Proceedings of the Institute of Radio Engineers Volume 20, Number 6

June, 1932

CIRCUIT RELATIONS IN RADIATING SYSTEMS AND APPLICATIONS TO ANTENNA PROBLEMS* BY P. S. CARTER

(R.C.A Communications, Inc., Rocky Point, L. I., N.Y.)

Summary-Expressions for the self and mutual impedances within a radiating system are developed by the use of the generalized reciprocity theorem. These expressions are given in terms of the distributions of the electric field intensities along the radiators. A method for the determination of the field intensities is outlined. Formulas for the self and mutual impedances in several types of directional antennas are given. Questions of practical interest in connection with arrays of half-wave dipoles, long parallel wires, and "V" type radiators are discussed. Different types of reflector systems are considered. Curves of the more important relations are shown. The mathematical development is shown in an appendix.

N THE design and the adjustment of antenna systems a knowledge of certain characteristics and relations is of great assistance. We should know the theoretical directivity, that is, the ratio of the intensity of radiation in a desired direction to the mean intensity in all directions. The contribution of each radiating element to the total radiated power and the interactions between elements are important. In a good system the ratio of heat losses to radiated power must be low. The intensity of radiation in the desired direction is relatively easy to obtain. To determine the total power we may, for mathematical purposes, imagine the system placed at the center of a very large sphere and compute the power flow through each element of area on the sphere. A summation gives the total. The average intensity is then this total divided by the number of units of solid angle contained in the sphere. The application of this method to long linear radiators and several types of directional antenna systems has been shown by the writer in detail.1 Upon completion of this process we have a complete knowledge of the power flow in every direction in space but are left in entire ignorance as to the portions of this power contributed by the various antenna elements and as to the interactions between these elements. To the communications engineer, who is quite familiar with the use of impedance operators in connection with ordinary circuit calcula-

* Decimal classification: R116. Original manuscript received by the Institute, March 1, 1932. Presented before Twentieth Anniversary Convention of the Institute, Pittsburgh, Pa., April 9, 1932. l Carter, Hansell, Lindenblad, "Development of directive transmitting antennas by R.C.A. Communications, Inc.," PROC. I.R.E., vol. 19, pp. 1773-1842; October, (1931). 1004

Carter: Circtit Relations in Radiating Systems

1005

tions, the question may arise "Why not determine the power at the circuit itself rather than at a great distance?" As a matter of fact, except for the distribution of the radiated power in space, we may determine everything we need to know at the wires. Indeed, we must work directly at the antenna system if we wish to determine what is happening in it. In the following discussion an attempt is made to reduce problems in connection with radiation systems to such a form that they may be handled by the ordinary circuit methods of engineering and to show the results of -certain applications of importance to directional antennas. PART I--GENERAL PRINCIPLES 1. CIRCUIT LAWS AND RECIPROCITY IN RADIATING SYSTEMS Assume that, inside of a closed box, we have a circuit network consisting of inductances, capacities, and resistances (vacuum tubes, iron core inductances, and other nonlinear devices excluded) connected in an unknown manner but having two pairs of terminals, a and b, at opposite ends of the box. Now suppose we apply an a-c voltage at a and measure the current through an ammeter connected to terminals b. Then let us reverse the process, placing the ammeter at a and applying the voltage at b. The current is alike in both cases both in magnitude and phase angle. This is the reciprocity theorem for ordinary circuits. If we call the ratio between the voltage at a and current at b, Zab and the ratio under the reverse conditions, Zba, the reciprocity law then states that Zab = Zba. This impedance is generally known as the transfer impedance. Carson2 and Ballantine3 have given proofs of a more generalized reciprocity theorem which holds without any restrictions in all cases to be considered in this discussion. Consider a linear radiator as shown in Fig. 1. We know from experimen-t that, to a fair degree of approximation, the current is sinusoidally distributed. We wish to know what voltage we must impress at a current loop in order to produce a given current. Assume that we know the distribution of the parallel component of the electric force E, along the radiator due to the sine wave distribution of current IA sin mZ where m = 2r/X. Since the e.m.f. or voltage is the line integral of electric force, the voltage induced in an element dz due to the current
2 Carson, "A generalization, of the reciprocal theorem," Bell Sys. Tech. Jour., July, (1924); "Reciprocal theorems in radio communication," PROC. I.R.E. June, (1929). 3 Ballantine, "Reciprocity in electromagnetic and other systems," PROC. l.R.E., June, (1929).

1006

Carter: Circuit Relations in Radiating Systems

IA at the loop position A must be Ezdz. By the reciprocity law a current IA in the element dz would induce a voltage E,dz at A. However, the actual current in the element dz is IA sin mz. Calling the voltage induced at A, dV, we then have the relation: dV E2dz ~~~~~~~~~~(1) IA IA sin mz or, (2) dV = E, sin mz dz. The total voltage at A, due to current in all elements of the wire, is then: V= J Esinmzdz. (3)
_
(S

The impressed voltage must be equal and opposite to the induced e.m.f. Let us define the self-impedance of a radiator as the ratio be-

Fig. 1

tween the impressed voltage and current at a current loop position when this radiator is under no influence from other radiators. We then have the relation: Vil VI
z11 =- =

yZ

__

(4)__

where Zi, is the self-impedance of radiator No. 1 due to its own current. VI is the impressed e.m.f., YV' the induced e.m.f., and I, the loop

Carter: Circuit Relations in Radiating Systems

1007

current, the dots indicating that the phase must be taken care of by the standard complex algebra method of electrical engineering. Assume that we have a second radiator No. 2 at any distance and oriented in any manner with respect to radiator No. 1. Assume we know the current in No. 1 and that this produces a known distribution of electric force along No. 2. Assuming that the current in wire No. 2 must be sine wave distributed, the effective induced e.m.f. at a current loop of wire No. 2 will then be:

V21=

E2

sin mz dz.

(5)

Let us define the mutual impedance as the negative ratio of the induced e.m.f. at a current loop of radiator No. 2 to the loop current of radiator No. 1. Because of the reciprocity law this mutual impedance is the same in the reversed direction and we have the relation:

Z12 = Z21-

-V21 * I,
-

(6)

Provided we know the values of the self and mutual impedances we can now determine the relations in any system having n linear radiators and impressed voltages V1, V2, , V,n at the current loops. The resulting relations are the same as for ordinary circuit networks, the difficulties in the way of the solution in the more complex cases being the same. The general relations then are: V1 = IlZl + I2Z12 + * + InZln V2 = IhZ12 + I2Z22 + + InZ2n

(7)
.

..

..

*.* ...

..

..

*.**

Vn = IlZln + I2Z2n + ** I +nZ4 n

As an example, consider the case of two like radiators 1 and 2, where a voltage V1 is impressed in radiator No. 1 only. We then have the relations: V1 = IlZll + I2Z12 = I1Z12 + U2N22( If no impedance is placed in series with either, Z11 = Z22 and the relations giving the currents become:

Z112

, zii - Z12

19

1 12 -

(9)

1008

Carter: Circuit Relations in Radiating Systems

If we disregard heat losses, the power radiated from the nth radiator in any system is VnIn cos 0, where 0 is the phase angle, or IFR where R is the real part of the complex impedance operator Z. In most cases in short-wave practice we are justified in neglecting heat losses within a radiator but in the feeding system they may be considerable. Before we can apply the method outlined to any actual cases, we must know the impedances and, as already shown, the values of these impedances are determined by the field conditions in the vicinity of the radiators.
2. FIELD CONDITIONS IN THE VICINITY OF A RADIATOR In the determination of the field conditions in the vicinity of a radiator, the writer has used an extension of the method outlined by Pistolkors.4 During this work a paper by Bechman' appeared in which he gives an excellent method of determining the fields. At the conclusion of this work a second paper by Bechmann6 was published. The latter covers a considerable part of the theory underlying the discussion to follow. Bechmann derives his electric and magnetic forces from a Hertz vector while the writer has used the scalar and vector potentials but, of course, both methods become identical in the results. Faraday's law of induction and Maxwell's revised form of Ampere's law, which form the foundation of electromagnetic theory, are: (a) The line integral of the electric force around any complete circuit is equal to the negative rate of change of magnetic induction through the circuit. (b) The line integral of the magnetic force around any complete circuit is equal to the total current, conduction, and displacement, through the area bounded by the circuit. The six partial differential equations known as "Maxwell's equations" are the differential form of these laws. We cannot directly determine the forces from these equations as they are in an implicit form. In the quasi-static, or low-frequency case, the magnetic force may be derived from a vector function called the vector potential. The reason for the name vector potential is that this vector is derived for a given current distribution in a manner similar to that used in deriving the scalar potential for a given charge distribution. To determine the electric force we need both the vector and scalar potentials.
I Pistolkors, "The radiation resistance of beam antennas," PROC. I.R.E., March, (1929). I Bechmann, "Calculation of electric and magnetic field strengths of any oscillating straight conductors," PROC. I.R.E., March, (1931). 6 Bechmann, "On the calculation of radiation resistance of beam antennas and antenna calculations," PROC. I.R.E., August, (1931).

Carter: Circuit Relations in Radiating Systems

1009

In free space Maxwell's laws result in the "wave equation." Hence, at a distance r from the source, the electric and magnetic field intensity vectors E and H must be functions of (t - r/c), c being the velocity of light, in addition to satisfying the particular conditions at the source. (t - r/c) is commonly called the retarded time. It is the time at which a disturbance, reaching the distance r at the time t, originated at the source. The method of deriving the electric and magnetic intensities mentioned above can be generalized so as to hold for all frequencies if the retarded time is used in forming the potentials. Potentials formed in this manner are known as "retarded potentials."

Fig. 2

For wires an integral number of half waves long the process described gives us the formulas shown below. These give H and the perpendicular and parallel components of E in terms of the cylindrical coordinates as shown in Fig. 2. The power and quadrature terms are given by the real and imaginary parts of these expressions after expanding the exponentials. The derivation is shown in Appendix I.

E- j30I [
Ep
= -

45-mr,

e-mrl

] volts per cm.

Z -

(10)

j301 1 __
[

r j-mr2 (Z
e-m2

(-.1)n -_
n

z)e-j77
] Gauss
E j

volts per cm. (1 1)

(12)

Ho = + j

1[_(

1)n

el)(

mr
-

1010

Carter: Circuit Relations in Radiating Systems

where I is the current in amperes, m=2ir/X, r2 =p2+ (lz)2, rl= p2H, n = the number of half waves on the wire, and E-imr = COs mr-j sin mr. From these relations we may determine the self and mutual impedances for the several conditions of interest which we wish to investigate. From the definitions of impedance given we have as general relations for the self and mutual impedances of any linear radiators whose lengths are multiples of a half wavelength: -1 r' (13) Z1=-Ii Ein sin mz dz

Z= -

-1

arI

E12 sin mz dz

(14)

where E11 is the component of the electric intensity parallel to radiator No. 1 due to the current I, in No. 1 and E12 is the component of the electric intensity parallel to radiator No. 2 due to the current I, in No. 1. The conditions of interest and the corresponding impedance formulas are as follows. The derivation is shown in Appendix II.

(a) Self-Impedance of a Linear Radiator Z = 30 [,r + logE 2ml - Ci2ml + jSi2ml] ohms where, u cosx sin x 2r = dx, Si(u) =dx m =_ Ci x JO 00 ~ ~~x )t r = Euler's constant 0. 5772

(15)

Jr
Fig. 3

Tables of the sine integral and cosine integral functions Si (u) and Ci (u) are given in Steinmetz' "Transient Electric Phenomena and Oscillations" and in Jahnke and Emde "Funktionentafeln mit Formeln und Kurven. " (b) Parallel Wires of Equal Length (Not Staggered) (See Fig. 3.)

Carter: Circuit Relations in Radiating Systems

1011

Z21= + 30{2Ei(- jmp) - Ei[- jm( /p2 + 12 + 1)]

- Ei[- jm(Vp2 + 12 -1)] } ohms

(16)

where,

Ei(- jx) = Ci(x) -jSi(x).

(c) Parallel Wires in, Echelon (Fig. 4)

where,

Z21 = 15 cos mh [2 Cih -2 Ci -h + Cih--I + Ci - -(h - ) + Cih + 1 + Cih - (h + 1)] + 15 sin mh [2 Sih- 2 Si - h - Sih - I + Si - (h - 1) - Sih + 1 + Ci - (h + 1)] - jl5 cos mh[2Sih-+ 2Si - h - Sih -1- Si - (h -1) - Sih + l - Si - (h + 1)] + jl5 sin mh [2Cih- 2Ci - h(17) - Cih - I + Ci - (h - 1) - Cih + 1 + Ci - (h + 1)] ohms.

h= M (/p2 +Ah2 +Ah), -(h-i) = m[/p2A+(h-1)2 - (h-l)],etc.

(d) Colinear Wires (Fig. 5) This is a special case of (c) in which p =zero where the formula fails due to giving rise to the indeterminate form o - oo. However, by taking the limit as p approaches zero, Z12 -15 cos mh [- 2Ci(2mh) + Ci(2m(h - 1)) + Ci(2m(h + 1)) /h2 _12\-_ l + 15 sin mh 2Si(2mh) -Si(2m(h -1)) (h2

Fig. 4

-Si(2m(h
-Si(2m(h

+
+

1)) -] j15 cos mh [2Si(2mh) -Si(2m(h - 1))

1))]
+ ji5 sin mh

[2Ci(2mh) -Ci(2m(h -1)

)] ohms

-Ci(2m(h
when, h>l.

1)) -log, (

(18)

1012

12Carter: Circuit Relations in Radiating Systems

4,~~~~~~Fg

Filg.

5a

(e) Two Wires Forming a "V" (Fig. 6) In this case the electric intensity along the second wire, due to the current in the first, depends upon both the parallel and perpendicular components with reference to the first wire. This is given by:

Es
where,

= j30I

1 s

e -jmr2

--i

r2

(- 1)n volts per cm

21s

( 19)

r2

\/2

+ S2

cos

and s is the distance along the second wire as shown in the figure. The mutual impedance is:

Z21

30(-1<7

[,f

1sinmr2 sin

ds

O r2
I
cos

+ jT

Jo

--

Mr2 sin ms
s

(20)

---ds ohms.

The integrals in this equation cannot be expressed in the form of any known tabulated functions and therefore are best evaluated mehanically in each particular case.
e

E/

Fig. 6

Carter: Circuiit Relations in Radiating Systems

1013

PART I1-APPLICATIONS 1. FIELD CONDITIONS NEAR RADIATORS It is not possible to build a model which will accurately show the field conditions near a radiator. If two forces, both of which are sinu soidal functions of time and which differ both in direction and in time phase are combined, the terminus of the vector representing the resultant force will describe an ellipse during a cycle. This is the condition which generally exists in the vicinity of a radiator with respect to the field forces. In general, the parallel and perpendicular components of the electric field intensity differ in time phase. Such a condition makes it quite difficult to get a clear physical picture of the phenomena existing.
450

/0

/
_

0/E

Zv

Frtrr/M

/L)000 60

GD

'IO0

90

/00

Fig. 7

At positions of interest we might plot the ellipses of polarization but to connect these pictures properly is a laborious procedure. However, in the case of a single linear radiator we may take any line in space parallel to the radiator, and plot curves of the power and quadrature terms of both the parallel and perpendicular components of the electric force along this line. A collection of such pictures gives us all the information necessary in a study of radiation phenomena. In Fig. 7 are shown curves of the power and quadrature terms of the parallel component of the electric force along a half-wave dipole of No. 4 wire for a wavelength of 15 meters, while in Fig. 8 are shown curves of the quadrature terms in both the parallel and perpendicular components of the electric force. The only forces which can do any

1014

Carter: Circuit Relations in Radiating Systems

It will be noted that the curve of the reactive term in the parallel electric intensity rises to infinity at the wire ends. In practice, this cannot be true. This result is due to the assumption of a filamentary wire

work on the conductor current are the parallel components. The average work per cycle done by the quadrature term is, of course, zero. Since we have assumed the wire to have no ordinary resistance, the work done by the power term represents energy radiated.

_ /

,SD00.1-

3042

60I 0 9 f

/8s

_. V P w '0",O A.'- 7-= srze

t vo

3o s 2:< o7O

go 90 /00o

-_2#0
Fig. 8

having no end effects. In practice, the length of such a radiator is about 6 per cent less than a half wavelength, the insulator cap capacity being equivalent to the remaining length, necessary to complete the tuning. Thus in the practical case it is seen from the curves that the maximum electric force parallel to the wire is only about 0.28 per cent of that in the perpendicular direction. It may be of practical interest to note the value of maximum electric force to be expected with 50 kw in a half-wave dipole at a wave-

Carter: Circuit Relations in Radiating Systems

1015

length of, say, 15 meters. From the curve it is seen that the maximum electric force is 236 volts per centimeter for a current of 1 ampere. At 50 kw the current would be 50,000/73.2 26.2 amperes. This will
=

Fig. 9

30a

a~~ ovc~' /MR 41>9ERtE

\c -

eU

ON,c 'WI& e EWrz;s v4D4Tox or4 wlr-C /adTWom

Rr'P'73,VE

#rbF',9rn<qLLgL
I8O 6tEe5

/C

_1

.6

41
__

O4
zoo
r

in result a force of 6200 v~~~~~~olts p6/err= centiete or1/0

o>rcFig. 10cs

ot

enh

Fig. 10

result in a force of 6200 volts per centimeter or 15,700 volts per inch. In Figs. 9 and 10 are shown similar curves for a radiator one wave long. It will be noted that the power term of the parallel component of

101 6

11(arter: (Circuit Relations in Radiatinq Systems

the electric force is greatest towards the ends of the wire. For wires several wavelengths long the resulting curves would be very similar to these for the one-wave wire and therefore have not been shown.
2. HALF-WAVE RADIATORS F'rom (16) we obtain for the self-impedance of a half-wave radiator:

Zi, = 73.2 + j42.5 = 84.5 /--- 30. 150 ohms. Thus, at the current antinode the radiator is the equivalent of a resistance of 73.2 ohms (the radiation resistance) and an inductive re-

Wg9k Fi

Fig. I11 actance of 42.5 ohms. This result indicates the necessity of decreasing the length a little in order to obtain a nonreactive load. The curves in Figs. 11 and 12 show the manner in which the mutual impedance and its phase angle vary with the spacing of two parallel half-wave dipoles when the ends remain upon a common perpendicular line. In the past there has been considerable controversy as to the best spacing for a free reflecting half-wave dipole from its antenna. By mak-

Carter: Circuit Relations in Radiating Systems

1017

ing use of the curves of mutual impedance we may answer this question. For zero radiation directly backwards the currents must be of equal amplitude and in quarter-phase relation and the spacing exactly equal to an odd multiple of a quarter wavelength. It is evident that such a condition is impossible to obtain when a free reflector is used. For best addition of the wave amplitudes in the forward direction the phase angle between the currents should be equal to 2rS/X where S is the spacing. In Fig. 13 are shown curves of current and of actual and

N~~~~~~~~~ I
/14'u,q

/t~ ThWC1JJ43 ST'R

IN

/~ ~

----- --1XX -t
, r ..r

f / /}

1\;
v,9&-e

,;

ia

ii ,/

J7>9C/'Vd IM

L'6P'.5S

Fig. 12

desired phase angles as a function of the spacing. It is seen that the phase angle curves cross at about 0.29-wavelength spacing but that the current is rapidly decreasing. At a quarter-wave spacing the phase angle is incorrect by about 28 degrees. This may be made zero by a slight change in the reflector tuning with little effect upon the current amplitude. Consequently, to the writer, there appears to be little basis to the arguments for the use of other spacings. With R.C.A. Communications it has not been the practice to use free reflecting systems. When a reflector is fed by a transmission line it is no more a reflector in the strict sense of the word and the problem is quite different. This condition will be discussed later.

1018

Carter: Circuit Relations in Radiating Systems

Consider the question "What is the best spacing for two parallel half-wave dipoles?" If the currents are held equal and constant as the spacing is varied, the intensity of radiation perpendicular to the array will remain constant, but, due to the variation in the mutual impedance, the total power will vary. It is apparent that at zero spacing the directivity is the same as for one radiator alone. At a very great spacing the mutual impedance is negligible and the power just twice that for one wire. At a great distance perpendicular to the array the electric

E4

"4X
/
1

94,

-T .

I.

59C/ /VG , I .5_P.

r zV L/HS I

Fig. 13

intensity is doubled, thus quadrupling the intensity of radiation. The directivity is then exactly twice that for a single wire. For a finite spacing the power in either wire is: W = I,2 Ril + II2 R12 where R11 = Z11 cos oil and R12 = Z12 cos 412. The total power is 2W. For one ampere in each dipole the total power is then 2(R1l+Rl2). The directivity is proportional to 1/W. Fig. 14 shows the relative gain as the spacing is varied. The same reasoning may be extended to include any number of radiators. Fig. 15 is a similar curve for three radiators.

Carter: Circuit Relations in Radiating Systems

1019

In treating radiation phenomena the assumption of a perfectly conducting ground is usually far from justified. A curve of radiation

l
._r6-\

1a L/n \-

1--<
_~~~~~~~~~~~~~V9 e/ _ '_

8--

4-~~~Fg

14

l/.

42.,

5P4C/1A'

.3

.7 .8 .6 . Vr1 J/V WIW'92 L2jV

.9

.0

//

/.

Fig. 15

resistance for a horizontal dipole above an ideal ground is shown in Fig. 16. This should, in this case, give a fair idea in a qualitative way,

1020

2(arter: Circuit Relations in

Radiating Systems

I-

Fig. 16

.w6ztY.'#P/K.7RA'V 72fw2,tGgEE5

/1/?LE/.9K4 H6.'/ZO/V7'?L 2./PdLg ?LZA'r/VE D/,97-/a/V4'T/Tta/6r/ //T..9&.VA? 626(/'V.L'

_ 22 Ku
/15/GA?

V. Y

(PL.rEcr v.'au/vD /?SSUtIED)

- -

RAOV5

5FOJFAg 17
D

5L57'/GrHS

Fig. 17

Carter: Circuit Relations in Radiating Systems

1021

wavelengths. The most common simple type of radiator is a half-wave dipole connected to a transmission line in the manner shown in Fig. 18. The

of the variation with height over an actual ground. In practice, if we add a few feet to the actual height, the results will approach those for the ideal case even though such a procedure has little theoretical justification. In Fig. 17 are shown curves of relative radiation per unit power input at elevation angles of 5 and 10 degrees for a horizontal dipole as the height is changed. Experience indicates the radiation most effective at a distant receiver to be that which leaves the transmitting antenna at angles in the vicinity of 10 degrees. If such is the case, a horizontal dipole should be placed at a height of between one and two

Fig. 18

length of the dipole and the spacing of the connections are then adjusted so that the characteristic impedance of the transmission line is matched. Such an arrangement is approximately equivalent to the sketch of Fig. 19 as the larger portion of the radiated power is contributed by the outer ends of the dipole. The effective characteristic impedance of the dipole is assumed equal to that of its feed line. The equivalent resistance load is then made equal to the ratio m of the char-

Fig. 19

acteristic impedance squared to the radiation resistance. It can be shown from transmission line theory that, in order to match the equivalent resistance of such a system to the characteristic impedance of the feed line, the total length of wire in the short-circuited line branch must be equal to X/2r tan-' (/m/(m- 1)) meters.

10)22

Carter: Circuit Relations in Radiating Systems

Fig. 20 shows the manner in which the distance between feed-line terminals varies with height for a horizontal dipole over perfectly conducting ground. Under actual conditions the variation follows a quite

Fig. 20

Fig. 21

Carter: Circuit Relations in Radiating Systems

1023

similar curve. The total theoretical length also varies with height in a similar manner, but in practice the actual length is less than the theoretical by an amount depending upon the type of insulator and size wire and must be determined by trial for each condition. In order to make the most economical use of supporting structures, it is often desirable in practice to arrange several horizontal dipoles end to end in the same span. Some of these are often operated at frequencies differing by less than one per cent. This raises the question as to what minimum spacing should be allowed when two dipoles in such an arrangement are operated with small frequency differences. The curve in Fig. 21 shows the variation in mutual impedance with spacing for two colinear dipoles. To obtain the per cent coupling the values given must be divided by 84.5. There are many more conditions in connection with arrays of halfwave dipoles which might be investigated by means of the methods under discussion but, due to the recent developments in directional antennas which we have made, arrays of half-wave dipoles have become of minor importance in R.C.A. Communications.

__

7-7r o>/Hf1V F g.S-y A q

r t7>iX-

_ -__varHs _ x~~~zmr -__ _wx

zx-m

Fig. 22

3. LONG PARALLEL WIRES (NOT STAGGERED) From (15) we may compute the self-impedance for a wire having a length equal to any integral number of half wavelengths. Fig. 22 is a curve of impedance versus length. The values shown are correct only for multiple half wavelengths. The power term of the impedance is identical with the value of radiation resistance as shown in the I.R.E. paper already referred to, when computed by the Poynting's vector method. Curves of the power term in the mutual impedance for wires

1024

Carter: Circuit Relations in Radiating Systems

4, 8, and 16 waves long, plotted as a function of the spacing, are shown in Fig. 23. By the same method as was used in connection with halfwave dipoles we may determine the relative directivity for two or more long wires in broadside as the spacing is varied. In Fig. 24 are shown

79~Ex7

uZgz

Sr0nSGGEtLDv

.1n

t~~~~~~~~ X44X4444X
1C/SA

Fig. 23

Fig. 24

curves for a pair of wires 4, 8, and 16 waves long. The 8-wave curve is of particular interest in connection with the broadsiding of two sections of the R.C.A. Type B antenna;, often spoken of as the vertical harmonic wire antenna. It will be noted that the optimum spacing is

Carter: Circuit Relations in Radiating Systems

1025

11 wavelengths. This also was the result obtained experimentally by Lindenblad for two bays each without a reflector. Analysis by the Poynting vector method, subject to more or less error because of the large number of curves plotted, resulted in an optimum value of 2 wavelengths for 2 bays complete with reflectors. It is apparent that the spacing should be a little greater for a system with reflectors than without. Therefore, we can state with certainty that the spacing should lie within the range of 14 to 2 wavelengths.
4. LONG PARALLEL WIRES IN ECHELON From (17) we may calculate the transfer impedance between two wires of any length in echelon. As this expression is rather tedious to evaluate, we shall here treat only one case, the calculation of the theoretical directivity for the Model B harmonic wire antenna, as a check upon the conclusions from the Poynting vector method of analysis. In a commercial model the dimensions are: length = 8 wavelengths, spacing (p) = 0.416X, stagger (or echelon distance) = 0.1311X. Since the current in each succeeding wire leads that in the preceding by 90 degrees, the total impressed voltage in No. 1 wire for a current of 1 ampere in each wire must be: Vl = Zll + jZ12 + j2Z13 + j3Z14 = Z11 + jZ12 - Z13- jZ4. Since Zab=Rab+jXa b, the power term of V1, or the power for one ampere, in No. 1 wire becomes: W1 = R, - X12- R13 + X14. In a similar manner we obtain for the power in the other three wires: W2 = R22 + X21 - X23 - R24 W3 = R33- R31 + X327- X3 W4 = R44 + X43 - R42-X41. Each of the R's in the above relation are equal to one of the four values, R11, R12, R13, R14. A similar relation is also true for theX's. Upon substitution of these equivalent values and addition of the W's, we obtain for the total power: W=4R11-4R13. Having previously calculated R1, we have remaining only R13 to determine. From (17) we get R13=-11.48 ohms. Therefore, W=4X167.1=668.4 watts per ampere. For one ampere in each wire the radiation in the direction of maximum is increased 16 times over that for one ampere in one wire alone. The power for one wire alone is 155.6 watts per ampere. Therefore, the

1026

Carter: Circuit Relations in Radiating Systems

gain ratio is 16X155.6/668.4=3.71. The gain ratio for one wire to a half-wave dipole is 4.42. Hence, the ratio for the one-bay harmonic wire antenna to the half-wave dipole is 4.42 X3.71 = 16.4. The above result is nearly the same as that obtained by the Poynting vector method as well as by actual tests.
5. V-WIRE RADIATORS

In the Model D antenna the radiating unit is a wire in the shape of a V. Several lengths have been used but we shall first consider a unit having sides 8 waves in length. The self-impedance of one side of a V, as obtained from (20), is 155.6+j 46.9 ohms. For the mutual impedance we get from (20), after changing the variable from S to ms = 163r sin mr2 sin ms Z12 = 30 -ml d(ms) Jo0 mr2 ms + 16r cos mr2 sin mds(22) Ohms ms mr2 where, mr2 = (167r)2 + (ms)2- 32rms cos 0.

+j

d(ms)

By graphical evaluation of the above expression we obtain Z12= 37.2-j 48.3 ohms. Since the current in the second wire is in phase opposition to that in the first, the total effective impedance of one side of the V is 155.6
.002
o6

Y 2,02
tt1 .0084 - ~s,v6a3zG f 6 zt-i d .o ___

2 I>

2 Y o," ssrtQcr0

w;vr
-9A
7

ov

k.os

tAS~~~~~V9wfA

A'c V0

+j 46.9 - (37.2/-j 48.3) =108.4+j 95.2 ohms. In the determination of the mutual impedance, curves of the integrands were plotted to a large scale. In addition to knowing the total efnect of one wire of a V upon the other it is of value to know just how this effect is distributed

Carter: Circuit Relations in Radiating Systems

1027

o/ _ _ _

7*LV w rM65 /SRW2r.AV.D//wCED VOL

A ij<
/12L?

i_

.001
.0oG __ _ --|

____I

__

~ ~ ~~.ao2~
2 2 , 2 2 Z # 42|

'3)

A40
_~

/.

OA|

82_

2 /2

Fig. 26

.002

t
:1

000 -0-4

WSS 2 \L.\ 71 a
tl6A VOU

r.-

.ao

-0.000.54.5as^z 8DCE -

V N

-~~o04

2r6D.4Fig.02
1

-.0/64et1

36Vea

Fig. 27

/A'CZ2/.V2
7_ 0

,9//GE -35 VD0"E05

W_ l_

___ll

_r 1

-y'07

-- c ||f

L L &aL0V07y26a) 1 LA mrrnoXtlr 1 t~ ~~Z(r9'25Av.(io-k

Fig. 28

1028

Carter: Circuit Relations in Radiating Systems

along the wire. For this reason, rough copies, drawn to a small scale, of these curves are shown in Figs. 25 and 26. The effect of decrease in the included angle of the V to 25 degrees is shown in Fig. 27. In certain cases we wish to know not only the total power radiated from a V unit but also the contribution of each element of the wire to this total power. The total power radiated from an element is deter-

Fig. 29

mined by the sum of the self-induced and mutual electric forces. A curve showing the relative power distribution along an 8-wave, 35-degree V is shown in Fig. 28. Although each element of the wire takes part in determining the directional distribution of radiation, its contribution to the power may, in some cases, even be negative. It will be noted from the curve that the larger portion of the power is contributed by the elements toward the diverging ends of the V. It was shown in the I.R.E. paper previously referred to that, if the current is held constant, maximum radiation along the bisector of ain

Carter: Circuit Relations in Radiation Systems

1029

8-wave V takes place when the included angle is 35 degrees. However, it does not necessarily follow that this angle results in the maximum directivity as the directivity depends upon the ratio of the intensity of radiation along the bisector to the total power radiated. In general, constant currents do not represent a constant total power when the conditions in a radiating system are changed. An investigation of the

4z

7_ 00' C('y''0y76'%r=t

777 007 f f'f-CA10WCtG7 _

007700L/

__ _ _ 7L o'rqwr -0 -1
7

D/A'L-

,___

7---

__ _

___

_-

I00

TO

7;z

<

5GEE

Fig. 30

effect of angle upon directivity was therefore made in connection with V wires of several different lengths. The total power radiated is, in each case, IPR, where R is the power term in the total impedance of the V. The power per unit solid angle radiated along the bisector can be obtained by the method shown in the same I.R.E. paper referred to. Curves of radiation in the optimum direction, total power radiated, and directivity, plotted as a function of the included angle of an 8-wave V, are shown in Fig. 29. The unit of solid angle has been taken here as a full sphere rather than a solid

1030

Carter: Circuit Relations in Radiating Systemss

J.-fK
&O0

-7
K

~~8p 3.0

~ ao
2509

4v0e

23oe-__ ;f
3O
? -434,0v2 38 IA'

/LA94A

Fig.E 31L2G'

Fig. 31

Zr-i-r

4 v or

Vt

. sows

a
ZE

iO

/iv

Nr15.

Fig. 32

Carter: Circuit Relations in Radiating Systems

1031

radian. The directivity then becomes the ratio of the watts-per-unit solid angle in the optimum direction to the total power radiated. In Figs. 30 and 31 are shown similar curves for one-half and onewave V's. It will be noted that the optimum directivity for a one-wave V is obtained when the angle is 90 degrees rather than 105 degrees as determined from the field amplitude patterns alone. In Fig. 32 is shown a curve of directivity vs. length of wire where the best angle is used in each case. The included angle of a V wire may be made obtuse such that maximum radiation takes place in a direction perpendicular to the bisector as shown in Fig. 33.

Fig. 33

It is of interest to find the gain ratio for such a system and compare it with an acute angle V having maximum radiation along its bisector. This has been done for a V having an angle of 140 degrees and sides equal to 8 wavelengths. In this case the mutual impedance is such as to increase the power for a given current in each leg over the value when the wires are at a great distance. This results in a gain ratio of

Fig. 34

considerably less than 2 over that of a single wire. The actual value obtained was 1.68. Therefore, this combination is quite inferior to a system where maximum radiation is along the bisector. If we form a diamond system (Fig. 34), we have a combination of the effects just considered. Maximum radiation takes place along one diagonal. The gain ratio is less than twice that for a single acute angle

1032

Carter: Circuit Relations in Radiating Sy'stems

V but greater than twice that for a single obtuse angle V. The usual broadside arrangement used in the Model D antenna is thus seen to be superior to the diamond. A rough copy of the curve of the mutual voltage distribution for a V wire having an included angle of 140 degrees is shown in Fig. 35.

.42

//V rTHE T

*D/E5 Burlolv or 1okVze 7oE1-7 /VS/ZH OINDUCED = 8 A__ i'OZ TEftE. _ Z Z'IV6

w/EAr

7a

-_4 A

X_
Fig. 35

the methods under discussion. First it can be shown that so long as the currents in the antenna and reflector systems are equal and in quarter-phase relation, the total power input is exactly twice that for one system alone when remote from any other, regardless of the spacing. This statement should not be taken to mean that the portions of the power contributed by the antenna and reflector are equal. As a matter of fact, with ordinary spacings the larger portion of the power is always contributed by the antenna. When the spacing is an odd quarter wavelength the electric intensity at some remote point along the continuation of the spacing line between the two systems, due to both systems, is either twice that

6. ANTENNA SYSTEMS WITH FED REFLECTORS In all the directional antenna systems in use in R.C.A. Communications, it has been the practice to feed the energy directly to the reflector by means of a transmission line, and several different spacings have been used between the antenna and reflector units. Several times the question has arisen as to what is the best spacing. If two like radiating systems are spaced in the desired direction of radiation by a distance equal to any odd number of quarter waves and the currents in the two units are equal and in quarter-phase relation, the directivity will be exactly twice that for one systein acting alone, provided the distribution of currents within either system remains unchanged by proximity of the other. This statement can be proved by

Carter: Circutit Relations in Radiating Systems

1033

due to one alone, or zero. Calling the electric intensity due to one system alone E, that due to both is 2E and the corresponding intensity of the radiation 4E2. Since the power input is twice that of one wire alone, the directivity ratio is 2 and the statement is proved. The same conclusion- is roughly obtained in any particular case by the integration of Poynting's vector. It is therefore clear that, in so far as directivity alone is concerned, the number of odd quarter wavelengths used for the spacing of an energized reflector is immaterial. However, small spacings result in a

.0. -

__________________I *41ELrM6r// 202f11ZOP70.1S./

siE D

17 /0 20 J5 Z5- 30 1 fE,V/G*/ 017L26P/N 1E717E7

=__

_T

35

4,0

Fig. 36

much greater difficulty of adjustment due to the higher values of mutual impedance obtained. In some instances radiation at some undesirable angle may be appreciable and by properly spacing the reflector this may be made a minimum. In practice, where the antenna and reflector each may consist of a rather complex arrangement of radiators, the mutual impedance may be determined from experimental curves of intensity of radiation vs. tuning of the reflector when free. In making such a test the coupling to the transmitter is kept low so that the impressed voltage on the antenna can be considered constant. An example of such curves, taken in both forward and backward directions, is shown in Fig. 36. These were taken during the process of adjustment of a V antenna system in which both antenna and reflector units each consisted of 2 V wires one above the other.

1034

Carter: Circuit Relations in Radiating Systems

APPENDIX I I. FIELD CONDITIONS IN THE VICINITY OF A RADIATOR In Fig. 37 let AB be a radiator whose length 1 is an integral number n of half waves. Call the distance along the wire zo and let the cylindrical coordinates of the point P, at which we wish to determine the field strengths, be: (p,6,z)
\3

Fig. 37

If the current at an antinodal point is represented by the real part of IEiwt, the current at any point zo along the wire is then given by the relation: i = Ieict sin mzo, (23) where m = 2rX =co/c, c being the velocity of light. Because of the law of continuity, the time rate of decrease of charge density, o-, along the wire must equal the space rate of change of current, or: _r _-,(24) bt 5zo where o- and i are both in electrostatic units, from which I (25) ejUt cos mzo. of _
c

The electric and magnetic field forces may be determined from the retarded scalar and vector potentials f and A where:

+ =j [ jdzo,
r

(26)

Carter: Circuit Relations in Radiating Systems

1035

A =c

-dzo,
r

I]

(27)

in which [a] and [i] indicate the charge and current taken at the retarded time (t - r/c), and K is unit vector in the z direction. Upon substituting the values of "retarded" charge and current in these relations and changing the sine and cosine functions to exponential form, the potentials become:

2c ~ki~Ei
A
=

1-~~m (rz,) r -oL f[

e i'o

e- im(r Z0)

dzo

(28)
(

X-i

AliKyEiwt J[

jm(r+Zo) i -

-jm (r-zo)

dzorr

(29)

where,
r =

V/2P + (zo

z)2.

The above integrals may be expressed in terms of the sine integral and cosine integral functions but this serves no useful purpose as the potentials are of no value to us except as a means of deriving the field forces. To form the components of E and H we have the relations: 1 dA E = - grad A(30) c dt H = curl A. (31) Since A has the direction of the wire axis A z= A and A p = As = zero.

Grad,,/ = -, gradp J =-, and grado VI sz

bp Curle A
= -

= 0

6A

Sp

, curlp A = 0, curl, A = 0 -

get

6A ,and Hp = Hz = 0. bp However, from Faraday's law, dH/dt =-c curl E, from which we

Hence, Ho = H

~ bEp bEzN E = ~~~ (32) =1curlo m m Lz bpJ so that, once we have derived Ez and Ep we may just as well obtain H from the electric force components as from the vector potential.
H

1036

Carter: Circuit Relations in Radiating ISystems

From (30) we have for the parallel component (E,) of the electric force: 1 dA E,- grad, 4--- c dt n-jm(r-zo) I ul3r - E rjm(r+zo)
=

j 2az
IE iw
f

r
E- jm(r-zO)

dz0

c+ l

2c220

[-E jm(r+zo)

~~r

-r

dz

(33)

From the diagram it is apparent that ar/lz = -br/lzo. If we make use of this relation and integrate by parts, after some manipulation we get:

E,

+i

EI= rJ~Li~[(El

mVP-i\P2+ (l_z)2 e-imVp2z'] (E jml + E- jml)P2 ( 2(34)

but (Eiml+e-iml)/2 cos ml= (1)n where n is the number of half waves on the radiator. If we let ri= Vp2+Z2 and r2= Vp2+ (l-z)2 -Ejmr2 E-1 I 1mr_ ) Ez j E (35) 1(
c

electrostatic cgs units. If I is the r-m-s current in amperes this becomes:

r2

Ez = j30I

e-jmr2

,_ r2

(- 1) n-

Ejmrl

rI

volts per cm (r-m-s.) (36)

=0 To obtain Ep we have, since A,=

Ep= -gradp l

= -

bp
E- jnt(r+zo)

(37)
r
E- jni -

Ep-j2c Eiwt J

z r1z

op
r2

r
r

(r-zO)1

j
Z

dzo.

(38)

After considerable manipulation this becomes:

Ep= -i
Eo = -j301
.
E

jiX L

-jmr2 Z-l

_
z

(_- 1)n

p-jmrr

rl
z

p _

esu.

(39)

or in practical units where I is the r-m-s current in amperes:.

jmr2
r2

\ p

l\

(- 1) -

E- mr1

ri

volts per cm.

(40)

Carter: Circuit Relations in Radiating Systems

1037

To determine H we may use the relation:

H
i
m

curle E

~~~j SEp
m Lbz

bEz bp j

(41)

After performing the operations indicated and combining terms we get, for the current in amperes (r-m-s):
H
=

30

IF- Eimr2

-(-

)n

E-

mr,

Gauss (r-m-s).

(42)

These relations give us a complete knowledge of the field conditions at any point in space.
APPENDIX II 1. SELF-IMPEDANCE OF A LINEAR RADIATOR The self-impedance has been defined by the relation: 1 rt Z=E, sin mz dz.

(43)

Substituting the value of E, from (10) into this expression and noting that, in this case, r =z and r2=1- z we obtain:

Zl=

-j30

r jm(Uz)
I

)n

e jmz

sin mz dz.

(44)

Since ml = nr, i-jmll=(-1) and this expression, after replacing sin z by its exponential equivalent, becomes: ZJ=
=
I

>-i2mz C_

i-2 mz-

2z

1)dz
(45

-30

r2mZ e-ju

or,

u Jou

du

[Tr + loge 2ml - Ci2ml] + j [Si2ml] } ohms, where X = 0.5772 . . . (Euler's constant).

Zli

= 30 {

(46)

2. TRANSFER IMPEDANCE FOR PARALLEL WIRES (NOT STAGGERED)

The transfer impedance was defined as

Z21=

-1 r1 I E21 sinmzdz.

(47)

1038

Carter: Circuit Relations in Radiating Systems

Substituting the value of E, from (10) into this expression it becomes, after replacing sin mz by its exponential form:

Z21= -15

[m( - 1)n
-Ejm (r1-z) _
jm (r1+z)

zm(r2-Z)

I-jm(r2+z)

ri

dz.

(48)

Changing the variable in the integrals by substitutions of which U = m(r2-z+l) is typical, we get:
Z21= + 15 ( 1)n
-J

mp

m(fAi
mp

_
p2+ 12+I1

f-j(u-ml)
-

du

-+(- 1)"
-

-- dufn(V/P2+l2 1) r m(\/p2+12+1)I-ju
mp
u

f-j(U+m0

m(,\/ p -+I-1)
p

-du u (49)d

ej

du.

Since ml= n7r, Z21 + 30 [2Ei(- jmp) - Ei

and + iml=( 1)n -jm(\/p2 + 12 + 1)1 - Ei {- jm( \p2 + 12-1)}]. (50)

3. MUTUAL IMPEDANCE FOR PARALLEL WIRES IN ECHELON By reference to Fig. 4 it is seen that if we consider Z as the distance along wire No. 1, the distance along the second wire becomes (z - h). Taking this into account, the expression for the mutual impedance becomes:

Z21

j30

*l+h-

jmr2

r2

(-1) n
jm(r2-Z)

jmr,l

ri

sin m(z - h)dz.

(51)

Changing the sine term to exponential form this becomes:

Z12 -

+ l56+ jmh

iE nhL(- 1)n J h _
Jh

l+h

r2

dzh

1 -h

E-jm(rl-z)

1++h h-jm(r1+z)

~~~~dz Jl

rl h _~~~'r

adz

(1+h E-jm(r -z)

ri

-dz.

(52)

By changing the variables in a manner similar to that in the preceding development and remembering that Eim = E-iml= (-1), we get:

Carter: Circuit Relattons in Radiating Systems

1039

-imh{-2Eil- jm( Vp2 + h2- h)I Z21- -15 + Ei[-jm(,\/p2 + (h - 1)2 - (h - ))] + Ei[-jm(s,/p2 + (h + 1)2 - (h +-l))]}
-

+ Ei[- jm( V/p2 + (h - 1)2 + (h - 1))] + Ei[- jm(Vp2 + (h + 1)2 + (h + 1))]} ohms. Expanding this expression into its power and quadrature terms it becomes: Z21= -15 cos mh[- 2Cih - 2Ci - h + Cih -1 + Ci -(h -1) + Cih + i + Ci - (h + 1)] + 15sinmh[2Sih - 2Si-h -Sih-I + Si-(h-l)-Sih + I + Si-(h + 1) (54) -jl5 cos mh[2Sih + 2Si -h - Sih - 1 - Si - (h - 1) -Sih + - Si - (h + 1) + jl5 sin mh[2Cih - 2Ci-h -Cih - 1 + Ci - (h - 1) - Cih + + Ci - (h + 1)] ohms where, h = m(/p2 + h2 + h), - (h-i) = m(V/p2 + (h-1)2 - (h - 1), etc.
4. MUTUAL IMPEDANCE FOR COLINEAR WIRES This is a special case of wires in echelon where the spacing (p) is zero. However, if zero is substituted for p in (54) we get the indeterminate form oo - oo. This difficulty is overcome by taking the limit of the expression as p approaches zero in the following manner: For small values of p we have by the binomial law:

15e+imhI - 2Ei [- jm(V/p2 + h2 + h)]

/p2 + h2-_ h
,p 2 + (h -1)2 - (h -1)
~~1
p2

=--

p2

2 h

(55)

h 1'(56) 2 h -i
1

Vp2 +(h + 1)2- (h+1) =2

For small values of x,

p2

(57)
(58)
(59)

and,

Ci(x) = r + log, x

p2 h 2 _128 MP 2 MP2 MP 1 log( Limit 2Ci. -C -Ci p- *o _ 2h 2(h- 1) 2(h + \) f h2

1040

Carter: Circuit Relations in Radiating Systems

and,

Z21= -15 cos mh[ - 2Ci2mh + Ci2m(h - 1) + Ci2m(h + 1)

-log h2 12 + 15 sin mh [2Si2mh - Si2m(h - 1) -Si2m(h + 1) - j15 cos mh [2Si2mh - Si2m(h-1) -Si2m(h + 1)] + j15 sin mh [2Ci2mh - Ci2m(h -1)

(60)

-Ci2m(h + 1) - l1g0 (
when,

)] ohms

h>l.

and having an included angle 0 between them. The distance along (2)

5. MUTUAL IMPEDANCE FOR V WIRES In Fig. 38 are two wires of equal length n (X/2) ,where n is an integer,

z"~~ ~

f~ ~

Fig. 38

is measured from the apex as origin. In order to determine the voltage induced in wire (2) by the current in wire (1) it is first necessary to determine the component of the electric force E in the direction of wire (2). The component of E, in the direction along s in E, cos 0. The component of Ep in the direction along s is E, sin 0. Therefore, E= E, cos 0 +E, sin 0. But z =s cos 0 and p = s sin 0.

Carter: Circuit Relations in Radiating Systems

1041

Substituting these values of Ez, E,, z, and p in (36) we obtain:

E= j3O -j30
ejE

Mr2

r2 E3rnr2

(- 1)ns Cos 0
-

i-ms

cosG
5-)m8
s coso

2
n

-(-1)_sinG ssin 0 s

s sinj

(61

(61)

= j30(_ 1)

1 -jmV2 IE i r
r2

volts/cm I where r2 = VYS2-F 2ts cos 6 (62)

Substituting this value of Es in the general formula for mutual impedance, and separating the expression into its power and quadrature terms we get:

Z21

301(- 1)n

~lsin mr2 sin ms

r2
s

ds

JO r2

r cos mr2 sin ms c s ds] ohms. (63)

s

It is not possible to express these integrals in any form convenient for calculation. As r2 is a function of 1, s, and 0 they must be evaluated mechanically for each specific condition.