Sei sulla pagina 1di 12

132

contributions

Analysis of Helical Transmission Lines by Means


of the Complete Circuit Equations*
1,ERNON J. FOWLERT
Summary--A set of integro-differentialequations,called the _\lost of the preliminary definitions and analysis lead-
Icomplete circuit equations, are derived from Maxwells equations ing to the complete circuit equations are given in the
and applied to the solution of the parallel-wire transmission line, the
literatureI.2 and will not be repeated here. For homo-
double-helix transmission line, and the singlehelix, or helical wave-
guide. These equations take into account the effects of inductance geneousisotropicmedia,themagnetic flux density B
and capacitance distribution, retardation, and outward radiation. A and the electric intensityE can be expressed in terms of
generalization of earlierconcepts of distributedinductanceand a vector magnetic potential A and a scalar electric po-
elastmce (or inverse capacitance) is manifest in the solution of the tential, as follows:
helical line, wherethese quantities become functions of the phase
efficient or wavelength of propagation and are Fourier transforms of R=VX=I
certainclosed-formdistributionfunctions. In general,phase ve-
locity is a complicated implicit function of frequency, but under a
hypothesis of mode segregation on the basis of wavelength the
phase velocityand frequency can be obtained parametrically in terms
of a third variable, called the phase parameter. Usingthis hypothesis, The divergence of A can be specified arbitrarily, thereby
plots of phase velocityand characteristic impedance versus frequency completing the definitionsof A a n d unique quanti-
were obtained for the double helix and the helical waveguide.
ties. By letting
DERIVATIOX
OF THE COMPLETE CIRCUIT EQUA-
TIONS FROM hl AXWELLS EQUATIOKS
(3)

c IRCUITAL relationships are generally expressed

defined onl). for slowly varying fields. However,


these quantities have alternative expressions which are
almost as universally applicable as the electric and mag-
the two wave equations Ain alone andin alone are ob-
in terms of currents and voltages which are well tained,

netic field intensities themselves: namely, the current


density and the scalar electric potential T h e cir-
cuitalrelationshipsthataretobeconsideredinthis
paper find their most exact expression in terms of these These ran be integrated directIy by Kirchoffs methodB3
two quantities.
S. Ramo andJ. R. Whinnery, Fieldsand \tTaves in Modern Ra-
Revised manuscript received by the PG;\P, May 17, re- dio, John 1Viley and Sons, New York, Y . ;19.18.
ceived by the IRE, July 1953. Sponsored by -Air Force Cambridge J. -A. Stratton, Electromagnetic Theory, McGraw-Hill Book
Res. Center, Contract -4F 19(604)-524. Company, Inc., New York, K. y.;1941.
t Res. Asst., Electron Tube Lab., L-niv. of Illinois, I-rbana, Ill. 3 Stratton, op. cit., chap. 8, pp. 424431.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
1954 Fowler: ,4nalysis of Helical T.ransmission Lines by Means of the Complete Ciwu.it Equations 133

complete circuit equations. Like the ordinary trans-


mission line equations, these equations are circzrital re-
lationships, in that they can be solved for the current
and potential at points in the conducting wire without
the need for first determining expressions for the fields
externaltothe wire. Eqs. (10) and (12) are complete
in the sense thattheytakeintoaccountbothwave
retardation (radiation) and inductance and capacitance
where distribution.Therefore, a particularsolution of these
equations which fits the given boundary conditions is
an accurate description of actual propagation along a
system of conductors, especially if those conductors are
and f2 denotes a closed domain, bounded b y a regular
extremelythin.For a systemconsistingentirely of
surface, containing all sources which contribute t o t h e
spaced thin wires, the correspondenceis extremely close
field.
at all wavelengths which are large compared with the
In the practical application of these potentials, the
largest wire diameter, regardlessof the physical proper-
condition of homogeneity is commonly violated by the
ties of the wire material. Thus, in most practical cases,
presence in the medium of conductors in and on which
where the wire diameteris much smaller than 3 mm, so-
flow the current However, since the actual current
lutions of and (12) are applicable up to a t least
distribution in the conductors is not sought for its own
100,000 megacycles.Sincethepermeabilityandper-
sake, i t suffices t o regard L as an equivalent current dis-
mittivity of actual nonferromagnetic metal wires closely
tribution in themediumdisplacedbytheconductors
approximate the free-space values, the validity of the
which matches the boundary conditions at the surfaces
complete circuit equations should extend even to much
of the conductors. The current-carrying conductors are
higher frequencies.
thussupplantedbyphantomcurrents flowing in the
mediumitself.Withinactualconductorsthecurrent
densityisrelatedtotheelectricintensitybyOhms
Law, but for nonferromagnetic metals the permeability
and permittivity are approximately equal to those of
free space.* Therefore, in all practical cases i t is a n ex-
tremely good approximation to identify L with the ac-
tual current distribution in the conductors.
Assume then that the conductors are indistinguisha-
ble from the dielectric medium, except that within them
Ohms Law applies.
I u,E
where is the conductivity of the conductors. Then
from (2) and (6) the first relationship between and is
obtained.

The second relationship is obtained by eliminatingp be- Fig. I-Definitions of retardation distance and angle
tween the time derivativeof (7) and the equationof con-
tinuity of charge, If the wire diameter is small compared with the free-
space wavelength and if the distance between any two
turns of wire is large compared with the wire diameter,
then i t is also a very good approximation to represent
The result is the current flow by means of filamentary currents flow-
ing along the axes of the wires, and to assume that
planes normal to the axis of each wire c u t t h a t wire in
equipotential circles. The only component of V 4 which
survives is then the one parallel to the wire axis, namely
For points in the medium displaced by the conduc- andthedivergence of thecurrentissimply
tors, (10) and (12) comprise a determinant set of simul- Therefore, the volume integrals in the complete
taneous equations in and which will be called the circuit equations can both be replaced with scalar line
Stratton, op. cit., chap. 5 , p. 326. integralsalongthecurrentpath,asshown in Fig. 1.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
134 October

The potential is evaluated a t a point on the circumfer-


ence of the wire a t position P,a distance along thewire
from an arbitrary origin The current element at
Q, a distance along the wire from P , or from the
origin, is idx, andthe correspondingretardationdis-
tance is convenientlJ- taken to be
PQ2)1/P (13)

where a is the wire radius. Thischoice leads to the prop- where and x are measured along the axis of the line
er distance r a in the limit asx approaches zero. If the rather than along the axis of the wire itself, and K1 and
radius of the wire were ignored, the integrands would K 2 are the lengthsof wires 1 and 2 per unit axial length.
approachinfinity x approaches zero, and the inte- Similar equations can be written for2 line by interchang-
grals would diverge. ing the subscripts 1 and 2.
T o obtain the component of the vector integral in If line 1 consists of twoparallelwires of radius a
(10) in the direction of the gradient of the scalar po- spaced a distance d apart, the parameters are follows:
as
tential a t P,the integrand is multiplied by cos where
is the angle formed by intersecting lines and OB, K1 K? 1, $11 $12 $21 922 0 Dz
parallel to thewire axis a t P a n d a tQ , respectivell;. T h e
rll r22 (0 x)l?,
resulting scalar line integral is
r12 r21 (02 xejl12.

The integrand of (12) requires no such multiplier, since


i t is a scalar quantity. Thus,(12) becomes

In these equations R is the resistance per unit length


of wire, and L indicated integration along the combined Fig. 2-The double helix.
lengths of all wires in the structure.
For a double helix, such as shown in Fig. 2, the fol-
UNIFORMTRANSMISSION
LINES lowing parameters apply:

There are two types of wiretransmissionlinecon-


figurations which areperfectlyuniform (Le., have no
geometrical singular points) the infiniteparallel-wire
line, for which is everywhere zero, and the infinite 27r
helical line, for which is a linear function of distance. 911 922 x
These lines can consist of any number of elements, but P
for a multi-element helical line to be perfectll- uniform, 2nx 27rx
912 90, $21
the various helixes must be concentric, wound in the P P
same direction, and have the same pitch. This paper
will be concerned mainly with two-elementlossless lines,
although the results will also be used to calculate the
performance of single-wire lossless lines.
The complete circuit equations for the first wireof a
uniform two-wire lossless line can be written as

w(t, S)

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
Fowler: Analysis of Helical Transmission Lines by Means of the Complete Circuit Equations 136

If and vary sinusoidally with time, particular solu- COXVERGENCE


OF THE WAVE PARAMETERS
tions of the complete circuit equations representing for-
L,, and S,, as the Fou-
I t mill be convenient to write
ward-wave propagation can be written in the form
rier transforms of
il Ilej(ut-8s: i2 I,ej(ut-Ds) (20)
+1 @lei(ot--8si +2 @zei(t-8s). 4rrmn
?Vhen these quantities are substituted back into the dif-
ferentialequations,theresultingexpressionsarefour
homogeneous linear equations in the four current and
potential coefficients 1 1 , 12, and a2. Thus, (26) and can be written as
/ w 1 WL1,Il wL12I2 (22)
0% WLZlIl WL22IZ (23)
W@l PSllIl PS1212 24)
w@2 0S21I1 Iss2212 (25)
where the Ls and Ss represent the following integrals: and i t is seen t h a t and are the analogs of t and re-
spectively, the variables commonly used in connection
with the Fourier integral to describe functions of time.
M, and T,, will be called the self- and mutual in-
ductance distribution functions and elastance distri-
bution functions, respectively; and their Fourier trans-
forms, L,, and S,,, will be called the self- and mutual
In order to obtain values of 11,1 2 , @ I , and t h a t wave inductances and wave elastances, respectively.
satisfy (22)-(25) and are notall zero, it is necessary that These wave parameters are generalizations of the dis-
tributed inductances and elastances (reciprocal capaci-
the determinant of these four equations be identically
tances) of conventional transmission line theory. How-
zero. This fourth-order determinant can be reduced to
ever, the wave parametersdifferfrompreviousdis-
the second order b y first eliminating @ 1 between (22)
tributed parameters in three respects: (1) they are func-
and (24) and between (23) and (25). The determinant
tions of the phase coefficient as well a s of frequency;
of the resulting two equations in 1 1 and 12 is then
( 2 ) the total numberof wave inductances and elastances
needed t o describe a n AT-wire lineis 2 N Z instead of
2 ( ~ V - l ) ~ ; a n d(3) the wave parameters are not neces-
sarily srmmetrical; i.e., L,, is not identically L,, and
The valuesof which satisfy this equation are the phaseS,, is not identically S,,,.
coefficients of the various modes of propagation of the However, examination of (18) and (19) reveals t h a t
two-wire line.
The phase coefficient for lossless parallel-wire lines is Mmn(x), M n m ( - ~ ) ; T m n ( x ) t Tnm(-x) (33)
readily obtained. By substituting for the Ks, and from which i t is readily shown that
rs in the right memberof (26) and (27) their values given
in i t is seen t h a t L,, Thus, the onlysolu- Lmn(P), L n m ( - - P ) Srnn(B), s n m ( - P ) -
tion of (28) forthin-wire lossless parallellines is
fwjc. Thus, i t can be seen t h a t if is a root of the charac-
Determining the phase coefficients for helical lines is teristic equation (28), then so is, -0,confirming the fact
considerably more complicated. In the first place, is t h a t uniform transmission lines, in common withall
not given explicitly; the expansion of (28) is a quad- other iterated passive networks, are bilateral.
ratic in p2, and the coefficients of this quadratic them- Forthewaveinductancesandelastancestohave
selves depend upon owing to the
term which meaning, it is necessary that the integrals they repre-
appears in (26) and (27). In general, there are as many sent converge. Since the integrands are bounded func-
major modes as there are helixes, and additional minor tions of x, their convergence depends entirely upon the
modes also can occur a s a consequence of the fact that behavior of the integrands for large x. Each of the Ts
L,, and S,, are functions of A second complication has an asymptote of the form
arises from the fact that these functions do not con-
e - j ( o / c ) dA~2+z2
verge for all values of the parameter This difficulty is T(x) (35)
treated in the following section. 47rq/A2 x?

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
136 TRANSACTIOLS-AI\
I,R#E
TESjkAS A X D PROPAGATI0:Y October

We will now consider theFouriertransform of this betweentwosuchintegrandsfromapproachingzero,


function, first for real values of and then for complex unless the two integrands are identical. Therefore, the
values. infinite Fourier integrals involved in determining scalar
For real values of the Fourier transform of T ( N )is5 potential and differences in scalar potential all neces-
sarilydiverge,unless a=O. From this it is seen t h a t
lossy waves cannot exist on a transmission line which is
perfectly uniform in the sense that the line is without
geometric singularities.
which exists for all real except k wic, t h a t is, ex- This restricts the domain of variation of for which
cept for zero arguments of the modified Bessel function solutions can be obtained to points on the real-/3 axis.
KO.Using the small-argument asymptote of this func- However, the presence of the retardationfactor
tion6 e-jurmn!c in ( 2 9 ) and (30) makes i t possible for the wave
inductances and elastances corresponding to real values
of to be complex. I t will beshown in the following
section that if is real and satisfies
where C is Eulers constant 0.5772157 Thus. i t is
seen that the wave elastances of all perfectlJ- uniform
lines (parallel-wire and helical) have logarithmic singu-
larities a t 0 forallintegers K , includingzero, then all thewave
Likewise, for parallel-wire lines the wave inductances parameters are real, except the mutual inductances and
have logarithmic singularities a t However,
elastances of unsymmetrical double helixes (viz., Fig. 2
for helical lines each of the Ms contains an additional for $0 not zero). Furthermore, it is found that to each
periodic factor, cos ,,,,)I of periodicit). 2 n i p , where p is
value of satisfl-ing (38) there correspond one or more
the helix pitch; consequently, the singularities frequencies for which that value is a root of the charac-
of the
waveinductances of helicallines are displacedfrom teristicdeterminant, ( 2 8 ) . Theunsymmetricaldouble
those of the wave elastances by the amounts 2 n yp.
helix is no exception to this, becauseL,, and S,, happen
These singular points are to be conjugates of L,,,and S,,,,, respectively, and as a
result the imaginary part of the characteristic determi-
2ii 2ii
PI=-+--: nant is identically zerofor all real valuesof that satisfy
P C P C (38).
2n 2ii For real values of that do not satisfy (38),all the
wave parameters are complex and not conjugate. I t is
P C
extremelyunlikely thatthese complexvalues would
As a consequence of these singularities of the wave lead to real-/3 roots of the characteristic determinant,
inductances and elastances, the scalar potential diverges except perhaps in isolated instances; that is, one would
for the singular values of This in itself does not pre- expect complex values of L,, and S,, to be associated
vent the propagation of waves having these phase co- with complex-/3 roots of the characteristic determinant,
efficients, because such waves are invariably excited b)- but L,, and for complex values of do not exist,
means of driving voltages which are diferetzces of po- because theintegrals which theyrepresentdiverge.
tential. These differences of potential can be determined Consequently, exponential propagation occurs only for
by taking the Fourier integralof the difference between real values of within the intervals defined b>- (38),
two distribution functions, and such Fourier integrals except perhaps for isolated points outside those inter-
converge for the singular valuesof owing to a mutual vals,where (28) mightbe satisfied coincidentallyby
cancellation of the logarithmic singularities. Therefore, real values of and complex values of the wave param-
the potential differences which occur in the neighbor- eters.
hood of a uniform line are convergent for all real values -An explanation of the physicalsignificance of the
of prohibited ranges of will not be attempted here. How-
For complex values of such as the Fourier ever, some pertinent facts upon which an explanation in
integrandcontainstheexponentialfactor This terms of outwardradiationor excessive velocity of
factor prevents the integrand from approaching zero as energypropagationmightbebasedarecontained in
x approaches Likewise, it prevents the difference (38). For K 0 the prohibited range is

G. A . Campbell and R. M . Foster, Fourier Integrals for Prac-


tical Applications, D. \:an KostrandCompany,Inc.,KewYork,
Y.,pair no. p.
A . Schelkunoff, Electromagnetic Waves,D. \an Kostrand
Company, Inc.,New York, Y . ,chap. 3, p. T h a t is, for propagation to occur, the axial phase ve-

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
locity must be less than the velocity of light. For the where is for LIZ and S I 2 , +#o for L21 and and
other ranges let the wavelengthof propagation be X and zero for LI1, and L22,Sll,and S22; i t is zero for all the
the wavelength of a plane wave in the medium be waveparameters of asymmetricaldoublehelix,for
Then which 0. The product K,K, will be simply K 2 .T h e
2r 2a distribution functions of the double helix are then
(40)
6

Cpon substituting for and w / c in (38) and rearrang-


ing, i t is found that the following prohibited ranges of e-j(u!c)r
T(x) (45)
propagationwavelengthcorrespond to theprohibited -IT
ranges of 0:
Since M ( x ) and T ( x ) are even periodic functionsof
regarded as an independent variable, they can be ex-
panded as Fourier cosine series.

These are small neighborhoods of the vaIues for which


(46)
the helix pitch p is a n integral number of wavelengths.
-Although certain ranges of are prohibited, this does
not necessarilymean that there are also gaps in the
range of frequencies over which propagation can occur.
In fact the plots of phasevelocitygivenlaterreveal
that propagation in one mode or another occurs a t all
frequencies. There are certain modes for which the plot
of phase velocity versus frequency terminatesin a point;
but it alsohappens t h a t some of theseplotsdouble
back on themselves, giving rise to extrema of frequency
and multivaluedness within a single major mode. Es-
pressed differently, the variationof the L ' s and S's with
generateadditionalminormodes;thedomains of
variation of frequency for all of the major and minor
modes form a set of overlapping line segments which
covers the entire frequency spectrum.
However, this is not to say that an actual helix, hav-
ing large but finite maximum-Q values, will propagate
effectively a t all frequencies. The modelbeingcon-
sideredhereis a n infinite-Qstructure.Foranactual
helix to be accurately represented by such a model, the
wave inductance must be very large compared with the
wire resistance per unit length. For certain ranges of the
phase coefficient 0the wave inductance is very low, and
consequently the propagation will be very lossy in the
mode segments associated with those ranges of
HELIXW A V E IYDUCTANCES AND ELASTANCES
To obtainexpressionsfor L,, and S,, suitablefor
numerical calculations, a more compact notationwill be where
used; the subscriptswill be dropped, and the retardation
distance Y,,,, will be represented b17
Lk(p) nz~(.x)e-jflzda
(42) (53)
r \:A2 B? x2 T.

where the numbers A and B assumedifferentvalues s k ( ~ ) J-mtk(.x)e-jdrn+ (54)


for etc. [cf. and
m k ( x ) ,t B ( x ) ,Zk(P), and will becalledthepartial
2ax (43j inductance distribution function, partial elastance dis-
P. tribution function, partial wave inductance, and partial

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
wave elastance, respectively, all of order k . Upon sub-
stituting from (45), (48), and (49), it is seen that
e-i(wlc)r
Zk(B) cos cos k$ej$ld$dx ( 5 5 ) and the total wave inductance and elastance become

1 e- j ( w l c )r

SP(B) cos k$ejsrd+dx. (56)


%a% 7-
The order of integration of these double integrals can
be reversed, providedthat for either orderof integration
thefirstintegral is uniformlyconvergentthroughout where
therange of the secondintegration.'Sincetheinte-
grands are continuous, the finite integrals are uni-
formlyconvergentwithrespectto x. T h e infinite x
integrals(Fouriertransforms)arethesamefunctions The modifiedBesselfunction KO, appearing in (61)
used earlier in this paper [cf. (35) and (36)]. and (62) is real for positive real arguments and complex
forimaginaryarguments.Therefore,it follows from
e - j ( w / c ) 4A'-.4icoa $+I*
e-isrdx (51), ( 5 2 ) , (58), and (59) that all of the and in-
d A 2 B2 COS X
' volved in the expressions for L(P) and S(8j will be real,
provided that
2ak WI
.B
They are uniformly convergent with respect to pro-
p
vided Thus, it is permissible to reverse the for all integers K (positive, negative, and zero). I t also
order of integration in (55) and ( 5 6 ) and the resulting follows from and ( 5 2 ) that since $1 is zero for Lll,
integrals are Lz2,Sll?and S??,thesewaveparametersare real for

'Thus, the wave inductances and elastances are ob- values of satisfying (66) and since is $0 for L1? and
tained as functions of bg; first determining the quan- Slzand for L-21and the mutual wave parameters
titieslo(P), L ( P ) , &(@I, and are conjugate for values of satisfying (66).
which turn out tobe Fourier seriescoefficients of quanti- Somecalculatedvalues of thepartial w a ~ einduc-
ties involving in the radical [P2 l:?. The argu- tances, elastances, and capacitances (inverse elastances)
ments of l k and sk arethenshiftedbytheamounts of single and double helixes are presented in graphical
27rk/p, and the corresponding valuesof the quantities form in Figs. 3-5. For clarity the notation L,,(k =O),
and are entered into thesums in and ( 5 2 ) . L,,(k l), S,.(k=O), S m , ( K = l ) , isused t o
Since appears only in the radical, i t is convenient to represent
the
partial
waveparameters lo. k ,
give t h a t radical a special name and symbol. I t will be SI, correspondingto L,, and S,,, of thedouble
called the phase parameter given by helix. A similar notation L ( k 0), L ( k l ) , etc., is used
(W/C)?]*/? (60) for the single helix.

Thenthepartialwaveinductancesandelastances, SEGREGATION MODES os


OF PROPAGATION
written as functions of 6, become BASISOF WAVELESGTH
Inhelixeswithsufficientlylargeratios of circum-
Zk(6) K o ( 6 d A 2 - B 2 cos cos cos k#d$ (61)
ference to pitch the calculated values of total wave in-
ductance L(P) and total wave elastance S(8) display a
peculiar "wavelength resonance" phenomenon. For ex-
G. A. Gibson, "Advanced Calculus," Macmillan and Company,
Ltd., London, Eng., chap. XIV, theorem I, p. 446; 1931. ample. when is very small compared with 25rjp, 60 is

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
Fig. 3-Si~~gle-helix partial wa\-e inductances, wave elastances, and Fig. 4-Double-helix partial ware inductances of orders 0. 1 ,
na\-e capacitances (inverse wave elastances) of nrders 0. and 2 and 2 vs phase parameter ( b l / b l = 0 . 4 ) .
\-a phase parameter.

small, and and &(Go) areneartheirmaximum


values. On the other hand, for the same values of the
other 6's are large, and the 8ther and are much
smaller than their maximum values, so t h a t for Kl and
K 2 both sufficiently large the calculated values of L(fl)
and are closely approximated by the first terms of
(63) a n d (64').

HELIX

Fig. 5-Double-helix partial


wavecapacitances
(inverse
wave
elastances) of orders 0, 1 and 2 vs phase parameter (b,/b, =0.4).

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
Utilizingtheseapproximationsgreatlyreducesthe
L@) lk(6k)eJk@1
number of terms required in subsequent calculations,
and more important, it enables the calculationof phase S(b) s~~(6~~)ejk+l.
velocity parametrically in terms of one of the phase-
parameter values the choice depending Thus, for near 2 n k / p the quantities Fo, F1, and FZ are
upon the range of wavelengths involved. For example, all explicit functions of and v 2 can be determined in
thephasevelocityandfrequencyarebothmultiple- closed form a s a function of 6r; by solving the quadratic
valued functions of a t wavelengths which are small equation, (73).
compared with 2 n / p (referred t o a svideo wavelengths
in the illustrations); at wavelengthsnear 2a/p (first-
resonancewavelengths) thephasevelocityandfre-
quencyarefunctions of 61, andnear 4 ~ l p(second-
resonance wavelengths) they are functions of Likewise, f= w / 2 a can be determined in closed form
In performing the calculations for the partial u-ave b>- substituting 2 7 ~ f for
j ~ in (63) and solving the re-
inductances and elastances, given in the previous sec- sulting quadratic equation in f . The result is

tion, it is found t h a t each of these quantities is practi- Plots of phase velocity versus frequency are madeby
cally independent of the wireradius a . On the other choosing a sequence of values for the parameter 6 k and
hand, i t can be shown that the error incurred b>- using determining the corresponding values off and v, which
(67)-(72) has an upper bound which is IargelJ- affected are then plotted against each other. For each value of
by the choice of a , and most likely that error itself in- there are two valuesof owing to the f sign in
volves the wire radius as a principal variable. Certainl!. which senyes to identify the two major modes; the plus
this is true in the limit as a approaches zero, for then sign corresponds to the major mode associated with the
the total wave inductance and total wave elastance di- smaller helix. and the minus sign corresponds to the
verge [cf. (13) 1. Thus, the partial wave inductances and major mode asscciated &h the larger helix. -Also, for
elastances individually display the attributes of exter- each value of there correspond two values of f,owing
nal impedance and collectively display the attributes to the f sign in (80), causing each of the major modes
of internal impedance.S The data of this paper n-ill t o be divided into two separate curves. The square root
take cognizance of externalimpedanceonly.andfor of z l ? hastwovalues: correspondingtoforward-
t h a t reason the wire radius has been set equal tozero i n travelingwaves,and -21, corresponding to backward-
thecalculatedpartialwaveparametersgivenabove. traveling waves. Thus, to each value of 6 e there corre-
Likewise, it does not enter into calculations given Inter spondeightdistinctpoints in the positive-frequency
in which the total wave parameters are each approxi- half of thef-21 plane, except that fork 0 there are only
mated by a single term from their expansions i n terms four distinct points, owing to the fact that each of the
of partial wave inductances and elastances. eight roots coincides with one other root, thereby re-
By substituting w l v for the characteristic determi- ducing the number of distinct roots to four.
nant (28) when expanded becomes a quadratic equation In the case of the single helix, is given by as
in 9 , before, but the phase velocitl- squaredis
Fc F 2 ~ 4
212 Sk(6k)/lk(6k) (slj
where the quantity w has been eliminated by factoring.
and the coefficients are the following functions of L,,, which does not contain a f sign. Thus, the number of
and Smn: majormodes is one, but this major mode subdivides
into two curves, owing to the sign in (80).
Fo S11S2, SI&rl
Illustrative plots of phase velocity, impedance, and
F1 L11S2? L2dll LI.?S?l LplS1. (75) coupling coefficients are given in Figs. 6-13. Figs. 6-10
F2 LllL22 L12L21. (76) apply to single helixes having various ratios of circum-
ference to pitch. Phase velocity is plotted Ko/c, the
At wavelengths near. 27rkip substitutions of the fol- ratio of thephasevelocity,measuredalongthewire
lowing typecan bemadeforeach of thequantities instead of axially, to the velocity of light. This ratio
L,, and Smn: tendstounity forlargephaseparametersor at fre-
Ran0 and Whinnery, o p . chap. 6 , pp. quencies far removed from the critical frequencies.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
Fotder: Analysis of Helical
Transmission Lines by N e a n s of the
Complete
Circuit
Equations 141

Fig. 7-Phase velocity of single helix vs frequency for


Fig. 6-Phase velocities of single helix vs frequency. video wax-elengths.

3.90 ,.PO
Frq- Critical

Fig. 10-Characteristic impedances of single helix vs frequency for video, first-resonance, and second-resonance wavelengths.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
This indicates a tendency for the propagated wave to
follow the contour of the wire. The characteristic im-
pednnce of the single helix, shown in Fig. 10, is simply

T o aid in identiiJ-ing the various curves in this figure


some of them have been extended far beyond the last
calculated points (2b8 24) by means of straight dotted
lines. Thecurvesrelatedtosecond-resonancewave-
lengths terminate a t points corresponding to 2bas=0.
This is indicatedbpverticalstraightlinesemanating
from the terminal points.
Curves applying to double helixes having bl;br=0.4
aregiven in Figs. 11-13. The electricandmagnetic
coupling coefficients, h and k, respectively, are plotted
in Fig. 11 functions of the phase parameter. These
coefficients bothapproachzero as 6 approachesin-
finit,-. Consequently, a t frequencies far removed from
all the critical frequencies of the individual helixes the
double helix behavesliketwoseparatesinglehelixes,
Fig. 11-Coupling coefficients of double helix vs phaseparameter eachpropagating a wavewhich follows its respective
for video,first-resonance. and second-resonancewavelengths
(bl/bz-0.4). contour.This is evidencedalso bl; theplot of phase

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.
Fotder: Analysis of H e l i d Tranmzission Lines by Means of the Complete Circuit Equations 143

I
I
0

.2 .3 .6 2 3 4 678910' 2 6789102 2 3 6 789103

Fig. 13-Characteristic network component impedances of double helix vs frequency (video wavelengths)

velocity,measuredalongthe outer helix, versusfre-presentthesameimpedance as doesthedouble helix


quency, given in Fig. 12. For large values of the phase when it is propagating a single fonvard-traveling wave.
parameters 61, and that ratio approaches unityfor
ACKNOWLEDGEMENTS
the outer helix and for the inner helix. T h e im-
pedance of a double helix a t video wavelengths is repre- The author wishes to thank Dr. H. M. Von Foerster
sented in Fig. 13 by three impedances, and for his many helpful suggestions and criticisms through-
which, when connected as a T-section between the inner out the course of this investigation, as well as Dr. E, C.
and outer helixes and a point of zero scalar potential, Jordan for reviewing the manuscript.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:025 UTC from IE Xplore. Restricon aply.