by entirely
different methods (although to the same order of approximation), and
well within the probable accuracy of experimental results.
* The numbers refer to the list of references at the end of the chapter.

Original from
Digitized by VjOQSIC UNIVERSITY OF MICHIGAN
146 CYLINDRICAL ANTENNAS 5.4

King and Middleton 10 attempted to choose W(z) so that the

second term in equation 16 would be as nearly equal to zero as possible.
Since I(z) is unknown, they tried to make the second term small when
I(z) was replaced by its first approximation. In the case of a symmetric
transmitting antenna the current approaches asymptotically the form
Iq sin p(l — \z\), and they chose

W*(z) - [sin - \z\)]~

l
G<{ - z) sin 0(Z - |{|) (35)
f^
If I > X/2, this function becomes infinite at some points which in-
validates equation 16 at these points. In any case equation 35 repre-
sents an extremely complicated function* which would make the
evaluation of the various terms in the series 20 difficult. Thus, while
starting with equation 35, King and Middleton used eventually a con-
stant parameter:

W-W 4 (0) = (sin/M)" 1

<*-*
sin 0(1 - |*|) df, I < IX;
(36)

= WA (l - iX) =f_G{$ - I + JX) sintft - |*|) I > JX.

In the vicinity of the first critical length I = X/2, this parameter is very
close to Gray's parameter even though the definitions 34 and 36 are
quite different. For this reason the values of the antiresonant im-
pedance obtained by King and Middleton are comparable to those
obtained by Gray and considerably smaller than the values obtained
from W = U.
Siegel and Labus evaluated the left side of equation 1 for the first

approximation, 7(£) Jo sin 0(Z = —

|||), to the current in the symmetric
transmitting antenna, and found that a substantial part of it contains
the factor Jo sin &(l - \z\). The coefficient of proportionality for
z > 0,

Wt(z) -
i[log 4z(l ~ z) - logtfV) + Ci 2/32 + Ci 2fi(l ~ z) - 2C], (37)

ifused in equation 16, will cancel in S[I(z)] a major part of the contri-
bution due to the sinusoidal component of I(z). We could level this

•J. Labus, Rechnerische Ermittlung der Impedanz von Antennen, Hoehfreq.

u. Electroak., 41, January 1933, pp. 17-23.
E. Siegel and J. Labus, Scheinwideretand von Antennen, ibid., 43, May 1934,
pp. 166-172.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
5.4 SPKCIAL ASYMPTOTIC EXPANSION'S 147

function by taking its average value,

W 5 = 2Iog-|--Cin2fl- 1 -
^J^- (38)

In the case of progressive waves on a wire extending from z = to

z = I, we have*
WJo(z) = Ae-"«
WV/ito = W6 S[Io(z)]
2
= AW e-"* - Aj'f
6
*
(2*r)-l e-"'-'« tf* (39)

= Ar*|V. - Ei ^{1 - * ) + Ei JgL] -

To make the squares of the real and imaginary parts minimum, we take

We = av |jEI 2j/3(Z - 2) - Ei -^-]

= 2 log ^- Cin 8^-1- - i Si 20+ J

L^gL^ .

m
This value comparable to W&.
is

It should be noted that King-Middleton's expressions 36 for the

expansion parameter as well as equations 38 and 40 depend on the form
of the first approximation to the antenna current. Hence, different
values will be obtained for transmitting and receiving antennas.
The various expansion parameters 31, 33, 34, and 38 are comparable
in magnitude when l/\ is not too large. But, as l/\ increases, the differ-
ences increase. If X = «>, then Wz = and W& = U — 2. As X
decreases, remains constant while Wz and W5
However, decrease.
Wz decreases more rapidly and eventually becomes smaller than W&.
It is difficult to decide a priori which particular value will give the best
asymptotic approximation for relatively small values of the expansion
parameter. For really large values of W
there is no problem since all
expansions are analytically the same and since for large we can take W
many terms of the series before it begins to diverge. Unfortunately, for
practical antennas ft is not very large.
The input admittance is a fluctuating function of the length of the
antenna, and point-by-point comparison of different approximations is
not very instructive. The following comparison is offered only to ex-
hibit the differences due to the differences in the expansion parameter
* It is to be noticed that Ei(jx) - C+ log jx - EinO'z) - C + log 1 + J;*
- Cinz -j'Six = Cix -jS\x + fcV.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
148 CYLINDRICAL ANTENNAS 5.4

if we limit the series 20 to two terms. We express all results for the input
admittance F; of a full-wave antenna (21 = X) in terms of the average
characteristic impedance K
a which occurs in the mode theory of antennas
when 21 < 1.5X, and which equals 60(£2 — 2). The mode theory gives

K a 2 Yi = 199.09 + ;125.41 + (29119 + jim2)KrK (41)

If we use equation 31 for the expansion parameter as in the King-Blake

paper, we obtain

K a*Yi = 199.09 -f j'125.41 - (42320 + flMM)Km~K (42)

If we use Gray's value (equation 34), we have

K a2 Yi = 199,09 +SWA1 + (49405 + /1M74)£.-* (43)

And, finally, if we use the King-Middleton value (equation 36), then,

K a
2
Yi = 199.09 + j'126.41 + (29352 + j29702)JC -
a
1
. (44)

The substantial difference between equation 42 on the one hand and

equations 43 and 44 on the other may be traced to the lower values of
the expansion parameter used in deriving the last two expressions. On
the whole the last expression the closest to that given by the mode
is

theory, although the reactive part of equation 43 is closer. Halleo

-2
points out that, if the terms depending on fi are included in equation
42, the conductance becomes equal to that given by the mode theory,
but the susceptance is still different. King and Middleton find that
for their expansion parameter the third term of the series 20 makes a
significant contribution. This indicates that the fourth term depending
-3
on ft should be computed before we could make an assertion concerning
the best value of the expansion parameter. In fact, the best value may
not exist, for it is possible that one value may give the best results when
only two terms of the series 20 are taken, and another value is best for
three terms. would be desirable to study more closely the most
It
obvious splitting of the integral, as shown in equation 15 and the corre-
sponding expansion parameter (equation 33). The expansion param-
eter plays the part of characteristic impedance in transmission line
theory, and it so happens that equation 33 gives very closely the asymp-
totic value of the input impedance of an infinitely long antenna. This
fact is very suggestive.
For further details concerning special asymptotic expansions and
their applications the reader is referred to the papers listed at the end
of this chapter.

Original from
UNIVERSITY OF MICHIGAN
 APPROXIMATE INTEGRAL EQUATIONS 149

5.5 Approximate integral equations

Usually assumed that equations expressing physical problems
it is

correctly have solutions. The assumption is safe provided no mathe-

matical approximations are subsequently made in the equations. An-
tenna theory provides examples in which such approximations lead to
equations with no solutions. Thus, when the radius of a cylindrical
antenna is small, the kernel in equation 7 for the Green's function is
usually approximated by a simpler function,

- -l
r({- z) r [«-z) 2
e-"* ro= 2
-f-a P*. (45)

If |£ — z\ kernel differs from the exact kernel (equation 2)

a, this

by a quantity of the order of the square of the radius. The difference

is largeonly in a small interval around £ = z. Furthermore, if /(£) is a
slowly varying function, the values of the integral in equation 7 for the
two kernels differ by quantities of the order of the square of the radius.
Presumably it was such considerations as these that led to an assump-
tion that equation 45 represents a reasonable approximation to G (£ — z).
However, the approximate integral equation with this value of G
has no solutions. For the approximate kernel the left side of equation 7
is a proper integral and is a continuous and differentiable function. On
the other hand, the derivative of the right-hand side is discontinuous at
z = 0. The nonexistence of solutions could have been anticipated
from physical considerations. After substituting T(£ — z) for (?(£ — z),

the derivative of the integral in equation 7 represents, except for a con-

stant factor, the exact potential on the axis of the cylinder due to a
charge on its surface; the same derivative also represents the exact
potential on the surface of the cylinder due to a charge on its axis.
Consequently, equation 7 requires a charge distribution which would
produce a potential discontinuity, and an infinite electric intensity, at a
place where there no charge whatsoever. Thus, the approximate
is

integral equation formulates an impossible physical problem.

As a approaches zero, the iterative process in Sections 5.3 and 5.4
leads to the same formal solutions of equation 7 for bothkernels, G(£ — z)
and T(£ — z). Since equation 7 has no solutions for T(f — z), these
expansions must be divergent.
At this place it may occur to the reader that some antenna problems
can be solved synthetically by analogy with electrostatics. In elec-
trostatics we may assume, for example, a positive charge uniformly
distributed on a straight segment between z =
and z = I, and an equal
negative charge uniformly distributed between z = and z = — I.
We can then calculate the potential of this charge distribution and re-
place one or more equipotential surfaces by conducting surfaces without

Original from
UNIVERSITY OF MICHIGAN
150 CYLINDRICAL ANTENNAS

affecting the field. Ia this way we can solve many problems of charge
distribution on conductors whose potentials are given (the shapes of
these conductors will be determined by the assumed charge distribu-
tions). It may seem, at first, that we can similarly assume arbitrarily
some current distribution and then calculate surfaces on which the
tangential components of the electric intensity vanish, thus solving some
antenna problems exactly. However, we shall find that for an arbitrary
current distribution there are no such surfaces.

REFERENCES
Hie following are basic papers expounding Hallen's method of antenna analysis:
1. E. Hallen, Uber die elektrischen Schwingungen in drahtformigen Leitern,
Uppsala Universitels Arsskrift, 1930, no. 1.

2. E. Hallen, Theoretical investigations into the transmitting and receiving quali-

antennae, Nova Acta, Uppsala, 11, 1938, no. 4.
ties of
3. E. Hallen, Further investigations into the receiving qualities of antennae: the
absorbing of transient, unperiodic radiation, Uppsala Universitets Arsskrift,
1939, no. 4.
The following papers discuss HalleVs theory:
4. C. J. Bouwkamp, HallcVstheory for a straight perfectly conducting wire, used
as a transmitting or receiving aerial, Physica, 9, July 1942, pp. 609-631.
5. L. Brillouin, The antenna problem, Quart. Appl. Math., 1, October 1943, pp.
201-214.
6. S. A. SchelkunofF, Antenna theory and experiment, Jour. Appl. Phys., 15,
January 1944, pp. 54-^0.
7. M. C. Gray, A modification of HalleVs solution of the antenna problem, Jour.
Appl. Phys., 15, January 1944, pp. 61-65.
8. S. A. SchelkunofF, On the antenna problem, Quart. Appl Math., 1, January 1944,
pp. 354-355.
9. S. A. Schelkunoff, Concerning Hall6n's integral equation for cylindrical antennas,
IRE Proc, December 1945, pp. 872-878.
10. R. King and D. Middleton, The cylindrical antenna; current and impedance,
Quart. Appl. Math., 3, January 1946, pp. 302-335; Corrections, ibid., 4, July
1946, pp. 199-200; Addit. corrections, ibid., 6, July 1948, p. 192.

11. E. Halle'n, Iterated sine and cosine integrals, Kungl. Tekniska Hogskolan, 1947,
no. 12.
12. K. Hallen, Properties of long antennas, Cruft Laboratory Report no. 44, Harvard
University, May 25, 1948.
13. E. HalI6n, Admittance diagrams for antennas and the relation between antenna
theories, Cruft Laboratory Report no. 40, Harvard University, June 1, 1948.
14. E. Halten, Traveling waves and uiwymmetrically fed antennos, Cruft Laboratory
Report no. 49, Harvard University, June 22, 1948.
15. E. Hallen, Properties of a long antenna, Jour. Appl. Phys., December 1948,
pp. 1140-1147.
The following papers are concerned with special applications of HalleVs theory
16. R. King and F. G. Blake, Jr., The self-impedance of a symmetrical antenna,
IRE Proc., 30, July 1942, pp. 335-349.

f~" Original from

_,. ..
Digged by V^OOglL
.-».-*!/-»
UNIVERSITY OF MICHIGAN
 REFERENCES 151

17. R. King and C. W. Harrison, Jr., The distribution of current along a symmetrical
center-driven antenna, IRE Proc., 31, October 1943, pp. 548-566; Correc-
tions, ibid., December 1943, p. 697.
18. R. King, Coupled antennas and transmission lines, IRE Proc., 31, November
1943, pp. 626-640.
19. C. W. Harrison, Jr., and R. King, The radiation field of a symmetrical center-
driven antenna of finite cross section, IRE Proc, 31, December 1943, pp.
693-697.
20. R. King and C. W. Harrison, Jr., The receiving antenna, IRE Proc, 32, January
1944, pp. 18-35.
21. C. W. Harrison, Jr., and R. King, The receiving antenna in a plane-polarized
field of arbitrary orientation, IRE January 1944, pp. 35-49.
Proc., 32,
22. R. King and C. W. Harrison, Jr., The impedance of long, short and capacitively
loaded antennas with a critical discussion of the antenna problem, Jour. Appl.
Phys., 15, February 1944, pp. 170-185.
23. R. King and C. W. Harrison, Jr., Mutual and self-impedance for coupled an-
tennas, Jour. Appl. Phys., 15, June 1944, pp. 481-495.
24. D. D. King and R. King, Terminal functions for antennas, Jour. Appl. Phys., 15,
February 1944, pp. 186-192.
25. C. W. Harrison, Jr., Mutual and self-impedance for collinear antennas, IRE
Proc, 33, June 1945, pp. 398-408.
26. C. W. Harrison, Jr., On the distribution of current along asymmetrical antennas,
Jour. Appl. Phys., 16, July 1945, pp. 402-408.
27. R. King and D. D. King, Microwave impedance measurements with application
to antennas, Jour. Appl. Phys., 16, August 1945, pp. 445-453.
28. D. Middleton and R. King, The thin cylindrical antenna: comparison of A
theories, Jour. Appl. Phys., 17, April 1946, pp. 273-284.
29. C. W. Harrison, Jr., A theory for three-element broadside arrays, IRE Proc, 34,
April 1946, pp. 204P-209P.
30. R. King and T. W. Winternitz, The cylindrical antenna with gap, Quart. Appl.
Malh., 5, January 1948, pp. 403-416.
31. C. T. Tai, Coupled antennas, IRE Proc, 36, April 1948, pp. 487-500.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 6

NATURAL OSCILLATIONS

6.1 Forced and natural oscillations

Once the steady-state solution of a given antenna problem has been
found, it is relatively easy to obtain the more general solution represent-
ing the response of the antenna to an impressed field varying arbitrarily
with time. Throughout the preceding chapters the exponential time
factor is in the form exp(jtat) which makes the equations more easily in-
terpretable when the radian frequency w = 2tt/ is real. However, there
isnothing in the analysis that requires u to be real and, if « is permitted
;

to be complex, we obtain the more general exponential solutions and can

use Laplace integrals to introduce an arbitrary dependence on time.
While analytically w can be complex, physically it is essentially real. For
this reason we first transform our equations formally by replacing ju
with p = £ jo>: that is, w with o> — j£.
+ The real part £ of the oscilla-
tion constant p is called the growth constant its negative is the damping
;

constant. Let F(x, y, z) be some field quantity when the function repre-
senting the impressed field is of the exponential form V exp(pf), and let

F(z, y, z) = V T(x, y, z, p)e«. (1)

Functions that are arbitrary for practical purposes may be represented

by Laplace integrals of the form

where the contour of integration (C) runs to the right of and parallel to
the imaginary axis in the p plane. For the impressed field represented
by this time function, we have

F(x, y, z, = S(p) T(x, y, z, p)e" dp. (3)

-^jf(c)
152

Original from
Digitized by kjOOQlC UNIVERSITY OF MICHIGAN
6.1 FORCED AND NATURAL OSCILLATIONS 153

As an example let us consider a perfectly conducting sphere of radius

a, concentric with the origin of our coordinate system, and let us assume
that, at the instant t = 0, a constant electric field of intensity Eo is im-
pressed on this sphere. If this field is parallel to the z axis, the meridian
component is E%* = — Eo sin 0. The total intensity tangential to the
sphere must vanish ;
hence, the component due to the charges and cur-
rents on the sphere is Eo(a t 9, t) = Eo sin when t > 0. When t < 0,
E 9 (a, 6, t) - 0. Therefore,

E $ (a, 0, f) = sin
§Jm
p- V dp. (4)

From equations 1-110 we can

obtain the transfer ratio E$(r, d)/E9 (a 6) f

for the exponential time variation, and, then,

r (paW + (pM Wr) + W3 f via - r + a) 1

J" W
exp
J(C p[(pa/c) 2 + (pa/c) + 1] L c

If c£ — r + a < 0, that is, if r —

be closeda > ct, the contour (C) may
with an infinite semicircle in the right half of the plane without changing
the value of the integral. Since the integrand is holomorphic inside the
closed contour, we have

E9 (r,B,t) = 0, r - a > ct; (6)

that is, the field is zero at points that cannot be reached in time t by a
disturbance traveling from the surface of the sphere with the speed c.
If r — a < ct, the contour (C) may be closed in the left half-plane. The
integrand is holomorphic except at p = and two other poles given by

(?)' + * + 1= 0,

Pi = -^. P2 = ^-. w = -i + *>V3. (7)

By the theorem of residues,

E?Qa T
fa
2 . vfl + w(a/r) + (a/r) 2
r"
L'
2 w(w — w*)

W*(w* — w) J

Digitized by V^OOgK UNIVERSITY OF MICHIGAN

154 NATURAL OSCILLATIONS 6.2

Substituting from equations 7 and simplifying, we find

( cr-.-.o«. C os [
1
^^ ~ cl)
- #] sin 6,
J
(9)
H
cos^ = iV3(l + -l)(l + -^ + -^)- ,

The transient vanishes at the surface of the sphere at all

part of 2?»

times. It represents free or natural electric oscillations of a perfectly

conducting sphere, and, except for the amplitude and phase, it can be
determined directly from equations 1-110 if we impose the boundary
condition Et(a, 6) = 0. From equations 7 we find that the damping
constant is c/2a, and the radian frequency w = cV3/2a. The wavelength
corresponding to this frequency is X = c/f = 4ira/\/3. It exceeds the
circumference of the sphere by 15.5 percent.

6.2 Equivalent networks

Any well-defined electric structure has a unique input impedance but;

the converse not true. Several networks, either discrete or continu-

is
ous, may have the same impedance Z(p) at all frequencies; and, of
course, many more may have approximately the same impedance in a
limited frequency range. The problem of network analysis is: Given a
network of resistors, inductors, and capacitors or a continuous physical
structure (such asan antenna), find its impedance (or admittance) function.
The problem of network synthesis is the inverse problem: Given a
physically realizable impedance function Z(p), find one or more networks
possessing this impedance. The problem of network synthesis is partic-
ularly important in the design of filters and impedance corrective net-
works. An equivalent network for an antenna may be useful in
experiments designed to test the terminal networks intended to operate
with the antenna, before the antenna is actually constructed.
In the case of discrete networks the problem of analysis is almost
trivial: We
have to solve a system of linear algebraic equations. But
the problem of synthesis is far from trivial. In the case of antennas even
the analysis is formidable. Fortunately, in practical applications simple

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
6.2 EQUIVALENT NETWORKS 155

approximations often suffice, and at present at least the complete solu-

tion of the problem of obtaining networks equivalent to antennas is in-
teresting only from the theoretical point of view.
The equations for discrete networks can be solved in general terms,
and it is relatively easy to establish the criteria for the physical realiza-
bility of Z(p) : that is, the conditions that must be satisfied by Z(p) if it

is to represent the impedance of any network whose elements consist of

positive resistances, inductances and capacitances.* Some of these
conditions are obviously applicable to either discrete or continuous
structures. For example, the zeros and poles must not lie in the right
half of the complex plane otherwise, the system would permit oscilla-
;

tions of increasing amplitude without the corresponding supply of energy.

If complex, the zeros and poles must occur in conjugate pairs. This
follows from the fact that the coefficients in Z[p) are real. On the
other hand, at infinity the impedance of a finite discrete network has
either a simple pole or a simple zero, or else it reduces to a constant. For
continuous structures the behavior of Z(p) at infinity cannot be specified
as simply. In this book, however, we are not concerned with the general
theory of impedance functions. Our functions are physically realizable
because they are obtained by the analysis of given structures and their
analytic properties are exhibited.
The admittance Y t of a spherical antenna is given by equation
2-196. It is seen to be a sum of an infinite number admittance func-
of
tions. Each component admittance is proportional to the radial ad-
mittance Y n (pa) of a zonal spherical wave (equation 2-198). Each
radial admittance is a rational fraction, in which the degree of the numer-
ator equals that of the denominator. The zeros and poles are in the left
half of the plane, excepting the simple zero at the origin. Hence, each
component admittance is physically realizable, and the admittance of
the spherical antenna may be considered as the admittance of certain
finite discrete networks connected in parallel.
To derive the network structures of the radial admittances it is
convenient to consider simultaneously the radial impedance of a TM
wave and the radial admittance of a TE wave of the same order (see
Sections 1.5 and 1.6),

Zn {a)
~ Yn {a) ~ 9

Kn n (w) '
Kn n (w) (10)

w = pa(^e)^.

*
Otto Brune, Synthesis of a finite two-terminal network whose driving-point
impedance is a prescribed function of frequency, Jour. Math, and Phys., 10, August
1931, pp. 191-236.

Original from
Digitized by VoOO^lC UNIVERSITY OF MICHIGAN
156 NATURAL OSCILLATIONS 6.2

The impedance of the TM wave is proportional to the admittance of

the TEwave, and the corresponding networks are dual. As w ap-
proaches zero, Kn n (w) approaches Aw~ n exp(-w). Hence,

and the impedance has a simple pole at the origin. Since n/pea is the
impedance corresponding to the capacitance ea/n, this part of Z is
physically realizable. Let us subtract it from Z and simplify the
remainder, using the first of the recurrence formulas for the normalized
Bessel functions,*

wKn n '(w) = —nKn n (w) — w Kn % -i(w),

wKn n -\{y)) = n Kn n -i(v>) — wKn % {w).

Thus, we have
_ v Kn n '(w) n_ _ wKn n -\{vj) >

Kn n (w) pea pea Kn n (w)

Using the second recurrence formula in the denominator, and dividing

both terms of the fraction by the numerator, we obtain the first of the
following recurrence relations:

7 tm _ _a_ j
r
i
'

pea (n/pMa) + K„_,™

(14)

pva (n/pea) + Z n -i TM
The second formula is obtained by interchanging p and e.
From equations 1-110 we obtain the radial impedance of the zonal
wave of the first order,

^ »(» + 1) pea
+ (l/pMa) + v ~* K ?

Interchanging /i and e,

Yt
r* = —+ppa
... \
(l/pea) +,
i|
• (16)

Equation 15 shows that Zi is the impedance of a capacitor in series with

the parallel combination of an inductor and a resistor (Fig. 6.1a)
Similarly Y\ is the admittance of an inductor in parallel with a series
combination of a capacitor and a resistor (Fig. 6.16).

• Applied Mathematics, p. 412.

Digitized by V^OOglC UNIVERSITY OF MICHIGAN

6.2 EQUIVALENT NETWORKS 157

From equations 14 we see that Z„™ is the impedance of the capac-

itor ea/n in series with the parallel combination of the inductor na/n
and r„_i TE and that F n TE is the admittance of the inductor pa/n in
,

parallel with the series combination of the capacitor ea/n and Zn-i™.
Using these rules repeatedly, we obtain the network representations of
the radial impedances of TM
and TE waves (Fig. 6.2).

(a) (b)

Fig. 6.1 Network representations of the radial wave impedances: (a) TM zonal
wave of the first order, (b) TE zonal wave of the first order.

ea ea ea
n 2n-3 2n-7

(a)

ea ea
2n-l 20-5

(b)

Fia. 6.2 Network representation of the radial wave impedance of the nth-order
zonal waves [and more generally of waves in which the radial field intensity is propor-
m
tional to Pn (cos $) cos mp]: (a) TM
waves, (6) TE waves.

If we
multiply each inductance and resistance and divide each
capacitance by a constant u, the impedance of the network is multiplied
by w.For instance, the impedance of the network in Fig. 6.3 is u times
that of the network in Fig. 6.1a. Hence, the admittance (equation 2-
196) of the spherical antenna may be represented by a parallel combina-
tion of networks of the type shown in Fig. 6.2a.
With the aid of such network representations Chu was able to

.. f~* I Original from

...
Digitized by V^OOglL UNIVERSITY OF MICHIGAN
158 NATURAL OSCILLATIONS
obtain important relationships between the gain, bandwidth, and size
of antennas.*
If the admittance has only simple poles in the finite part of the
plane, it can be expressed (the Mittag-Leffler theorem) in the following
form,

n \V P* Pn Pn • pn /
(17)

where F(p) is an integral function, p = p n is a typical pole of Y(p),

~
and a is such that T,A n p n a is convergent.
Equation 2-203 for the admittance of the
spherical antenna is of this form.
In the case
1 of finite networks we can retain in the sum-
mation only the partial fractions A n /(p — pn)
*

and include the remaining terms (needed only

Fig. 6.3 A network whose for convergence purposes) in F(p) after this ;

impedance is u times that

of the network in Figure
^
rearrangement the degree of F(p) cannot ex-
one for Jf there ig ft po , e ftt infmitV( it
must be simple. If the network is continuous
and nondissipative, it is possible to obtain directly from Maxwell's
equations the following expansion, f

W-EoTT^ < 18 >

where o>„ is the natural frequency of the nth mode when the input
terminals of the network are short-circuited and £ „ is the energy stored in
it when the current through the input terminals is unity. Similarly,
for the input impedance we have

= S '
(19)

where o>„ is the natural frequency of the nth mode when the terminals
of the network are open and & n is the stored energy when the voltage
across the input terminals is unity. These expansions are special cases
of equation 17.
If the network is slightly dissipative, the following approximate

*L. J. Chu, Physical limitations of omni-directional antennas, Jour. Appl.

Phys., 19, December 1948, pp. 1163-1175.
f E. U. Condon, Forced oscillations in cavity resonators, Jour. Appl. Pkys., 12,
February 1941, pp. 129-132.

Denized by V^OOglL UNIVERSITY OF MICHIGAN

6.2 EQUIVALENT NETWORKS 159

expansions may be obtained,*

*

Y °a) ~
? 2S.(«. -« a 8
+i«*.Q.-
1 )

3
Z<j») Z 7nn—2 2 j. -
n -n '

where the Q„'s are the " quality factors " of the corresponding modes,

and £n = re(p n ). If Pn is the average dissipated or radiated power in

the nth mode, the relative rate of energy loss is and the relative
rate of decrease in the amplitude of the oscillations is P„/2£ n Hence, .

= «' = ^r- (22)

Each term in equation 18 represents the admittance of a series

combination of an inductance and capacitance; and the entire series is
the admittance of an infinite number of such simple circuits connected
in parallel. Similarly, the series 19 is the impedance of an infinite
number of parallel combinations of inductances and capacitances con-
nected in series.
The approximate expansions 20 represent similar networks with
small resistances added in series with series combinations of inductances
and capacitances, and small conductances in parallel with parallel
combinations. But it is not safe to use these approximate formulas
except in the immediate vicinity of a resonance.
These equations are
based on the assumption that the damping constant of each natural
mode is and they do not take into consideration the precise
small,
mechanism by which the damping is introduced. Consider, for in-
stance, the circuits in Fig. 6.4 in which R is very small and R' very large
compared with (L/C)^. Both circuits have small damping constants,
and their quality factors are equal if RR' = L/C. Their impedance
representations in the form 20 are identical, and yet the circuits behave
very differently for very low and for very high frequencies. If R' is
constant, the equivalent series resistance of the circuit b is proportional
to the square of the frequency. The input resistance of a has a small
constant value at all but at high frequencies the input
frequencies;
resistance of 6 is very large. Near a resonance, however, one term in
the expansions 21 becomes dominant and controls the changes in the
impedance with frequency. For this reason expansions 21 are im-
*Ibid.

Digitized by V^OOgK UNIVERSITY OF MICHIGAN

160 NATURAL OSCILLATIONS 6.3

portant. The parameters aj B , & n Q» can

,
often be obtained approxi-
mately without solving the complete boundary value problem.

R'

C L R C
Hf—'W 1
VAr—o o
|f

(a) (b)

Fia. 6.4 Two typical series resonant circuits.

If Y(p) has only simple zeros ay, a% and simple poles b\ 62,- • • •,
t
•

in addition to a zero or pole at p = 0, then by the Weierstrass theorem,

Y(p) = e*<p>p« x

n(l-p6„- )exp[p6 n -»+ hP 2 b n -2 +ip3 b n ~*+- + (^1)-^-^-^

'
1 • -

(23)

where F(p) is an integral function, a = ±1, and is such that

and Y,b n ~* are convergent. This expression lacks the simplicity of the
-

corresponding product expansions for finite networks, in which the

exponential convergence factors are not required and F(p) = 0.

6.3 Small antennas

Consider a perfectly conducting spherical surface joined to conical
surfaces extending to the origin (Fig. 6.5) where the cones are con-

Fxa. 6.5 An inductively loaded sphere.

nected in series with two other cones forming a cavity. The condition
for natural oscillations is

Z+(j«) + Z~(jo>) = 0, or F+(jaO + Y~(ju>) = 0, (24)

Original from
Digitized by VoOO^lC UNIVERSITY OF MICHIGAN
6.3 SMALL ANTENNAS 161

where Y + is the admittance seen outward from the gap A, B and Y~

is the admittance seen inward. The former (for circularly symmetric
oscillations) is given by equation 2-196 and the latter, except for a small
capacitance edge effect, is

v-, -ir -x &2 sin 0a sin 0b — K\ cos 0a cos 0b

Y (jw)\ = jKi
' . .

(25)
K2 cos 0a sin 0b + id tin 0a cos 0b '

where K\ and K% are the characteristic impedances of the conical guides

seen inward from the gap.
As to —* 0, we have

F-(j«)-~- . U= (Kia 4- K 2 b)(i*)H,

(26)
Z '

where L< is the internal inductance of the spherical oscillator and C # ah.

is the external capacitance, representing the higher modes of external

C E .sr=4? f a

T C ^h o 2M&
vl*--—
, < ^
5.R raa= _
27?
= e0

L_J
Fia. 6.6 Network representation of the inductively loaded sphere.

I 37T

LJLi
Fig- 6.7 A circuit equivalent to a conducting sphere oscillating in its gravest mode.

oscillations on the sphere, shunt with the network representing the

in
2
first mode. The latter network is shown in Fig. 6.3 with u = —- The •

on-

complete network is shown in Fig. 6.6.

If we short-circuit the cones, Li = and the network becomes that
shown in Fig. 6.7. This network still represents correctly the gravest
mode of oscillations on the sphere. Incidentally, with the values of the
circuit parameters as indicated, the energy stored in the capacitor equals

Original from
_.. ..
D.g.t.zed by ^OOglL UNIVERSITY OF MICHIGAN
162 NATURAL OSCILLATIONS 6.3

the external electrostatic energy associated with the charge on the sphere, and
the energy stored in the inductor equals the external magnetic energy associ-
ated with the current.
The equivalent circuit of a small antenna can be obtained from
general considerations without solving the particular antenna boundary
value problem. Consider, for example, a wire of length 21 loaded in-
ductively in the center. From the law of conservation of energy, we
have

where q is
4
the charge on one
HIM *]--«arm of the antenna. The first term inside
the brackets represents the stored magnetic energy and the second the
stored electric energy; P is the rate at which
the circuit loses energy by dissipation in heat or
by radiation. Power lost by dissipation in heat
rao is R(dq/dt)
2 but, since we are interested in ra-
;

diation, we shall assume that the conductors

are perfect.* In Chapter 1 we have seen that
Fig. 6.8 Network repre- a small circuit radiates a small fraction of its
sentation of a small per-
gtored energy, and its electrical behavior in a
X UCt mg
!tenna Ti •
T relatively short time interval
j^tt
must be deter-
Hence, Ai
with the terminals
t
, r
short^ircuited. mmed i »
primarily by and C. L tenta-
tively we neglect P 27 and obtain
in equation
the natural frequency w = (LC)~
H . Having determined this fre-
quency, we calculate the radiated power from the external field of the
circuit. For the wire of length 21, for instance,

The coefficient of proportionality,

is called the "radiation resistance" because it plays the part of resistance

in the circuit. Thus we have the equivalent circuit shown in Fig. 6.8.
The next problem is to construct a circuit whose resistance is pro-
portional to the square of the frequency when the frequency is sufficiently
low (in the present case pi 1, which isthe definition of a "short" wire).
* It must be remembered, however, that ohmic losses in small antennas are
often much larger than radiation losses, and they should always be included in the
complete circuit.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 6.4 THIN CONDUCTORS — 163

Consider the impedance Z of a constant resistance ft' in parallel with an

inductance L so small that wL ft' Then, < '.

2= i"**
ft' + juL

.^(1+ ^)-'^ + ^ +0 [^r|. <30)

Hence, the circuits shown in Fig. 6.4 are equivalent when wL < ft', and
the series resistance of the circuit a is proportional to the square of the
frequency. Using ft„d for ft, we obtain

L =
ik'-v-' K£) (31)

This equation suggests that space may be considered as a pure resistance

inductively coupled to our antenna. Equation 31 does not give this
resistance and the coupling inductance uniquely. This is natural since
space has no well-defined terminals, and it has to be provided with them
if we wish to include it in our networks. A short wire and a small spher-
ical antenna excite the same type of
c
external field (the zonal wave of the
first order) except in their immedi-
ate vicinities. For the spherical /u
antenna (Fig. 6.6) the shunt re- 37r

sistance equals 2t)/3tt. Hence,

choosing ft' = 2^/3*, we find that - . . „, .

* 1G 6.9 Network representation of a

i- •
j »/« -

the coupling inductance is tf/fcr,

shorl wire i oadcd inductively in the
and the network in Fig. 6.8 is center.
transformed into that in Fig. 6.9.
The condition wyl/Z-n <S 2jj/37t, implicit in the transformation, is satisfied
since it is equivalent to pi *C 2 which
weaker than the condition is

^I<1 required for the original network in Fig. 6.8. However, the
practical range of application of these networks can be stretched.

6.4 Thin conductors — I

Natural electric oscillations on thin wires resemble vibrations of strings

under tension, since these phenomena are basically the samo. When a
string with two fixed ends (Fig. 6.10a) is displaced from its neutral state,
the restoring forces of tension tend to return it to its neutral position;
but, when it does return there, it has acquired a momentum which car-
ries beyond the neutral position. The displaced string has potential
it

energy which is transformed into kinetic energy when the string returns

Digitized by V^OOgK UNIVERSITY OF MICHIGAN

164 NATURAL OSCILLATIONS 6.4

to the neutral position ; the kinetic energy is gradually transformed back

into potential energy as the momentum carries the string away from the
neutral position. Similarly, when some electrons are displaced on a
wire (Fig. 6.106), the restoring forces of attraction between the positive
and negative charges tend to pull them back but, when the wire is once
;

more electrically neutral, the electrons are in motion, and their electro-
magnetic momentum carries them on until the ends of the wires are again
oppositely charged. Here the energy associated with the sep-
electric
arated charges is gradually transformed into the magnetic energy associ-
ated with the moving charge, and vice versa.
In the case of strings under constant tension and with uniformly
distributed mass, the various natural modes of oscillation are given by

k a, *|

Fig. 6.10 Diagrams representing various modes of mechanical oscillation of

stretched strings and electromagnetic oscillation of thin conductors.

sinusoidal displacements (Fig. 6.10a, c, d). The natural wavelengths

are integral submultiples of a length which is double that of the string,

X„ - —n > n = 1, 2, 3, (32)

The natural frequencies are given by

= (33)

where the velocity of waves on the string. If either the mass or the
c is

tension or both are not distributed uniformly, the displacements cor-

responding to the natural modes are distorted sine waves and the natural
wavelengths deviate from equation 32.
We have a similar phenomenon in the case of oscillations of electric
charge on wires. In a cylindrical wire neither the inductance nor the
capacitance is uniformly distributed. At a point near the center of the
wire, the electric intensity due to the moving charge is larger than it is

Original from
Digitized by VoOO^lC UNIVERSITY OF MICHIGAN
6.4 THIN CONDUCTORS — 165

near the ends, for in the firstproduced by the current on

case the field is

both sides of the point in question, whereas in the second case it is pro-
duced by the current on one side only. Hence, the inductance of a
cylindrical wire is smaller near the ends than it is near the middle. The
capacitance, on the other hand, is larger near the ends. Inasmuch as
the inductance increases and the capacitance decreases with the decreas-
ing radius of the wire, we may obtain a uniform distribution by making
the wire thinner toward its ends. The optimum shape turns out to be
very nearly spheroidal. The loss of power by radiation has only a
second-order effect on the natural frequencies.
Abraham was the first to study quantitatively the electric oscilla-
tions on thin spheroids.* It is relatively easy to obtain the exact solu-
tions of Maxwell's equations in terms of spheroidal functions satisfying
certain ordinary linear differential equations; but the evaluation of
these functions is not easy, and it constitutes the real problem. At the
time Abraham did not seem to appreciate the importance of shape, for
he apparently believed that his formulas applied to thin rods in general
(by taking, of course, some mean value of the radius of the wire). Ac-
tually the orders of magnitude of the deviations from equation 32 are
different for the spheroid and for other shapes even though in all cases
the deviations are relatively small. As the radius of the wire decreases,
the natural wavelengths of spheroids approach rapidly those given by
equation 32; but for other shapes they approach these values slowly.
Abraham gives the following expressions for the oscillation con-
stants corresponding to the various modes of oscillation on a thin
spheroid of length 21 whose maximum radius is a,

2xj - 9.74,4 - 47.4A»

? lV>* =
.

4(1 + 5.6^)
» = 1.

- - 77.8A 2
n _
4*j 12.54
4(1 + 3.34 2 ) '

_ 2n*j- (9.66+ 4 log nM - 2(4.8 + 21ogn)»i4»

4(1 + n-'(4-8 + 2 log n)A
2
]
9 n* *» w .

where Abraham's parameter A is related to the average characteristic

impedance of a thin spheroid, defined in Section 1.15, as follows:

4 Iog(2i/fl) K + 83.18 K a" = 120 log

° a
- (35)
a

* Die electrischen Schwingungen um einen stabformigen Leiter, behandelt

nach der Maxwell'schen Theorie, Ann. d. Phys., 66, October 1898, pp. 435-472.

Digitized by V^OOglL UNIVERSITY OF MICHIGAN

 166 NATURAL OSCILLATIONS 6.4

From these expressions we may obtain the quality factors of thin

spheroids:

Q
V"
J*n_ = *(Ka + 83.18) '
2£ n 292 [1 + 146(tf + 83.18)-*]

*{K a + 83.18)
'
187 [1 + 187(A" a + 83.18)- 1
]

nx(/T« + 83.18)

n> 2. (36)

These quality factors for natural oscillations will differfrom the corre-
sponding quality factors in the case of forced oscillations in the terms of
order X/K m in the denominator.
A few years later Marcel Brillouin published* his analysis of natural
oscillations on spheroids of all eccentricities. His results for thin
spheroids differ very substantially f from those given by Abraham.
In 1930 Halten obtained a general formula for the natural oscillation
constants of thin conductors of arbitrary shape by a method described
in the preceding chapter. His result for thin spheroids! is the same
as Abraham's. In 1941 Schelkunoff obtained another general formula
as a byproduct of the mode theory of antennas§ described in Chapter 2.

Finally, in 1944 Page obtained the oscillation constants of spheroids for

the first three modes. Both Abraham and Page solve Maxwell's
||

equations in spheroidal coordinates; but in the later stages their

methods are different.
A
comparison of the logarithmic decrements, 5 = tt/Q, will enable
us to bring out several important points. Page's and Schelkunoff's
formulas give only the first order terms ; these terms will be compared
first. We have according to:
* Propagation de V Electricity, vol.
1, Hermann, Paris, 1904.

f L6on Antennae
Brillouin, for ultra-high frequencies, Elec. Comm., 21, 1944,
no. 4, pp. 257-281; and 22, 1945, no. 1, pp. 11-39.
X Tiber die elektrischen Schwingungon in drohtforcnigen Lei tern, Uppsala Uni~
versitets Arsakrift, 1930, no. 1, p. 18.

§ S. A. Schelkunoff, Theory of antennas of arbitrary size and shape, Proc. IRE,

29, September 1941, pp. 493-521.
Leigh Page, The electrical oscillations of a prolate spheroid,
|| III, Phys. Rev.,
65, February 1 and 15, 1944, pp. 98-117.

_. Original from
..
Digged by V^OOgK
f~"m-*t*\r*
UNIVERSITY OF MICHIGAN
6.4 THIN CONDUCTORS — 167

Abraham
2.43 1.56 1.17
=
81
log(//a) + 0.69 log(Z/a)+0.69 log(J/a) + 0.69 1

Marcel Brillouin

= 1-22 0.77 0.472

62 63 =
log(i/«) + 0.69 log(Z/a) + 0.69 log(J/a) + 0.69 :

HalUn
2.43 1.17
= .

«i
log(Z/a) + 0.69 log (i/a) + 0.69 ;

Page
2.43 1.54 1.24
Sl = ;77
_

logtf/o) - 0.52
>

log(i/a) - 1-15 logtf/a) +0.94 ;

Schelkunoff

2.43 1.17
«i =
log(//a) log (//a) - 0.2

The reason for the omission of the second mode in HalleVs and
Schelkunoff's tables is this: Abraham, Brillouin, and Page considered

n = 1 n = 2 n = 2

(a) (c)

Fig. 6.11 Modes of oscillation in thin conductors: (a) first mode, (b) second mode
in a continuous conductor, (c) second mode in a conductor broken in the center.

natural oscillations of an unbroken wire and their even modes corre-

spond Hallen and
to antisymmetric current distributions (Fig. 6.116) ;

Schelkunoff obtained their formulas as a byproduct of the theory of

centrally driven antennas, and their even modes correspond to the
symmetric current distributions that would exist in wires broken in the
center (Fig. 6.11c). Both Hallen's and Schelkunoff's methods may also
be applied to the antisymmetric even oscillations.
The numerators in the formulas for the decrements can be cal-
culated exactly and unambiguously, for they depend on the power
radiated by an infinitely thin wire. In such a wire the current distribu-
tion is exactly sinusoidal, irrespective of its shape, and the radiated
power can easily be calculated to any number of decimal places. The
expressions given by Page for the second and third modes contain
small errors those given by Brillouin, quite large errors.
;

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
168 NATURAL OSCILLATIONS
The denominators cannot be determined unambiguously because
allthese formulas give only the first terms of asymptotic expansions,
valid for " large values " of \og(l/a). There is no way to distinguish
between " large values " of \og(l/a) and " large values " of log(fc//a) =
log(J/ a ) + where k is some positive constant which may be
log k,

either greater than or less than unity. Thus, the differences between
the denominators are not significant without the simultaneous considera-
tion of the higher order terms. For example, Abraham's expression
for the decrement of the first mode including the second order term is

2.43 2.96
1
log(//a) + 0.69 [log(//a) + 0.69] 2
The " first order " term given by Page gives

_ 2.43 2.94
61
log(Z/a) + 0.69 + [log(J/a)+0.69] 2
*

Thus, Page's " first order approximation " is equivalent to Abraham's

" second order approximation." This is an excellent illustration of a
theoretically obvious fact that one first order approximation may be
just as good as another second order approximation. Schelkunoff's
first order term gives

- 2.43 1.68
dl
log(Va) + 0.69 + [logy/a) + 0.69] 2 "

This is better than Abraham's first order approximation but not as good
as his second order approximation. Finally, for the high order modes,
Abraham's second order approximation is

log n + 2.41 (log n + 2.4) 2

n[log(i/o) + 0.69] 2n[log(Z/a) + 0.69]

2 '

whereas Schelkunoff's first order approximation is

log n + 2.41
n[logff/a) - 0.25 log n}

= log n + 2.41 (logn + 2.41)qogn+ 2.76)

n[\og(l/a) + 0.69] 4nflog«/a) 0.69] 2 +
Hence, Schelkunoff's first order term includes somewhat more than one
half of Abraham's second order term.
Another important point concerns the asymptotic character of the
above equations. If we were to obtain the complete series for 5„, we
should find that it is a divergent series. For small values of log(//a),

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
6.5 THIN CONDUCTORS — II 1G9

the series is completely useless. For large values of log(//a), its terms
decrease at first, but eventually they begin to increase indefinitely.
As \og(l/a) becomes larger, the number of terms in the first group (the
group of decreasing terms) also becomes larger. For any given value
of log(J/a) there is a definite limit to the accuracy with which we can
calculate S n from the scries; and this accuracy deteriorates as log(Z/a)
decreases until finally the series breaks down altogether. Although
all expansions, regardless of themethod by which they are obtained,
are formally identical, in the sense that any one may be transformed
into any other by a simple change of the expansion parameter, some
expansions begin to deteriorate for smaller values of log {Ifa) than
other expansions; that is, not all expansions are equally useful when
\og{l/a) is relatively small — as it must be for even the thinnest practi-
cable wires.
For higher order modes the series for 6 begins to deteriorate for
larger values of log(i/a). This may be seen from Abraham's expres-
sion in which term " exceeds the first order term for
the " second order
sufficiently large values of n as long as l/a has a fixed value. Similarly,
in Schelkunoff's expression the denominator becomes negative for
sufficiently large n. As n increases, the expressions remain equally
valid only if log (l/a) is increased correspondingly.
The methods used by Abraham and by Page depend on the solution
of Maxwell's equations in spheroidal coordinates and are applicable
only to spheroids and near spheroids but the spheroids may be of all
;

eccentricities. The methods used by Hall&i and Schelkunoff are more

general as far as shapes of conductors are concerned but for computa- ;

tional reasons they are restricted at present to thin conductors. For

fat conductors (excluding cones)
Hallen's method offers more hope. ^-'i

6.5 Thin conductors — II „-'' ^

Fairly simple and general formu- ^^^T" ^(s^ias^^Zl

*
las for resonant wavelengths and ^^Z^^
% *

quality factors of thin conductors \\^^

may be obtained as follows. We ^
shall assume that the conductor fio. 6.12 A thin wire,
may be bent (Fig. 6.12) and that
its radius may vary arbitrarily with position s on the conductor. Let
the current and the charge per unit length be

f(s, = Z(*)e-" cos at,

?(s, = q(s)e~ kt sin(w* + tf),

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
170 NATURAL OSCILLATIONS 6.5

where w is a natural frequency, fc the corresponding damping constant,

and t? is a phase angle approaching zero with A;. Since

(38)

wc have
I'($)e- kl cos wt = - ff (s)e-*<[« cos(«* + &) - k sin(«t + *)]. (39)

Since the bracketed expression equals

(w 2 + k2 ) H cos (at + & + tan" £j 1

p

we have

* = -tan" 1 - , /'(*) = -(«* + * 2 )* «<•). (40)

We assume now that the conductor

shall is so thin that k2 is negligible
in comparison with a 2 then, ;

/'(*) = -OJ^W. ( 41 )

Let us now consider two instants, separated approximately by

one quarter of a period, such that at one instant the conductor is elec-
trically neutral and at the other no current is flowing in it. Associated
with the current at the first instant there is a certain quantity & m of
stored magnetic energy,

&m = d*2i
^ff r~' /(Sl) /(S2) 008
* (42)

where
r= l(si-s2 )2 + [p( Sl )m (43)

and where p is the radius of the conductor.* Similarly, at the second

instant we have stored electric energy

~l
6. = r v( Sl ) ?( S2 ) dsi ds2 (44)
~j^ff -

Substituting from equation 41, we have

~l
5 - = 855^" r r<*° '
7 (S2) dSl dS2
ff -
(45)

* On
account of the surface distribution of current the integral in equation 42
is a quadruple integral to begin with; but, as long as / (s) is a slowly varying function

of s, we may reduce it to the double integral. The assumption of a " slow variation
of I {a) " restricts us to the natural modes whose wavelengths are large compared
with the radius.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 6.5 THIN CONDUCTORS — II 171

Since k is assumed to be small, the magnetic energy stored at the first

instant is substantially equal to the electric energy stored at the second
instant; therefore,

*-(»Y- ff^wi**. .
(46)

//r_1 /(si) /(*2 > cos * ^ dS2

When the radius is small, the integrands are large in the vicinity of
62 = where r
Si is Hence, we shall approximate equation 46 by
small.
assuming that, in the integration with respect to S2, the current I (82)
equals I(si),

„„.(*)• j^wg* r
~ l
C0S ds2 [/(sl)12
.
m
J\J * )
For a straight conductor cos >p = 1.

The first parenthetical factor in the numerator equals the product

of 4tt6 and the potential at s = si due to a uniformly distributed charge
on the conductor ;
hence,

- - -sSr '
(48)

where C(si) is substantially the capacitance per unit length. Similarly,

J r-i cos * ds2 = ~ L( Sl ), (49)

where L(si) is the inductance per unit length. Hence, equation 47

becomes

"V =
/arY =
1-x- I
—/e[C(.)]- [Z'<,)f* ( 5 °)
V / J^L(s)[I(s)\>ds

In the case of a straight conductor L(s) C(s) = Me, and

o>V=KH =~ (51)
fL(s) [I(s)]*ds

Let us now consider more specifically a straight conductor and

assume that s is the distance from one of its ends. The principal parts

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
172 NATURAL OSCILLATIONS 6.5

of the current and its derivative are

nrs dl rix - nvs

I(s)
. T .
= /osin-^-. ^ = Jocos-g- >
/rrt .
(52)
~zi

where 2/ is the total length of the conductor. Substituting in equation

51, we obtain

£' L(s) cosHnrs/21) ds

'
So'
L (g ) 8in 2 (nxV2/) ds

£ L(s) ds L(s) cos(rtT S/l) ds

(53)

f*'
o
L(s) ds -£
jo
L(s) cos(nxs/0 ds

We also have
L(s) « log 4s(2J - s)[p(«)l-
2
. (54)

For a spheroid
[p( S )]2 = ah(2l - s)l~2 ,
(55)

where a is the maximum radius and 21 is the total length. Hence, L(s)
is constant, and

"-Y.
y
x. 4- ; (56)
2/ ' «
that is, there is no first order deviation of the resonant lengths from
(4Z/n). For a cylinder, p(s) = a. Evaluating the integrals, we find

J« |\ Si 2nx 1
A
x
" n L
+ ,

2nx [Iog(2l/a) - 1 + log 2]

'

J
(57)
Z_ n Si 2nw
A, "4 8x [log(2Z/o) - 1 + log 2]

These expressions agree with those obtained by other methods except

for an additive constant in conjunction with log(2Z/<0> which as we —
have seen —
cannot be fixed uniquely.
The quality factor Q„ associated with the nth natural mode may be
obtained from equation 22. To calculate its principal part, we evaluate
the stored energy (equation 42) on the assumption that the current dis-
tribution is sinusoidal. We also calculate the corresponding radiated
power (by the Poynting vector method or some other method) and ex-
press it in the form
Pn = WJo 2
,
(58)

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
6.5 THIN CONDUCTORS — n 173

where 7o is the maximum amplitude of the antenna current and R n is the

radiation resistance referred to a current antinode. Therefore,

n 60* r 21 r 21 . nirsi . nw82 cos >f> , , ,_ n .

where 21 is the length of the conductor and 8 is measured from one of its
ends (or from a current node in the case of a loop).
The integral may be approximated in the same way as in passing
from equation 46 to equation 47,
„ 60r r 21 ( r 21 cos ip - \ 2
nirsx ,
fan.

Replacing the integral in parentheses by its average value, we have

60x
Qn *

Noting that at low frequencies the integral,

>2l n2l

represents the inductance of the current filament, we have

Q. = (63)
fj£-
In the derivation of this formula for Q„, we have pointed out the
various mathematical approximations. If we rely on physical reason-,
ing, we can obtain equation 63 much more simply. Let us return to
equation 22 which defines Q n and to equation 58 which gives the
,

radiated power. If L is the inductance of the conductor, the stored

magnetic energy for a uniform current equal to Io will be \LIq 2 ; for a
sinusoidally distributed current the stored energy is one half of this
value,
& n = \Lh2 . (64)

Substituting from equations 58 and 64 in equation 22, we obtain equa-

tion 63.
If we apply either equation 61 or 63 to a spheroid, we obtain
exactly the first term of Abraham's expression. If we use equation 59,
we find
Q n = 30ht ^log - log n + Ci nxj B.-» (65)

Original from
Digitized by VsOQslc UNIVERSITY OF MICHIGAN
174 NATURAL OSCILLATIONS 6.5

Substituting I = nX„/4, we find

Qn = 30nir (log ^+Q (66)

The limit of R n as the radius approaches zero is independent of the

shape of the antenna and can be calculated exactly,

Rn = 30 Cin 2nx = 30(log n + log 2x + C - Ci 2nx),
(67)
n = l, 2, 3, 4,---.

These values of the radiation resistance correspond to various modes of

oscillation on a continuous wire (Fig. 6.11a and 6). If the wire is broken

in the center, then the limit is

Rn =
'
120 Cin mr - 30 Cin 2nx, n = 2, 4, 6,-
-

(68)
= 90(log n+ log x + C - I log 2) + 30(Ci 2nx - 4Ci nr),
where n is the number In
of half-waves in the oscillation (Fig. 6.11c).
the preceding equations we have assumed that the intrinsic impedance
of free space is 120x ohms (based on the velocity of 3 X 108 meters per
second). In the mks system of units the permeability of free space is
exactly 4xl0"7 ; to find the intrinsic impedance, this value must be
multiplied by the velocity of light. For the first five odd modes we
have

n 1 3 5 7 9

Rn 73129602 105.494231 120.766134 130.845547 138.378883

For the first five even modes,

n 2 4 6 8 10

Rn 93.430697 114.088669 126.226587 134.847874 141.537922

For a wire broken in the center, 9

n 2 4 6 8 10

Rn' 199.087710 259.634117 295.750339 321.506803 341.523773

In comparing the logarithmic decrements at the end of the pre-

ceding section, we pointed out that the numerators may be calculated
exactly. These numerators are equal in fact to R n /30n, where R n is
given in the above tables.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 PROBLEMS
1. Obtain the most general solution of Maxwell's equations which is independ-
ent of the z coordinate.
Arts. This solution is the sum
two solutions one of which is derivable from
of
Hz and the other from E„ Each
component solutions is given by series (or
of the
integrals) involving an arbitrary parameter n and arbitrary seta of constants. If

Rntfp) = A n Jn (fip) + Bn Kn (0p), <M*0 = Cn cos n<p + Dn sin tup,

then one component solution is of the form,

H, = £ ft.tfp) flww, E„ ?L
7^r= Rn'ifo) fbM

Ju£p n

The other is of the form,

E, - L #„(&>) *W(#). ff, = T1" - -jV^E *n(*>),

Hp =-i-ZR*(0P)*n'(<?).
wpp n

2. Calculate the field such that the components of E tangential to a cylinder of

radius a assume given values Et (a, .- \ E v (a, *>). In addition the field is required to
-
be finite inside the cylinder and to vary as p ^ exp(—j0p) at infinity.
Ana.

E,-(p, *) = E
n-0
cos n*> + BI(n < 2)
sin n*>)
Jn(0a)
. p < a,

E,+ (p, <p) = 2- (£*.» tu cos n* -f Ea n ,

(2i
sin n?) . » p > a,
n-o -J AT„(0a)

«T(* *) = £ (V cos + £„. n 2 <

> sin «*) »

E„+ (p, *>) = L OW" cos 7^ + E ™

n=Q
Vin Sln n^) ,
(0a) - J An
,
(fa)
I

where
2

*MW = X- " -f '

Et {a
2*
f
Jo
*>) a>, =
tJo
t <p) cos <Lp,

£*.o
(2)
= 0, = - f EM v) '
sin n*> <b,

with similar equations for E^^ and £f,« (t

\

W^-(P, *>> = E KTE.n-(£ , n U) COS

,

>
71* + SID fl?) »

Fib.,

175

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
176 PROBLEMS

H,+ (P, *>--£ K TE ..+ (B,„<» cos + *.,<» *> »,) ffgg iffijff) •

„ ._, J.'{fia) -jN.'m

¥TE + ' - i N.<fia)
1

J.ffia)

H,-(p, *) = - f K™,-^" cos

n-0
+ sin «*•)
J»(PO)
'

_ j J,(go)

r) - £ W(*,,<» cos n, + W- *) ^gfgjgg

„ + Jnffia) - j N.Va)

3. Show that the field obtained in the preceding problem requires an electric
current sheet on the surface of the cylinder of radius a. Calculate the components of
the linear current density.
Ann. C t {a, 9) = ff*+ («, *) ~ *>)

- - £
n-0
(y TE .n+ + rTE .n-)(E,.»(1 '
cos *> + sin FI*).

C,(c v ) =H,-(a, v ) -H+{alV >)

+
= "
n-0
E (KTM .n + KTM ( n-)(^r.»
(,)
COB TUp + B,,w<« sin fly).

Note that, if CflB (1 \ C, (B (2) CV.« (1)

, , are the Fourier components of the current
density, then,

/* (1)

^ + .
V ' etc,

Kte, b+ + K TEl n- T-TTr :rrT"

4. Calculate the field of a given cylindrical current sheet of radius a (axially

uniform) in the presence of a perfectly conducting coaxial cylinder of radius 6 < a.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 PROBLEMS 177

Ana. The field ia the sum of the field given in Problem 2 and the following
field reflected from the cylinder.

Vfe r) - - E «J» cos + sm „) - -- - — ,

•jNn'tfifi)]

JnifiaKiJnW - jNn (fib)}

where °, £,,» <S) , etc., are expressed in terms of the current density as in Problem
3. The radial components are obtained by differentiating the tangential components
as in Problem 2.

5. Calculate tho field Btich that the component* of H tangential to a cylinder of

radius a assume given values H,{a,f>), H 9 (a,<p). The field is required to remain
finite inside the cylinder and vary as exp(-j0p) at infinity.

An*.

Hrifi, v) - L <ff«.»
(l)
«* "* + H <.» W sin nv-) ~ t

/•(ft.) -jNnifip)
H.+ (f>, v) - £
n-0
(ff,.»
(I >
cos + sin tup)
Jn (fla) -jNn (0a)

H9~(P, P) - t C*W» cos n* + sin n,) ,

,) - t cos „ + H„<» sin „,

g^fgj ,

B*.o
(,)
=
^ f*" B*{<h V) <b, =
jf* H t (a, *) cos rup

- 0, * -i-
J^' ff.(a, „) sin n* d*;

*) - -L 2w.W(ir#..(I) cos n* + ff..„ ( « sin ny>)

^ ^ i

ES(p, *)-£ W(ft-« cos nv + £f...<» sin

^"ffij*? -

«Tfa = E 2TE.n-(^.» (1)

cos r*> + sin n*) '

Original from
Digitized by LaOOvlC UNIVERSITY OF MICHIGAN
178 PROBLEMS

E,+ { PtV>) = - L +
^TE.» (ff,V
l)
C08flV + ffr .» ( »MnfI^)— — >

where the radial wave impedances ZxM.n and ^TE.n of the various modes are tl«
reciprocals of the corresponding wave admittances Ytm.h and KxE.n given in Problem
2. The radial components of E and // are determined by differentiation with respect
to as in Problem 2.

6. In the preceding problem the field

such that the electric intensity tan-
is

gential to the cylinder p o is discontinuous. This implies a source of the field in

=-

the form of a cylindrical magnetic current sheet (or double electric current sheet).
The linear density of the magnetic current sheet equals the discontinuity in E^ a \

thus,

M„(a, ?) - Et+ (a, *>) - Br (a, *>), MM 9>) - ^"(a, - E+(a, v ).

Express these quantities in terms of H at the surface of the cylinder.

Ana.

irpaMJn - jNn )

Jf^>*fcrt = - (W + «m.-»-M - -
„a JnV" _ JAr/) •

7. the answers to the preceding problems may be expressed com-

Note that
pactly in terms of the following transfer ratios and wave impedances and admittances

S» (ft>, /So)
= > SS(gp, /3a) =

T -
IR ^ J"' iM
. t+(* art
A'w-yjy.'w

7 ,-.
L
•/" (ga)
7 +,«^ , J.(flo) -iN,{0a) .

zTM z TE = KT„yT E = r*.

Id addition, we need the Fourier coefficients of a typical boundary function

= ^" (I> = FM 008

27Jo 7J
f*0(,) = 0, /»<» = -1 £' F(V ) sin dp.

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 PROBLEMS 179

Problems involving fields inside a wedge formed by perfectly conducting half-

planes *> = and ip - ^ require similar Fourier coefficients,

*V l) =-
\ F (*>) d*>, FM™ = 4 cos rupdv,

*V 2) = 0, /V a> = jfjf^ 8i » "V ^

where n = 2mjr/^, m = 0, 1, 2, 3, * • •

Some problems involve the sums of the wave impedances (or wave admittances),
looking in opposite directions. These sums may be simplified thus, ;

[2tm «+ ( W +-ZxM,n-(ito)r
1
= — TcQa
Jn(0a)[Jn(0a) -jNnifia)l

+
[2TE.» (^a) + ZTE.n'W)- 1
= Jn'mWnVia) ~ j

Z m+ + Ztm~ = vHYrn* + YTE -), Zte+ + Zte~ = i

2
(rTM+ + KTm").
To obtain the real and imaginary parts of the wave impedances we may con-
veniently use the following identities

Jn ~ j^n JnJn' + Nn N n + j(2/*j}g)'

Jn —jNn JnJn + ^n^n ~ jW^a)

Un> +
2
Jn-jNn
8. Solve Problems 2 and 3 for a typical wedge.
Arts.
E,-{p, E *»W Sn Pa)
= -(fip, sin n<p,

E,+ = E E..n S+(0P

( P , *>)
(2)
, fia) sin n*.

E<T(p, v) = E «mW ^) 003

V<ft *) - E rB+ (^P, 0a) cos n*,

it

#,-(p, *>) = E ^TE n-^a)£? n

n
l
n*. J,
<2>
Tn-f/Sp, fia) sin

#r
+ G>, *>) = ~ E YTE + ™T+ , n (fia)E. in n ((3p, fia) sin
n

^.-(P, ¥>> - ~ E r^.n-^oJ^.n^'SH-^p, 0a) COS TUp,

n

ff.
+ 0>, *») - E YTM n+ mE*.n S + COS <l)
n (fip, fia) n*,
,

where n - 2mw/^, m - 0, 1, 2, 3, ' . The radial components may be obtained by

differentiation with respect to v as in Problem 2.

The answer to Problem 3 for the wedge is the same as that for free space.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
180 PROBLEMS
9. Solve Problem 4 for the wedge.

Ant. The field is the sum of the field obtained in Problem 8 in which the
coefficients are expressed in terms of the Fourier coefficients for the current density
and the following reflected field

E/(p, *) = -L* £..* (2)

Sn-(fib, 0a) S.+ (0p, fib) ein m>,

H/( P , r) -En Sn-(0b, fia) 06) rTE .» + (0p) sin tup,

Vh *) = ~E ***<U 7V-03&, 0a) 7\+(0p, 06) cos n*>,

ff/fa - -Z ^.» <l)

*0 rB+ (0a, 06) Y-TM^iffp) cos n*>.

10. Solve Problems 5 and 6 for the wedge.

Ant. Hrip, *>) = E n

Sn-(fip, 0a) cos n*

H,+(p, *>) =En S«+ (0p, 0a) cos iup,

#r~0>. *) = - E ZTM.»-(0a)H,.n ™ Tn
n
-(fiP, 0a) cos n*.

+
V0». #) - E 2T M.» (0a)H.. n <» r.+fe* 0a) cos n*;

*) -E ***** rB-(|3p, 0a) sin n*

n

= E #r.*
n
(2)
r.+OJp, 0a) sin

ft~<A V) = En 2 TE.n-(0a)//^ n«> ( •S.Ttfp, 0a) sin n*>,

+ 0>, + (0a)^ <» 5„+(0p 0a)

tf. *>) = -E 2TEn
,« t(1 l sin n*.

The solution of Problem 6 for the wedge is identical in form with that for free
space.

11. Calculate the Fourier coefficients of a typical boundary layer function

defined by
P (*>) =0, < 9 < n,

- y <P1 <V <<P2,

= 0, VI < * <

for the wedge between *p - and <p =

JP.O-l- Bin *"^-">

Arts. fV»=4> cosfr(*.+**),

2 - <pi)
sin $n(*> 2
f" (2) = ,
.

7 *»(«-«)
where n = 2mr/^ m - f
0, 1, 2, 3"

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 PROBLEMS 181

12. Calculate the Fourier coefficients of a typical boundary layer' function

defined by
FW) = 0,

= J v>i < ifi < Vt,

fl — VI
= 0, <P2 < * < 2t,
for free space.
Ana. The same as in the preceding problem with ^ = 2x.

13. Consider a perfectly conducting half-plane, y> = 0. At distance a from the

edge there is an infinitesimal slot parallel to the edge. Let the voltage impressed
across the slot be Vq. Calculate the field and then the current in the plane.
Am.
HriP, *) = -0rr VoZlJn(iia) - j
l

n
Nu {fia)\ MM cos n*>, n = I1 1,

H+{P , V) = -/fcT'KoE Jn (0a)[JH (tip) -jNn P )) cos

((3 n*

E9~(p, *) = -^FoE^n^a) - j Nn m\ Jn'W cos TUP,

n

E++ {p, <p) * j^VoE Jn(0Q)Wn'(0p) ~ j COS n<p,

n

^+ (P, *) = -Jp-^oE ru/n (0a)|.M0p) -

n
j Nn (0p)) sin ny,

C p-(p) = //r(p,0) -//.-(p^x) = -2^- I

V'oE[-/ n (lSa) - j Nn m\ Jn(0P ),

n

Cp + (p) = tf,+ (p, 0) -H +(p, 2»r) - -2^-'KoE Jn{Pa)[J H {tSp) - j Nn (ttp)l

n

14. Calculate the field produced by an electric current filament I parallel to the
edge of a wedge formed by half-planes, >p = 0, ip = and passing through the point
(po, <po) where < <p < «A and < ^ < 2r.
Ana.

Et~{p,<p) = — El-AiGW — J AnOWl ^n(0p) sin n-p sin ny>,

* n

E,+ (P, *>) - _^E A(^Po)|/n (^) -J ATn ((Jp)| sin n^o sin n*,

where n = wt/^, m= 1,2, 3,' **. The remaining components of the field may
be obtained from

join dp wpp dip

15. Calculate the field produced by a magnetic current filament (a solenoidal

electric current filament) V parallel to the edge of the wedge defined in the pre-
ceding problem.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
182 PROBLEMS

An*. Hrb,*) - |[Jo(A») -iJV,(fr )] +

2E[./ B 0W -jNn (fiM)] Jn (0p) cosn^ocosn^i,

2 £ JniMWn<fip) - j X*(0p)] cos n*o cos

where 11 = mv/4>, m - 1, 2, 3/ • •. Tlie components of B may be obtained from

16. Using the results of the preceding problems, obtain the field of a plane
wave coming toward the wedge from the direction given by = w Consider two

Problem Fig, 1

cases: (1) E is parallel to the edge; (2) // is parallel to the edge. Let E and HQ
be the intensities of a plane wave 'at the edge on the assumption that the wedge is
absent.

Am. E, - —£ £
V n
V"*' 2 Jn (pp ) sin n* sin n*>, if E is parallel to the edge,

H. = — WoI^oOSp) +2 ./*(&,) cos n*, cos n*], if Wis parallel

to the edge,

17. Problem show two infinitely long wedge antennas. Assume that
Figs, la, 6
the boundaries are perfectly conducting and that the voltage is impressed between
the edges, .1, K uniformly in the axial direction. In this case the non vanishing com-
ponents of E and // are E 9t B //„. At first let us consider the low-frequency case in
m
which the solution may be expressed in terms of elementary functions. This will

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 PROBLEMS 183

illustrate the pattern of the general solution which differs from the essentially static
case largely in the type of functions involved.
At very low frequencies the magnetic field is small, and in the limiting case of
zero frequency it vanishes. The electric intensity may then be derived from a static
potential. Consider the double wedge (Fig. 16), and assume that the potentials of
the upper and lower wedges are, respectively, \ Vq and - 2 Vo- Show that the general
forms of the potential function in the antenna region and in free space are, respec-
tively :

ran
sin mp, n =
'-fi'+5[*(f)"+*(7)l' >

m-1,2,3,"-,

and V =£6*^*8***, ft = 1, 3, 5, • - •
.

Explain why C n must equal zero when the distance between the edges of the
wedges is vanishingly small.

18. Obtain the fields around the wedges.

Ana. If — Jir < <p < then, in the antenna region,

pEp = -L, nan 1

— ) sin rup, n = — 1
— » • • -
,

whereas, in free space,

pEP - - Z M>k cos hp,

pEp = - £ kb k sin hp, k m 1, 3, 5, • • . .

If |^| > \irt the field is obtained from the above by symmetry.
19. Express the 6 coefficients in terms of the a coefficients and vice versa.
Obtain independent sets of equations.

2Vn 4
Ana. 6* » —KKr + —T sin M £ o^tf
n
2
- n 2 )" 1 cos ntf,

a„ = - — cos n*£ 6*fc(fc

2 - n 2 )- 1
sin fttf.

20. Find the charge on the upper wedge per unit length of the wedge.

Ana. q - —* log — + 2eE

Pi n
(1 - Pina""K cos n# + 2e£ 6* cos k#.
A

The first sum represents the effect of the discontinuity on the charge distribu-
tion on the plane surfaces of the wedge and the second sum gives the charge on the
cylindrical surface.

Dig-t.zed by ^OO^k UNIVERSITY OF MICHIGAN

184 PROBLEMS
21. Obtain the initial approximations to the a's and 6's.

Ana. bk r
1 * * - 1,3,5, ••• ,
TK

m- 1, 2,3,.-.
22. Eliminate the o's from the equations in Problem 19.

2
. 2V 16 n« cos ni> sin at?

*,a = ,
1,
n r
3, 5, • • •
f
»- —TV 2r
—V,
3*
V

If we neglect those terms in the double sum for which a ^ k, then,

h = ^ [l +^ ** L " »"T ,
J*-
23. In the general case (as far as the frequency is concerned) the boundary
conditions are
A. At in6nity E„ H 9 vary as p"^ exp(-jftp).
B. In the antenna region the radial component of E vanishes at the surface of
the wedge; that is, for the antenna in Fig. la, Ep (0) - E^-d) - and for the
antenna in Fig 16, :
=- £„<-<») - « £,("i* - 0.
C. On the output boundary cylinder So, Ep {a) - when |?| > for the
antenna in Fig. la, and E 9 (a) - when & < \<p\ < \* + i> for the antenna in Fig.
16. Over the rest of the boundary E 9 and H, are continuous.
Z>. At the input boundary cylinder St, E 9 is given.

E. From symmetry considerations in Fig. 16, we find the additional condition

that the central plane MN
has in effect the boundary properties of a perfect mag-
netic conductor since

*,(J*) = Ep {-h) = and Ht <h*) « ff.(-Jir) = 0.

Obtain the proper values for n (see Problem 1) and the corresponding angular
wave functions for the wedge antenna in Fig. la.
Ana. The symmetric case: in the antenna region n — mr/#, m — 0, 1, 2,- «,

and 4> m (y) = cos (mry/d); in free spacen = 0, 1, 2,--', and 4> n (^) cosn*». The
antisymmetric case: in the antenna region n (2m + 1)t/2i>, m — 0, 1, 2 t
- , •

+
= sin[(2m 1W/2*|; in free space n - 0, 1, 2,- - •, - sin n*».

24. Obtain the angular wave functions for the wedge antenna in Fig. 16.

Ana. The symmetric case: $k(p) = cos rup, where n = mir/tf, 0, 1, 2,- m=
in the antenna region and n = 1, 3, 5,* • in free space. The antisymmetric case
* n (v>) = ain n*>, where n » (2m + \)t/2&, m
= 0,1, 2,- • in the antenna region
and n = 2, 4, 6, - • •, in free space.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 PROBLEMS 185

25. Show that the transverse voltage

'9

V{P ) = E9 (P
j ^ P , *>) aV

across the wedge depends only on the principal wave (the wave corresponding to
n = 0). Show that, as the radius pi of the input region approaches zero, the trans-
verse voltage V* 'p.) approaches a limit.

26. Obtain the field in the antenna region for the case in which the radius of the
input region is very small. Let Vo be the impressed voltage.
Am.

Br = «o Jo(0p) +V ( 4~ ) Xo(0p) +T,An Jnifip) cos nv t

^-f m -1,2,3, ••• ,

E,~ = jw Jo'Wp) - V tfo'CfcO H-^E A % Jn'ifip) cos tup,

Er ~ -~— L n^n Jnifip) sin n?,

27. Obtain the field in the free space region.

Am. H+ =IB» [Jnifip) - j Nn {0p)) cos n*>,
n

Ep+ - tiZ Bn Un'ifip) ~ j N n '{fip)} COS TUp,

n

E + = - -A- E nBn [Jn (fip) - j Nnifip)] sin n*,

3<&P n

where, in the case shown in Fig. la, n = 0, 1, 2, 3, - • •

, and, in Fig. 16, n = 1, 3, 5, • •

28. Express the B coefficients in terms of the A coefficients (Problems 26 and

27) for the case in Fig. la. Let the radius of So be a.
Ana. If n 0,

WrBnWniPa) - j N»' (*»)] =jT cos n*> a> =

j£ V cos n* aV

- -| ^flo /fl'Ote) - Vo AT/flta) sin n* +

J

Ja'lfia) I cos a<p coa rup d<p

a J-6

- |mo Jo'flSa) - Vo-^- (fia) J sin -

2j»j£ n(a* - n J rM, Ja'ifia) sin nt? cos a«J,

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
186 PROBLEMS
where the summation is extended over the set a = x/d, 2x/d, •

29. Express the A coefficients in terms of the B coefficients (Problems 26 and

27) for the case in Fig. 16.
Ans.

2da JoW+^-Vo No(fia)

- 2dB Q[Jo(Pa) - j Noifia)) +2 £
n— 1,2,3,...
n-'B^tfia) - j Nn (fia)] sin nd,

dA Ja ifia) -
tt 2 £
n-l.2,1,...
n(n 2 - a 2 )- 1
^/,^) - j Nn (fia)\ sin nd cos ad,

T 2x
""7'T
the equations in Problems 28 and 29 we can eliminate either the A' a or
From
the B's and obtain an infinite set of linear equations for either of the two sets of
coefficients. If the conditions are such that the A's are small, we can obtain the
B's by first neglecting the A's. Then we can calculate the A'b and recalculate the
B's, etc.

30. Calculate the input admittance per unit length of the dihedral horn in
Fig. la for the case in which the A's may be neglected. Let b be the radius of the
input boundary.

An*. Yt - JoWb) - N o (0b),

where
ao fog -Si
m j

V ton S2 '

2d
v sin 2 nd Jn (ga) - j Nn (pa) ,

2d „ sin'nd AQto) -jNnifia)

r -ft.... nv jjm-irtm %m *

r^r.r.rtl^ Original from

_.. ..
Digitized by V^OOgK UNIVERSITY OF MICHIGAN
 PROBLEMS 187

31. Show that, aa Sa-»«, the ratio Si/S2 -» -j and that, consequently,
the input impedance of the dihedral horn becomes independent of 0a.

32. Obtain a set of equations for the A'a alternative to that in Problem 29.
Ana.

&A a Ja'ifla) =2 £
•-1,8,...
n(n 2 - a 2 )" 1
Bn [Jn '(fia) - j N%'(0a)] sin n# cos

Note that these equations are not independent of those in Problem 28 and, hencei
are not sufficient for calculation of the A'a and B'b without the equations in Problem
29. They are useful as supplementary equations at points where J a (fia) vanishes.
Thus, from the two sets of equations for the A'b, we may obtain

*A a[Ja (fia) +jJ a '(8a))

=2
n—
Z1,2,...
Mn 2 - a 2 )" B n [(Jn + Nn') +j(Jn - Nn )] sin
1 '
nd cos atf.

In this set of equations the coefficients of the A a 'a cannot vanish.

33. Consider N+ 1 thin cones with their axes along (0 U n), (flj, tpi), '"i
(Pn, v>n)$ and (*-, 0). The cone angles are ^i, ^2,- and ^o- The currents in
the first N
cones are I\ t In, and the transverse voltages between these cones
"
and the ground cone " along (t, 0) are Vi, V2 , Fjf. Show that, for progres-
sive waves,

Vm = £
n-l
Kmn In ,

where

Kmn = -~ log (cos \en cos yn esc i&mn esc n^m,

= — log(cos 2 \em csc csc \$o), n — m,

where i?mB is the angle between the axes of the mth and nth cones.

Prove that, either in free space or in a region bounded by perfectly conduct-

34.
ing cones emerging from a common apex, the transverse fields of and TE modes TM
are orthogonal ; that is,

TM E9 TE + ^r.TM^.TE)
JJ (E $ , ,
= 0,

(H0 iTM IIe .TE + ff r .TMtf*.TE)tta =

JJ 0.

These equations may be UBed to determine the from the electric or magnetic field
intensities tangential to a given sphere (see Section 1.10). Thus, if et, c r are the
normalized components of the transverse electric intensity of a typical or TE TM
mode so that

2
+ eS)dn =
ff (et 1,

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
188 PROBLEMS
then the corresponding components of a given transverse field Ee. E p are Aft, Ae 9
where

{E*et + Ejb9) dU.

35. Show that, in the case of two cones, not necessarily coaxial, equation
2-54 for the normalized terminal admittance must bo replaced by

where ho represents the normalized distribution pattern for the magnetic intensity
of the TEM mode; that is,

So • to dU * 1.

36. Consider a sinusoidal current filament extending from I " li to s li

Let r\ and from the ends of the filament and a typical point. Let
r 2 be the distances
0i and 02 be the angles between the vectors drawn from the ends of the filament to
thio point and the filament. Show that, except for the electric intensities of the end
charges,

4*j<xE, - ffcijrfV"** - J'(« 2 )rr 1 e~*r»,

irjc^E, - I'(zi)e-'^tco6 02 - /'(zije-^cos*! -j$Hz s )e>'** +i0/(*i)e-*r >.

The retarded potentials of the end charges are

_ Kti)e-'^ I(z t )e-t**

Vi = f Fj m
4rjWi "
4*Ju*r 2

Show also that

**j&pH p - l'(z 2 )e- it,r* - I'(zi)e-'^ - jff I(z 2 )e-*r* cos0 2 +j0 /(*i)c
_,i!r»
cos t .

37. Consider a circular disk in an impressed field which is a function of the

distance from the center of the disk. Obtain the differential equations connecting
the scalar and vector potentials.

Ans = ~ j""ii4 ' + Ei{p)t {pAp) = ~jo"pV;

-

it

hence, P* ^ ^ +p - Ap + = -JmpV{p) t

38. Obtain the Green's function for the preceding problem (the response to a
concentrated radial voltage at p = po).
Am.
V = PJoWp), p <po,
= P Jotfp) + hr&poVtkJxifipo) N (f3p) - Nitfpo) / o 0?p)l, p >p ;

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 PROBLEMS 189

= /i05p) + fe-^poVolJViOSpo) /iOp) - /iOJpo) p > po;

where P is determined by the condition that the radial current at the edge of the disk
vanishes. In the first approximation,

*/i<M
where a is the radius of the disk.
39. Consider natural oscillations of an conducting plane
infinitely thin perfectly
plate of arbitrary shape. Show that the and both Car-
scalar electric potential V
tesian components Ax Av , of the vector potential satisfy the two-dimensional wave
equation,

d*W d
2
W , m

where p one of the natural oscillation constants.

is

40. Show that, if the plate described in the preceding problem is driven by a
given field, /7*'(x, y), Ejix, y), then,

-J-
= -M, - Ba *{z, V), = -M» " Vfc v)-

dAx dA v
-jwV.
dx dy

Hence,

2 2
dx dy dx dy

41. Consider an infinitely long, perfectly conducting strip of width 2a. Let
the strip be in the xz plane between the lines x = — o and x « a. Assume that the
impressed electric intensity is parallel to the x axis and is uniform in the z direction.
Obtain an integral equation for the linear current density C{x).

= A cos fix + B sin fit - 2*jrTx f sin /3(£ - x) d£, —a < x < a,

where A and B determinedby the end conditions C(— a) = C(a) — 0.

42. In the preceding problem assume an infinitely narrow slot along the z axis
and a voltage V across it. Obtain an integral equation for the current density.

Ans. C(£) Ko{jfi\x - $|) dt = A cos fix - jnT^O sin fi\x\.

J*
43. Obtain an integral equation for the case in which the strip of Problem 41
is in a uniform field of intensity Eq parallel to the z axis.

2t£
An$m J^^^tf* —a<x<a.
join

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
190 PROBLEMS
44. From Manueback'a solution for waves on a semi-infinite perfectly conduct-
ing wire of infinitesimal radius show that, when the radius is finite, the current at
large distances from the end is

I(r) - Ae~& ^21og-^- - * •

1^

Hcnco tho onorgy is slowly diverging from the wire. Thus tho powor flow within a
cylinder of radius b, coaxial with the wire, varies inversely as 2 log(2r/a) — 1.

Further analysis would show that the distance r is " large " if it is comparable to or
larger than X/2.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 APPENDIX I

Integrals in Antenna Theory

In Fig. 1, let

r = Vp 2 + z
2
, n = Vp2 + 2i 2 ,
r2 = Vp 2 +z 2
2
;

then,

C0Sg( r + 2) cfe = Ci + *) - Ci «rx + *x),

r

Z=Z :

Si/9(n + »i).
Changing z to — z,
z = z
-
J-'i
cos£(r
^ ™
z)
dz
,
= Ci /j(n - zi)
.
-
z= o
Ci 0(r2 - 22),

sin ff(r - z)
- z = z,
dz = Si/3(ri - zi)
J- FlQl
Si0(r2 -z2 ).
-

To prove the above, we introduce a new variable of integration,

t-r + w, dt = ^+ dz=
{Z
+ r) dz
= 4- d.

t
-= 4r
r

Also note that

f^i dl= fJlLfL m) = Siffli etc.

If is vanishingly small, the cosine integrals become

*
r
Also,
Vp2 + r
1<>g
r7+V J_,V?Tp-
2l0g
P

X* V(f-z) 2 +P 2

[Z 2 - Z + V(Z2 " *)
2
+P 2
]
[«-*! + V(Z - Zl)
2
+p 2
]
log 2
P
191

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
192 APPENDIX I

Let ri and r2 be the distances from points

and 2 (Fig. 2). Simi- 1

be the distance between points m and n. The following

larly, let r mn
integrals are taken along the straight line passing through point 2 be-
tween points 3 and 4

f cosfffa + r2 - ri2 cos t?) ^

= Ci 0(r i4 + r24 - r i2 cos tf) - Ci /3(r, 3 + r23 - r i2 cos tf),

r
.7(3,4 )
sin g(rj + r2

n
- r 12 cos fl)
^
= Si /3(ru + r2 4 - r« cos t?) - Si 0(ri 3 + r23 - r J2 cost?),

— +
J*(3.4)
cos &{ t \ r2
Tl
ri2 cos fl )
^
= Ci /3(n 3 - r23 +r J2 cos ) - Ci /3(r u- r24 + r l2 cos t?),

sinff(ri - r2 + ri 2 cost?) ^2
J*(3.4) n
= Si /3(ri 3 — r23 + ri 2 cos tf) — Si £(r i4 — r24 + n 2 cos <?).

To prove these results, use the following substitutions,

0(n + r2 — n 2 cos ^) = t and #(n — r2 + r\ 2 cos ^) = respectively*

Fia. 2.

In the more general case of two inclined coplanar segments (Fig. 3),
the exponential integrals may be expressed in terms of certain combina-
tions of various distances typified by

tu = r J4 -h Si - s4 " cos Ui = tax + s4 " - «i' cos tf.

Note that ri 4 — r 4 i is an essentially positive quantity while si' and s*"

may be either positive or negative. Typical exponential integrals

Digged by V^OOglC UNIVERSITY OF MICHIGAN

 APPENDIX I 193

associated with inclined segments are

[1-2)
— *4
ds' = e-W'«* [Ei(-j#24 ) - m-jftu)],

- [Ei(-i^i) - Ei(-i^31 )].

These integrals are evaluated by introducing new variables of integra-

tion, r4 + $' — $4" cos t? and n + s" — 81' cos The integral involv-

ing r4 — s' may be reduced to the first of the above integrals by reversing
the sign of the variable of integration 5' and changing t? to v — Thus,

= eW~# [Ei(-#d -
x
'ci-2) r4
-ds' 4) Ei(-i#24)].

A typical double integral of the exponential type is

Sa-2) J «(3-4) r

i^ (,+fl0e5,v [Ei(-^ 2 ) - B(~jN»)] -

g (l + coBtf)
- B(-jMu)] +
e -»(i+«o.*).,'[Ei(-i/3f4i)

e -Wi^*)n''[Ei(-#( 24 - Ei(-i^i4)] - )

-Wi+*.^.."[Ei(-^23 ) - m-j&iz)]) +
fi

/3(1 +
1
cost?)
m-mru + + «*")] - Ei[-#(r ft' 24 + «*' + *")] +
Ei[ -#(r23 + s2 ' + sb")] - Ei[-#(r13 + Si' + *")])
To evaluate this integral we introduce a new variable of integration,

t»r+e"-r cos
, , _«
^"
-y = —eft
•

Original from
Digitized by VjOOOIC UNIVERSITY OF MICHIGAN
194 APPENDIX I

This enables us to reduce the double integral to a simple integral. The

latter is integrated by parts and then another variable of integration,

fc = r4 + S '- S4 "cos,>,
-f
1 ---^'
is introduced. It should also be noted that

r4 " -r' =
2n"u
r4 + n + n" + r'
The integral involving r — s' -f s" is obtained from the above
integral if we change the sign of the variable of integration s' and
replace 1? by v — «?. The integral involving r — s' — «" is obtained by
changing the signs of both variables of integration without changing t?.

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 APPENDIX II

Inverse Radiation Impedance of a Thin Cone,

Zn = Rn+jXu
2x1
Rn = 30(Cin 2L - sin 2 L), L=—

Xn = 30Si2L - 15 sin 2/.

Rn Xu i/x Rn Xu i/x Rn Xu
a aa n oooo 000 .14 44 8031 04 tit iS 07 55 0500 35 Q8Q
a oooi 1 887 47
XI it7434
1(11 04 555 08 54 2315 37 453
n A9 A AA19 1 783
O./OO 10 K(\ 0931 04 901 09 53 5Q5S 1't 004

n fi'i n 0004 R oqq 37 RSt 4173 03 624 70 53 1591 40 7'H

o 04 7 043 38 56 1013 62 826 71 52.9313 12 607
o 05 0479 9 624 39 58 6521 61 816 0.72 52 9313 44 491
o oo 0985 11 649 40 61 0475 60 613 73 53 1562 46 412
a 07 AO- 728
13 *
1 ') 41 03 2fi74 59 232 74 53 0121 48 337
u.uo A 1A51 49 05 9019 57 08') 75 54 9Q81 50 944
a no 48.10 IS 017 41 07 Al»OiJ
Uli 1085 50 019 70 55 9101 59 0Q7
A lA 79(11 90 970 44 08 0QQ7 54 218 fl 77 50 3402 53 875
A
U. 11 1 1 0477 99 577 45 70 0550 52 338 n 78 57 0773 55 547
A 19 i 4 ri<i(t 24 430 40 1080
1 A. JUOv
71 50 390 79 59 2071 57 MS
A 11 97 197 47
XJ.tl 79 011Q 48 410
TO .111) 80 00
UU. Q 191
Jlti) 58 470
0.14 2.6065 29.768 0.48 72.6454 46.429 0.81 62.7727 59.693
0.15 3.3647 32.243 0.49 73.0097 44.462 0.82 64.7658 60.713
0.16 4.2605 34.734 0.50 73.1296 42.546 0.83 66.8670 61.526
0.17 5.3031 37.234 0.51 73.0129 40.702 0.84 69.0500 62.118
0.18 6.5003 39.730 0.52 72.6706 38.960 0.85 71.2872 62.482
0.19 7.8581 42.202 0.53 72.1168 37.345 0.86 73.5498 62.611
0.20 9.3806 44.634 0.54 71.3684 35.878 0.87 75.8090 62.502
0.21 11.0696 47.008 0.55 70.4450 34.584 0.88 78.0356 62.163
0.22 12.9248 49.309 0.56 69.3685 33.475 0.89 80.1993 61.588
0.23 14.9435 51.509 0.57 68.1630 32.572 0.90 82.2765 60.796
0.24 17.1205 53.602 0.58 66.8542 31.888 0.91 84.2385 59.850
0.25 19.4483 55.557 0.59 65.4694 31.428 0.92 86.0593 58.598
0.26 21.9168 57.365 0.60 64.0366 31.205 0.93 87.7171 57.230
0.27 24.5135 59.002 0.01 62.5848 31.217 0.94 89.1917 55.709
0.28 27.2237 60.458 0.62 61.1427 31.464 0.95 90.4654 54.054
0.29 30.0306 61.715 0.63 59.7386 31.944 0.96 91.5239 52.292
0.30 32.9153 62.760 0.64 58.4013 32.648 0.97 92.3557 50.456
0.31 35.8574 63.587 0.65 57.1574 33.566 0.98 92.9531 48.571
0.32 38.8348 64.184 0.66 56.0324 34.686 0.99 93.3116 46.664
0.33 41.8244 64.544 1.00 93.4307 44.766
195

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 APPENDIX III

Inverse Mutual Radiation Impedance of Two Arms

of a Thin Biconical Antenna

l/x i/X

0.00 0.0000 0.000 0.34 34.0367 13.694 0.67 -11.8799 -0.655

0.01 0.0001 0.729 0.35 35.4562 11.804 0.68 -11.9102 +2.604
0.02 0.0012 1.467 0.36 36.6873 9.739 0.69 -11.5482 5.863
0.03 0.0063 2.221 0.37 37.7065 7.458 0.70 -10.7940 9.073
0.04 0.0197 2.999 0.38 38.4922 5.094 0.71 -9.6535 12.187
0.05 0.0476 3.806 0.39 39.0258 2.698 0.72 -8.1386 15.158
0.06 0.0978 4.649 0.40 39.2912 0.177 0.73 -6.2669 17.941
0.07 0.1792 5.531 0.41 39.2759 -2.440 0.74 -4.0615 20.492
0.08 0.3017 6.453 0.42 38.9710 -5.049 0.75 -1.5510 22.771
0.09 0.4760 7.417 0.43 38.3716 -7.642 0.76 + 1.2311 24.741
0.10 0.7134 8.420 0.44 37.4770 -10.182 0.77 4.2469 26.369
0.11 1.0251 9.459 0.45 36.2906 -12.634 0.78 7.4544 27.627
0.12 1.4222 10.527 0.46 34.8204 - 14.964 0.79 10.8080 28.493
0.13 1.9154 11.618 0.47 33.0789 -17.136 0.80 14.2579 28.949
0.14 2.5144 12.720 0.48 31.0827 -19.117 0.SI 17.7592 28.984
0.15 3.2277 13.822 0.49 28.8528 -20.875 0.82 21.2548 28.592
0.16 4.0622 14.910 0.50 26.4143 -22.382 0.83 24.6944 27.773
0.17 5.0231 15.967 0.51 23.7957 -23.612 0.84 28.0260 26.541
0.18 6.1136 16.982 0.52 21.0293 -24.541 0.85 31.1988 24.903
0.19 7.3345 17.931 0.53 18.1503 -25.150 0.86 34.1636 22.883
0.20 8.6840 18.799 0.54 15.1962 -25.424 0.87 36.8739 20.309
0.21 10.1579 19.567 0.55 12.2068 -25.354 0.88 39.2866 17.610
0.22 11.7492 20.216 0.56 9.2232 -24.933 0.89 41.3625 14.814
0.23 13.4482 20.728 0.57 6.2874 -24.161 0.90 43.0673 11.580
0.24 15.2423 21.085 0.58 3.4414 -23.044 0.91 44.3718 8.147
0.25 17.1165 21.272 0.59 0.7272 -21.590 0.92 44.2525 4.564
0.26 19.0530 21.274 0.60 -1.8146 -19.815 0.93 45.6922 0.884
0.27 21.0319 21.079 0.61 -4.1452 -17.739 0.94 45.6804 -2.837
0.28 23.0312 20.676 0.62 -6.2278 - 15.250 0.95 45.2130 -6.547
0.29 25.0269 20.058 0.63 -8.0288 -12.645 0.96 44.2930 -10.186
0.30 26.9938 19.220 0.64 -9.5182 -9.977 0.97 42.9304 -13.701
0.31 28.9054 18.161 0.65 -10.6700 -6.991 0.98 41.1418 -17.036
0.32 30.7346 16.883 0.66 -11.4627 -3.869 0.99 38.9505 -20.142
0.33 32.4540 15.390 1.00 36.3864 -22.967

196

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 APPENDIX IV

The Inverse Radiation Resistance, A*„, and Reactance, X a of a ,

Thin Biconical Antenna with Equal Arms Making a 90° Angle,

and the Mutual Radiation Resistance, B12, between Its Arms
l/X Ka Xa fill I/X Ra xa I/X Ra xa
ft ft A
ft *} A4 fl "
C A*>
143. 1103
* '
A CT**> CA '):. 1A fl 1 A
U J.34 1

2.6743 J.D7 I30.3i!4 59.256 12.61^

A Al a aaao a ni 1 aJ. ok 1(11 CQA 1 Al Ot A an TtA O AIT
0.01 0.0002 4.041 u 3D 143. 340 3.0490 J. DO I
1^.017
A AJ. OC AO 1157
17 A1 ARO inn cd7
ft /M»
0.02
f\ ftft»* A
0.0024 1.
8.093 ftft*>
u 3D 1
lOo. 1
143.068 3.4331 AJ.JSA
by lzy.887 i*0 flO
6J.581 1 11 O jIO
11.348
OO
1 ft 1 ft AJ.37
it 1 AO OAO O OA AO A 7A 1 07 r> r
Q A TAI 1A t> AO
0.03 0.0128 12.188 V 114. 61b 142. ZdZ 3.8U0W 0.70 127.533 64.794 10.608
16.342 ft1 120.914 141.020 4.3556 0.71 125.478 67.327 9.8018
0.04 0.0394 D.38
0.05 ft AA4? ft
0.0960 AA
JO. ETfi
bib U.00U1 A OA 1 Ofi AAA 1 on ono 0.7^ 1
' 7U.143 O A4\OA
*.»43y
ft 1 ftT J ft QftT ft ftft/Vl ft J A 1 OT AEA Ota
ftn
73.194
1

0.06 0.1974 24.8971

0.0002 1.40 132.8^5 1 37.009 5.3652 J.73 1Z2.376 8.0316

0.07 OA J
...'.,.'_>
'
1
u.oum .

J.41
i 1
138.345 134.41 1
1 AAA* ALI. 7*
/4 IOI OOA 1ft A OK
/6.430 T AT TO
7.0778
1 jo rn OOA
ft ftO OO DC 4 o.oouy A O 1 1 A CI A A TC OA 7*Tfi TA AOAO
1 <'
0.08 0.6120 33.864 i
143. Oil 131.3/3 6.4674 U.75 1
1^0.776 7y.ojo 6.0898
0.09 0.9694 38.524 0.0017 0.43 148.308 127.964 7.0457 0.76 120.575 83.297 5.0776
I A J AO OA OAA
0.10 1.4.100 43.306 O.0O.i2 0.44 152.673 124. 2.i
1

1 7.6367 A TT
0.77 1 AA Till
120.7al 86.820 ttft rft
4.0504
0.11 2.1068 48.210 0.0057 AJ.45
4 C 1ta eoc
156.585 1OA O*iy
120. A 1
8.230C 0.7 A 70a IOI 00c A A OIC
y0.33b O AOAO
3.0202
A 1 O 1" h
Oft "-"I A V|
a «n
.1
IRA "in 1
1
1 nil
io.y/3
r O OA 1 K U.7V
8-8410 A TO 1 OO AMI
1 1 1
1.)
fil TOO
A A \ (\ A OA at en A AOAA
0.0 100 A 4 T lit KA1 A 1 OO SAO 1
0.1.1 58.360 111.547 V.4441 J. 80 1^3.803
0.14 5.2590 £0 coo
D3.582 A ao*ja J.48 1 ££ OTA 1 AAO
loo.yy^
1 u"* 1A AM 1
10.041 001 IOC ccc 1AA OTfl
1O0.376 A A1 AA
0.0100
0.15 6.7980 68.883 a jo 0.49
0.0343
<
1

167.268 t ao ocn 1A £0 1
10.624 A11.82
OA 1AT flTO
127.673 1AO A AO
103.403 —0.9294
ft ftftftj

0.16 8.6202 74.240 0.0496 0.50 168.640 97.714 11.190 0.83 130.092 106.205 -1.8208
0.17 10.746 79.630 0.0700 0.51 169.489 93.102 11.731 0.84 132.795 108.748 -2.6525
0.18 13.194 85.024 0.0967 0.52 169.825 88.579 12.242 0.85 135.746 110.998 -3.4140
0.19 15.978 90.389 0.1309 0.53 169.665 84.200 12.716 0.86 138.902 112.939 -4.0990
0.20 19.109 95.694 0.1740 0.54 169.029 80.017 13.146 0.87 142.225 114.542 -4.6965
0.21 22.595 100.905 0.2278 0.55 167.943 76.075 13.526 0.88 145.668 115.797 -5.2014
0.22 26.437 105.981 0.2938 0.56 166.441 72.423 13.852 0.89 148.186 116.692 -5.6061
0.23 30.634 110.884 0.3737 0.57 164.560 69.099 14.117 0.90 152.744 117.227 -5.9043
0.24 35.180 115.573 0.4694 0.58 162.341 66.140 14.316 0.91 156.291 117.401 -6.0930
0.25 40-061 120.016 0.5823 0.59 159.829 63.577 14.445 0.92 159.781 117.224 -6.1690
0.26 45.263 124.169 0.7148 0.60 157.073 61.438 14.500 0.93 163.174 116.707 -6.1302
0.27 51.764 127.997 0.8683 0.61 154.123 59.743 14.476 0.94 166.431 115.870 -5.9764
0.28 56.537 131.461 1.0449 0.62 151.031 58.505 14.373 0.95 169.514 114.731 -5.7082
0.29 62.554 134.534 1.2465 0.63 147.851 57.732 14.187 0.96 172.392 113.322 -5.3277
0.30 68.781 137.185 1.4750 0.64 144.637 57.429 13.917 0.97 175.033 111.669 -4.8390
031 75.177 139.385 1.7311 0.65 141.444 57.590 13.564 0.98 177.413 109.807 -4.2466
0.32 81.701 141.116 2.0157 0.66 138.322 58.204 13.129 0.99 179.509 107.772 -3.5570
0.33 88.308 142.360 2.3298 1.00 181.308 105.602 -2.7765

197

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 APPENDIX V
Inverse Radiation Impedance, Za = R a + iXc , of Two Thin Cones
Making an Angle t?

Ra
.% M set" 80° 100° 120° 140°
A 4V 160° 180°

o
01 0.000 o.ooo 0.000 0.000 0.000 0.000 0.000 0.000
02 0.001 0.001 0.002 0.003 0.004 0.004 0.005 0.005
003 0.003 0.006 0.010 0.015 0.019 0.022 0.024 0.025
04 0.009 0.020 0.033 0.046 0.059 0.070 0.076 0.079
05 0.022 0.048 0.079 0.112 0.143 0.169 0.185 0.191
06 0.046 0.099 0.163 0.231 0.295 0.347 0.381 0.393
0.07 0.085 0.182 0.300 0.425 0.542 0.637 0.699 0.720
0.08 0.144 0.307 0.506 0.717 0.914 1.074 1.178 1.214
09 0.229 0.487 0.803 1.136 1.446 1.698 1.861 1.918
10 0.345 0.734 1.208 1.708 2.174 2.550 2.795 2.879
0.11 0.499 1.062 1.746 2.466 3.134 3.674 4.024 4.145
0.12 0.698 1.484 2.436 3.437 4.365 5.113 5.597 5.764
13 0.948 2.013 3.303 4.653 5.903 6.908 7.557 7.782
v» A14* 1.256 2.665 4.366 6.144 7.784 9.100 9.949 10.242
15 1.629 3.452 5.648 7.936 10.041 11.726 12.811 13.185
16 2.072 4.387 7.167 10.055 12.704 14.819 16.177 16.645
0.17 2.593 5.483 8.942 12.525 15.800 18.406 20.078 20.653
18 3.197 6.749 10.989 15.364 19.349 22.511 24.533 25.228
19 3.888 8.196 13.321 18.588 23.368 27.147 29.558 30.385
20 4.671 9.831 15.948 22.208 27.865 32.322 35.157 36.129
0.21 5.551 11.662 18.877 26.229 32.844 38.035 41.328 42.455
22 6.529 13.692 22.113 30.654 38.302 44.277 48.057 49.348
023 7.608 15.923 25.656 35.476 44.225 51.031 55.321 56.783
24 8.789 18.356 29.502 40.686 50.596 58.267 63.086 64.726
0.25 10.072 20.990 33.644 46.267 57.387 65.951 71.310 73.130
26
VI A#w 11.457 23.819 38.069 52.198 64.565 74.037 79.940 81.940
27 12.941 26.838 42.764 58.450 72.088 82.471 88.913 91.091
0.28 14.523 30.038 47.709 64.990 79.907 91.193 98.160 100.510
0.29 16.198 33.408 52.881 71.779 87.968 100.131 107.603 110.115
30 17.963 36.936 58.254 78.775 96.210 109.213 117.155 119.818
0.31 19.811 40.606 63.798 85.930 104.567 118.355 126.728 129.526
0.32 21.737 44.404 69.482 93.192 112.968 127.474 136.225 139.138
33
WlWW 23.734 48.310 75.270 100.507 121.340 136.480 145.549 148.557
34 25.794 52.305 81.126 107.817 129.607 145.283 154.600 157.678
0.35 27.910 66.370 87.011 115.064 137.692 153.791 163.279 166.399
u.oo OA A7QO
iW.UI 92 886 A Mfl A188
122 <
' i
145 516 161.915 171.489 A174
W liUb
621A
0.37 32.274 64.621 98.709 129.128 153.004 169.566 179.135 182.248
0.38 34.505 68.762 104.442 135.825 160.080 176.659 186.127 189.187
0.39 36.757 72.886 110.042 142.222 166.673 183.115 192.383 195.356
0.40 39.021 76.968 115.472 148.262 172.716 188.861 197.826 200.677
0.41 41.288 80.989 120.694 153.894 178.147 193.831 202.391 205.087
0.42 43.550 84.927 125.671 159.069 182.911 197.968 206.022 208.528
0.43 45.798 88.762 130.371 163.743 186.960 20J.227 208.676 210.960
0.44 48.025 92.477 134.762 167.878 190.254 203.570 210.321 212.353
0.45 50.225 96.053 138.818 171.441 192.763 204.975 210.940 212.692
0.46 52.390 99.477 142.515 174.405 194.465 205.429 210.529 211.977
0.47 54.516 102.734 145.833 176.753 195.350 204.933 209.098 210.222
0.48 56.597 105.815 148.758 178.471 195.418 203.501 206.674 207.456
0.49 58.629 108.709 151.277 179.556 194.677 201.160 203.296 203.725
0.50 60.611 111.409 153.385 180.011 193.150 197.949 199.018 199.088

198

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 APPENDIX V 199
4

APPENDIX V {Continued)
Inverse Radiation Impedance, Za = R a -f iX a , of Two Thin Cones
Making an Angle
Ra
Q AO !
120° 140° c
i/X 00 80 100 lfxr 180

62 539 113 913 1 55 080 179.847 190.867 193.922 193.908 193.617

52 64 413 116 217 156 366 179.081 187.869 189.143 188.047 187.400
53 66 232 118
XA 322
—1 1
*-> *-* 157 249 177.741 184.209 183.687 181.527 180.534
54 67 997 1 20 230 157 743 175.860 179.947 177.642 174.451 173.129
55 69 710 121 947 157 864 173.477 175.152 171.101 166.932 165.304
n 56 71 374 1 23 480 1 57 632 170.638 169.900 164.169 159.091 157.183
57 v — 990
72 -
if \J
if 1 24 836 1 57 073 167.395 164.275 156.955 151.053 148.901
58 74.564 126.028 156.214 163.806 158.366 149.575 142.950 140.591
59 76 100 1 27 066 155 087 159.930 152.267 142.146 134.915 132.393
60 77 60 'I 1 97 966 1 53 794 155.833 146.072 134.790 127.082 124.444
o in 7Q 078 1 98 749 1 59 1 6.3 151.582 139.881 127.626 119.582 116.879
62 80 532 129 100 150
A W
441
* * A * 147.245 133.792 120.774 112.544 109.829
63 81 969 1 29 984 148 595 142.891 127.901 114.347 106.090 103.420
64 83 397 1 30 484 1 46 667 138.590 122.305 108.455 100.336 97.766
65 84 891 1 30 926 144 694 134.410 117.093 103.202 95.387 92.975
66 86 247 131 326 142
X 1 716
I X w
*- n 130.417 112.353 98.680 91.338 89.139
67 87 681 131 701 1 40 770
J 1 IU '
. 126.674 108.163 94.974 88.271 86.340
68 89 128 132 066 138 893 123.239 104.596 92.157 86.253 84.643
69 90 593 132 436 XV f * X
137 119
Xc 120.169 101.716 90.289 85.337 84.095
70 92 079 132 824T
A \fA* \
~ 1 35 480 117.510 99.575 89.417 85.560 84.730
71 93 509 1 33 243 1 34 003 115.308 98.218 89.571 86.940 86.562
72 95 133 1 33 702 X KJ *0 716
132 f X U * 113.597 97.677 90.770 86.480 89.585
73 96 705 A134i
' > 212
Ad A Ad + 131 639 112.408 97.972 93.015 93.164 93.779
74 98 309 134 77R 130 792 111.763 99.114 96.292 97.959 99.101
75 99 946 1 35 406 1 30 1 87 111.674 101,097 100.572 103.813 105.494
n 76 , 101615 136 100 1 99 836 112.148 103.908 105.811 110.660 112.882
77 103 314 1 36 860 199 744 113.184 107.518 111.948 118.417 121.174
n 78 105 049 1 37 685 199 919 114.772 111.889 118.911 126.986 130.263
79 106 796 138 574 130 338 116.894 116.970 126.615 132.256 140.030
n so 108 571 1 39 521 131 015 119.524 122.702 134.962 146.105 150.344
81 1 1 363 140 590 122.633 129.013 143.843 156.400 161.064
82 112 167 141
A * A .563 (Uw 133 076 126.179 135.827 153.142 167.001 172.041
0.83 113.976 142.641 134.429 130.120 143.056 162.736 177.760 183.123
0.84 115.785 143.743 135.971 134.406 150.608 172.496 188.529 194.152
0.85 117.585 144.856 137.677 138.982 158.387 182.289 199.155 204.972
0.86 119.371 145.969 139.523 143.791 166.291 191.981 209.488 215.427
0.87. 121.134 147.069 141.483 148.771 174.219 201.440 219.380 225.366
0.88 122.867 148.141 143.527 153.861 182.066 210.534 228.690 234.644
0.89 124.561 149.171 145.626 158.996 189.732 219.139 237.283 243.127
0.90 126.210 150.147 147.752 164.114 197.115 227.134 245.035 250.690
0.91 127.806 151.055 149.875 169.150 204.121 234.408 251.833 257.221
0.92 129.341 151.882 151.966 174.044 210.658 240.861 257.579 262.624
0.93 130.810 152.617 153.999 178.738 216.643 246.402 262.189 266.819
0.94 132.206 153.250 155.948 183.176 222.000 250.955 265.595 269.744
0.95 133.524 153.771 157.789 187.308 226.662 254.459 267.750 271.357
0.96 134.759 154.173 159.500 191.089 230.571 256.866 268.622 271.634
0.97 135.907 154.450 161.064 194.478 233.683 258.145 268.200 270.572
0.98 136.965 154.597 162.465 197.441 235.962 258283 266.494 268.190
0.99 137.932 154.613 163.691 199.950 237.385 257.282 263.530 264.524
1.00 138.805 i
154.497 164.733 202.041 237.942 255.162 259.357 259.634

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
200 APPENDIX V

APPENDIX V (Continued)

Inverse Radiation Impedance, Za = R a -f iX„ of Two Thin Cones

Making an Angle t?

l/\
*/ ,%
(/
= 4n°
t\i 100° 120° 140° fin°
1i\nj i cn°

0.01 2.219 3.059 3.746 4.292 4.708 5.001 5.175 5.232

0.02 4.444 6.129 7.509 8.608 9.445 10.034 10.385 10.501
0.03 6.680 9.222 11.306 12.969 14.238 15.132 15.664 15.841
0.04 8.932 12.346 15.153 17.398 19.114 20.325 21.045 21.284
0.05 11.206 15.514 19.065 21.914 24.096 25.637 26.556 26.861
0.06 13.506 18.732 23.056 26.533 29.204 31.094 32.221 32.596
0.07 15.836 22.008 27.135 31.271 34.454 36.712 38.059 38.507
0.08 18.200 25.349 31.312 36.137 39.860 42.503 44.083 44.609
0.09 20.600 28.760 35.592 41.138 45.426 48.475 50.299 50.906
0.10 23.037 32.242 39.979 46.277 51.154 54.626 56.705 57.396
0.11 25.515 35.798 44.472 51.550 57.040 60.950 63.292 64.071
0.12 28.031 39.425 49.069 56.952 63.071 67.432 70.043 70.912
0.13 30.587 43.123 53.763 62.471 69.233 74.050 76.932 77.892
0.14 33.179 46.885 58.543 68.091 75.501 80.775 83.928 84.977
0.15 35.806 50.704 63.396 73.789 81.846 87.571 90.990 92.126
0.16 38.465 54.573 68.307 79.541 88.233 94.396 98.068 99.288
0.17 41.151 58.481 73.255 85.317 94.623 101.200 105.110 106.407
0.18 43.859 62.415 78.218 91.083 100.969 107.930 112.056 113.421
0.19 46.584 66.362 83.172 96.802 107.225 114.528 118.840 120.264
0.20 49.318 70.306 88.089 102.433 113.336 120.930 125.394 126.864
0.21 52.055 74.230 92.940 107.935 119.247 127.073 131.647 133.150
0.22 54.788 78.118 97.694 113.262 124.903 132.890 137.527 139.045
0.23 57-508 81.951 102.321 118.369 130.246 138.313 142.962 144.477
0.24 60.207 85.709 106.788 123.211 135.217 143.278 147.880 149.373
0.25 62.877 89.374 111.062 127.743 139.762 147.719 152.213 153.661
0.26 65 509 92.927 115 112 131.920 143.825 151.578 155 897 157 278
0.27 68.094 96.347 118.906 135.699 147.357 154.797 158.874 160.165
0.28 70.625 99.619 122.416 139.042 150.311 157.327 161.091 162.269
0.29 73.093 102.722 125.612 141.911 152.644 159.124 162.505 163.546
0.30 75.490 105.642 128.469 144.274 154.322 160.152 163.082 163.963
0.31 77.809 i 108.362 130.964 146.105 155.317 160.384 162.796 163.495
0.32 80.044 110.869 133.077 147.379 155.606 159.803 161.634 162.130
0.33 82.189 113.152 134.792 148.080 155.178 158.403 159.594 159.868
0.34 84.239 115.199 136.096 148.199 154.027 156.185 156.684 156.721
0.35 86.189 117.003 136.981 147.729 152.158 153.165 152.928 152.714
0.36 88.035 118.559 137.442 146.675 149.585 149.369 148.360 147.884
0.37 89.776 119.862 137.479 145.045 146.330 144.833 143.027 142.283
0.38 91.410 120.912 137.096 142.856 142.427 139.604 136.987 135.972
0.39 92.936 121.710 136.303 140.130 137.916 133.742 130.310 129.027
0.40 94.355 122.259 135.112 136.899 132.847 127.315 123.079 121.532
0.41 95.667 122.567 133.541 133.197 127.279 120.399 115.381 113.582
0.42 90.877 122.040 131.613 129.067 121.276 113.081 107.381 105.281
0.43 97.986 122.489 1

129.352 124.558 114.911 105.453 98.995 96.740

0.44 98.998 122.128 126.788 119.723 108.263 97.616 90.525 88.075
0.45 99.919 121.569 123.953 114.618 101.414 89.672 82.024 79.409
0.46 100.754 120.830 120.884 109.305 94.451 81.729 73.611 70.863
0.47 101.509 119.928 117.617 103.848 87.464 73.896 65.406 62.560
0.48 102.191 118.880 114.192 98.314 80.545 66.281 57.530 54.627
0.49 i
102.807 117.708 110.652 i 92.769 73.784 58.992 50.097 47.175
0.50 I
103.363 ;
116.431 107.037 1 87.283 i
67.271 52.135 43.219 40.324

_.. .. r*r«r»rtl/» Original from

D,g,t*ed by VjOOglL UNIVERSITY OF MICHIGAN
 APPENDIX V 201

APPENDIX V (Continued)

Inverse Radiation Impedance, Z = Ra +

a iX a , of Two Thin Cones
Making an Angle i>

l/\
f
& - 40° 60" 80° 100" 120° 140" 160° 180°

0.51 103.868 115.071 103.391 81.922 61.095 45.808 37.004 34.180

0.52 104.329 113.647 99.756 76.753 55.340 40.107 31.546 28.840
0.53 104.753 112.181 96.174 71.839 50.084 35.116 26.935 24.391
0.54 105.148 110.693 92.685 67.240 45.401 30.914 23.247 20.911
0.55 105.521 109.202 89.327 63.014 41.357 27.567 20.545 18.460
0.56 105.878 107.728 86.138 59.212 38.009 25.131 18.881 17.087
0.57 106.225 106.288 83.151 55.879 35.407 23.648 18.290 16.824
0.58 106.568 104.897 80.396 53.058 33.589 23149 18.791 17.689
0.59 106.911 103.571 77.901 50.779 32.584 23.648 20.390 19.679
0.60 107.259 102.322 75.689 49.070 32.408 25.148 23.074 22.779
0.61 107.615 101.160 73.778 47.950 33.067 27.636 26.815 26.954
0.62 107.981 100.095 72.184 47.428 34.556 31.085 31.568 32.153
0.63 108.360 99.133 70.917 47.509 36.857 35.452 37.273 38.309
0.64 108.750 98.279 69.982 48.185 30.042 40682 43.856 45.341
0.65 109.154 97.534 69.381 49.445 43.770 46.708 51.228 53.153
0.66 109.569 96.900 69.110 51.267 48.292 53.447 59.286 61.634
0.67 109.993 96.373 69.162 53.623 53.445 60.808 67.918 70.666
0.68 110.424 95.951 69.523 56.478 59.162 68.689 77.000 80.118
0.69 110.858 95.628 70.180 59.788 65.363 76.980 86.403 89.852
0.70 111.292 95.395 71.110 63.508 71.964 85563 95.988 99.725
0.71 111.720 95.243 72.292 67.582 78.874 94.316 105.614 109.590
0.72 112.137 95.163 73.700 71.953 85.996 103.113 115.139 119.298
0.73 112.538 95.143 75.304 76.561 93.233 111.826 124.419 128.702
0.74 112.916 95.169 77.074 81.339 100.482 120.327 133.313 137.658
0.75 113.265 95.229 78.978 86.224 107.642 128.491 141.687 146.027
0.76 113.580 95.309 80.982 91.146 114.612 136.197 149.409 153.677
0.77 113.854 95.396 83.053 96.039 121.295 143 329 156.360 160.488
0.78 114.080 95.475 85.156 100.837 127.595 149.780 162.428 166.349
0.79 114.253 95.533 87.258 105.475 133.424 155 451 167.518 171.164
0.80 114.367 95.558 89.327 109.891 138.699 160.25o 171.544 174.8o3
0.81 114.418 95.538 91.331 114.029 143.346 164.116 174.438 177.350
0.82 114.400 95.462 93.242 117.833 147.297 166.972 176.149 178.609
0.83 114.311 95.321 95.033 121.256 150.497 168.777 176.6-43 178.602
0.84 114.146 95.106 96.680 124.255 152.900 169.497 175.904 177.319
0.85 113.904 94.812 98.161 126.794 154.473 169.117 173.935 174.771
n Rfi 1 ^st
11 1io.Ooo GO A PJ1
yy.^ou 1 OQ QAt
I/O. ©to 1 00.11*0 ID 1 wo 1 7A 7>"ifi 1 (U.'JOo
0.87 113.183 93.967 100.562 130.381 155.049 165.075 166.409 166.018
0.88 112.704 93.414 101.457 131.394 154.045 161.461 160.951 159.929
0.89 112.140 92.773 102.139 131.873 152.195 156.845 154.457 152.806
0.90 111.514 92.049 102.604 131.820 149.526 151.290 147.020 144.751
0.91 110.809 91.245 102.855 131.244 146.077 144.875 138.746 135.882
0.92 110.035 90.370 102.896 130.160 141.898 137.690 129.755 126.327
0.93 109.198 89.431 102.735 128.592 137.050 129.839 120.180 116.229
0.94 108.303 88.438 102.385 126.569 131.602 121.435 110.162 105.738
0.95 107.357 87.404 101.861 124.127 125.635 112.601 99.851 95.013
0.96 106.365 86.341 101.181 121.309 119.234 103.467 89.401 84.217
0.97 105.337 85.264 100.365 118.160 112.493 94.166 78.971 73.514
0.98 104.279 84.186 99.437 114.732 105.508 84.836 68.720 63.070
0.99 103.201 83.124 98.420 111.080 98.382 75.617 58.805 53.045
1.00 102.110 82.093 97.340 107.259 91.217 66.646 49.377 43.596

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 APPENDIX VI

The Inverse Mutual Radiation Resistance of Two Inclined Cones

The mutual radiation resistance of two inclined wires (Fig. 4) with

reference to the maximum current amplitude is

R12 = 15 cos(Li - L2 )(Cin x\ + Gin x2 ) -

15 cos(Li + L2 )(Cin x3 + Cin xt) +
15sin(Li - L2 )(Sixj - Six2 ) +
15 sin(Li + L2 )(Si x + Si s + z 4)

30 sin Li sin La(l - Cin 2Li - Cin 2L2 ) -

30 sin Li cos L 2 Si 2L2 — 30 sin L2 cos Li Si 2Li,

where

U = /Mi L 2 = ph = ——i '

d2 = li
8 - 2W 2 cos i? + ia
2
,

x2 = fi(h - h + d),
*s - P(h + h - d), X* = fi(h + h + d).

202

Original from
Digitized byGoOglc UNIVERSITY OF MICHIGAN
 APPENDIX VII
The Inverse Mutual Radiation Resistance of Two Thin Cones,
of Lengths and /, Making an Angle 180°
/L .

203

Original from
Digitized by Google UNIVERSITY OF MICHIGAN
 APPENDIX VIII

M(0l) and N(fil) for Cylindrical Antennas, for Use in

Equation 1-136

i/X Miel)
\f"/ N(8l) l/\
"i M(8l) l/X M(8l) N(Bl)

0.34 46.83 156.48 0.67 64.02 46.65

0.01 -0.2364 0.01322 0.35 56.22 157.64 0.68 59.34 51.79
0.02 -0.9399 0.1054 0.36 65.63 157.87 0.69 55.32 57.59
0.03 -2.094 0.3542 0.37 74.95 157.19 0.70 52.05 63.95
0.04 -3.671 0.8339 0.38 84.09 155.59 0.71 49.58 70.76
0.05 -5.634 1.615 0.39 92.93 153.10 0.72 47.97 77.94
0.06 -7.934 2.761 0.40 101.37 149.76 0.73 47.26 85.36
0.07 -10.516 4.331 0.41 109.31 145.61 0.74 47.46 92.91
0.08 -13.31 6.373 0.42 116.66 140.71 0.75 48.60 100.49
0.09 -16.26 8.929 0.43 123.34 135.14 0.76 50.66 107.96
0.10 -19.28 12.03 0.44 129.27 128.97 0.77 53.62 115.21
0.11 -22.28 15.68 0.45 134.38 122.31 0.78 57.46 122.14
0.12 -25.20 19.92 0.46 138.63 115.24 0.79 62.12 128.63
0.13 -27.93 24.72 0.47 141.96 107.87 0.80 67.55 134.59
0.14 -30.41 30.07 0.48 144.35 100.32 0.81 73.68 139.92
0.15 -32.54 35.95 0.49 145.78 92.68 0.82 80.41 144.54
0.16 -34.25 42.32 . 0.50 146.26 85.09 0.83 87.66 148.38
0.17 -35.47 49.14 0.51 145.79 77.64 0.84 95.33 151.38
0.18 -36.12 56.34 0.52 144.40 70.46 0.85 103.30 153.50
0.19 -36.15 63.86 0.53 142.13 63.65 0.86 111.48 154.69
_
— )O0.01
C El 71 A1
I l.DO
A £1
U.U'i 1 dVf.Uo D/.ol U.oi '•H.JJ
0.21 -34.15 79.56 0.55 135.16 51.53 0.88 127.95 154.26
0.22 -32.04 87.57 0.56 130.61 46.42 0.89 136.02 152.65
0.23 -29.17 95.56 0.57 125.45 42.03 0.90 143.83 150.12
0.24 -25.52 103.44 0.58 119.78 38.45 0.91 151.25 146.73
0.25 -21.10 111.12 0.59 113.71 35.71 0.92 158.19 142.53
0.26 -15.93 118.49 0.60 107.34 33.88 0.93 164.56 137.57
0.27 -10.031 125.47 0.61 100.79 32.96 0.94 170.25 131.95
0.28 -3.446 131.96 0.62 94.17 32.99 0.95 175.20 125.74
0.29 +3.771 137.88 0.63 87.59 33.95 0.96 179.34 119.04
0.30 11.560 143.16 0.64 81.18 35.83 0.97 182.61 111.96
0.31 19.85 147.71 0.65 75.04 38.60 0.98 184.96 104.60
0.32 28.55 151.48 0.66 69.29 42.23 0.99 186.39 97.09
0.33 37.57 154.42 1.00 186.86 89.53

204

Digtlized by V^OOgK UNIVERSITY OF MICHIGAN

 APPENDIX IX
M(Z) and N(l) for Tapered Antennas

Original from
Digitized by Google UNIVERSITY OF MICHIGAN