fEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. MTT-16, NO.

2, FEBRUARY 1968

Synthesis of Symmetrical Branch-Guide

Directional Couplers

RALPH LEVY, SENIOR MEMBER, IEEE, AND LARRY F. LIND, STUDENT MEMBER, IEEE

A&s trac t—A synthesis procedure is described for the design of results for the Chebyshev cases. In fact, the Chebyshev case
hrancb-gnide directional couplers which gives results showing a signifi- yields VSWR and directivity characteristics which are al-
cant improvement over previous approximate methods. The synthesis most equal ripple, and the exact performance is given by
technique adopted gives exact Butterworth characteristics and almost computer analysis. The new theory also gives designs which
exact Chebyshev equal-ripple characteristics, the deviations in the latter have considerably superior performance for a given number
case being so small that in most practical cases they may be neglected. of branch guides as compared with previously published
The design of branch-guide couplers for bandwidths of greater than one results.
octave is demonstrated.
The design information for a large uumber of cases of practical inter-
est is presented in tabular form, and experimental results for several
branch-guide couplers constructed in waveguide and in stripline are in
good agreement with the theory. The technique could prove valuable in
the design of microminiature stripline hybrids and couplers.

INTRODUCTION Fig. 1. Cross section of the branch-guide directional coupler.

RANCH-GUIDE couplers may be constructed using

THEORY
many types of waveguides or transmission lines. In
B the branch-guide coupler using rectangular wave- It is assumed that the coupler is symmetrical about the
guides, the coupling region consists of a number of series plane indicated by the dotted line in Fig. 1. The analysis
branch lines, each XgO/4 in length at the midband frequency, of the coupler may then be based on an analysis of the nor-
connecting the two waveguides. These branch lines are them- mal modes of the four-port, i.e., the even and odd modes. [~1
selves spaced at intervals of AgO/4, as shown in Fig. 1. To The even and odd two-port networks are shown in Fig. 2(a)
obtain an optimum broadband performance it is necessary and (b), respectively. The even-mode circuit consists of a
to specify the impedances of both the branch guides and the number of series or shunt open-circuited stubs of electrical
connecting or main line waveguide sections. In the case of length 0 connected by unit elements of electrical length 20.
TEM-mode transmission lines, the branch lines are in shunt The immittances of the stubs are denoted by al, az, . . ., an
rather than in series with the main lines, representing the and the main line immittances by bl, bz, . . ., &_l. The Odd-
dual of the waveguide case. mode circuit differs only by the termination of the stubs in
A short survey of the previous work on the design of short circuits rather than open circuits.
branch-guide directional couplers has been given by one of For the even-mode circuit, a reflection coefficient I’, and
the authors. II] It is concluded that present theories are all a transmission coefficient T. are determined, and similarly
based on certain approximations which lead to considerable for the odd-mode circuit, 170and TO are determined. Super-
uncertainty in the realization of a broadband branch-guide position gives the following vector amplitudes of the signals
coupler. In the latest of these papers, [z] empirical correction emerging from the four-port in the case of a unit wave
curves are presented which enable a designer to realize a amplitude incident on port 1 with all ports terminated in
broadband device with maximally flat or equal-ripple VSWR their characteristic impedances 131(in this case all normalized
or directivity characteristics. These curves enable a good esti- to unity)
mate to be made of actual bandwidths but are less accurate
A, = ;(re + ro)
for the prediction of the magnitudes of the VSWR and
directivity in the band. A, = *(r, – r.)
In this paper a method is presented which gives exact A, = +(2’, – To)
results for maximally flat characteristics and almost exact
AA = *(T’. + To). (1)
Manuscript received April 28, 1967; revised September 18, 1967.
The values of the reflection coefficient 1? and transmission
This work was supported by a research agreement with the Ministry of
Aviation. coefficient T for a network are given in terms of the transfer
R. Levy was with the Department of Electrical and Electronic Engi- matrix parameters by the well-known formulascsl
neering, University of Leeds, Leeds, England. He is now with Micro-
wave Development Laboratories, Needham, Mass.
L. F, Lind is with the Department of Electrical and Electronic
r=(A– D)+(B–c)
(2)
Engineering, University of Leeds, Leeds, England. A+ D+B+C
 ,..
LEVY AND LIND : SYNTHESIS OF DIRECTIONAL COUFLERS 81

and of a short-circuited shunt stub of the same admittance by

10

(a)

A restriction imposed
11
[
;1”

on the branch-guide coupler

(7)

is that

-=1
any main line connecting two stubs of electrical length 20
should be of uniform impedance, i.e., should consist of a
double-length unit element of characteristic admittance bi
having transfer matrix

Fig. 2. (a) Even-mode two-port network.

(b) Odd-mode two-port network.

g
T = —-———– (3) The restriction imposed above has a most valuable property,
A+ D+B+C
namely that (8) is invariant with respect to the transforma-
tion
where it is assumed that the terminating impedances are
normalized to unity. For a lossless network, we have also
J- (9)
t
lr[’+l~l’=l. (4)
except for a change of sign. Note that (7) may be obtained
For a given branch-guide coupler, the even- and odd-mode from (6) by using this same transformation (9). Thus if
networks may be analyzed and their respective transfer the transfer matrix of the even-mode circuit is calculated,
matrices determined. By application of (2) and (3), I’e, Te, then the transfer matrix for the odd-mode circuit is obtained
and I’o, T. may be evaluated. For each pair of coefficients (4) using (9). Mathematically, if the even-mode transfer matrix
holds, since the even- and odd-mode networks are Iossless. is
Equations (1) are used to determine the waves emerging
A,(t) Be(t)
from the four-port.
not represent
preferably
This is a process of analysis,
a satisfactory
be one of synthesis
design technique,
from specified
which
and does
should
performance
[ c.(t) D.(t) 1 (lo)

then the odd-mode transfer matrix is

characteristics. Such a synthesis technique is presented in
this paper for specified reelection coeficien t and isolation,
but it is found that in common with previous design tech-
+) B.(:-)1

niques the coupling characteristics are not controllable for

A. B.
= (–~).-l
the symmetrical branch-guide couplers considered.
[1Co D.

In order for a synthesis procedure to be formulated, it is [ C6(+) D.(’:)J ’11)

necessary to prescribe functions representing the input re-
When the transfer matrices of the stubs and the double-
flection coefficient and isolation. These will be rational func-
length unit elements are multiplied in the appropriate order,
tions in the Richards’ variable
then the even-mode transfer matrix is found to be of the
form

where
A. B. 1 An_,(t’) tBn_2(t2)
(12)
[1C. D, = (1 – [
t’)n-’tC._,(t2)D._,(t’) 1
fl=tan O (5)
where the subscripts indicate the degree of the polynomials
o being the commensurate electrical length indicated in Fig, A(t’), . . ., D(t2). Hence from (11), the odd-mode transfer
2. The polynomial forms may be determined by analyzing matrix must be
even- and odd-mode networks of low-order couplers, and by
A. B. (– l)7L+1~2(n-1)

induction, the general results for n branch guides are ob-

tained. [1Co Do = (1 – t’)”-’

Now the transfer matrix of an open-circuited shunt stub

of admittance a~ is given by
[‘n-w+(+)
r
La,t
01
lJ
(6)
“1+C4) “ ’13)
‘n-(+) 1
82 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, FEBRUARY 1968

Subsequently, only a symmetrical form of branch-guide where P~_l(t2) is a polynomial in tz of order (n — 1). From
coupler will be considered, i.e., referring to Fig. 2 we have the t+l/t transformation of (9) as expressed in (13), it
follows that
av = afi+r_l
T =1,2,.... (14)
b, = bn_,

()
}
(– 1)~-1~2~-3pn_1 1
t2
r.
The symmetrical coupler is the only form which has been (19)
(1 _ ~2)n-1 “
previously considered, but it is believed that asymmetric F.=

couplers may be of interest and this could form the subject

of a future paper. The band center frequency of a branch-guide coupler occurs
The symmetry restriction in terms of the transfer matrix when the branch guides are kg/4 in length corresponding
representations of (12) and (13) is expressed by the condi- to 0 = 45° in Fig. 2, and to t=jl.
tion
THE BUTTERWORTH SPECIFICATION
A._,(t’) = Dn_,(t’). (15) If all the zeros of the F* functions in (17) are located at
the center frequency given by t=jl, then (17) becomes
As stated previously, it is desirable to prescribe the input
reflection coefficient and isolation [A ~ and A.2 of (l)] which
are sums and differences of r. and r., i.e., functions like ~+ = ~ ~* (–1)”-1 (1+ p)n_, (20)
“(1 _ ~2)?z-1
(2) with a numerator (B– C). This means that when Al and [ t 1
A~ are formulated, they each consist of ratios of rational
where K is a constant which determines the coupling. Since
numerator and denominator polynomials. Hence it is very
a splitting of an F+ function into its component even- and
difficult to specify Al and A z in this way since the coefficients
odd-mode parts is unique, (20) may be readily decomposed
of the numerator polynomial are closely related in a com-
to give
plex manner to those of the denominator polynomial.
Consider, however, the function [see (2) and (3)] I’e t(l – P)”-1
=K (21)
r B–C E (1 – ~z)n–1
—.— (16)
T 2 and

assuming the symmetry condition A = D. It will be found

that the functions (22)

Q=++’
() eo
(17) in agreement with
there are (n – 1) zeros of rJTe
the form of ( 18) and (19). It is seen that
and I’O/T. at t2= – 1 cor-
have a much simpler form which maybe specified by simple responding to those of F* given by (20), as previously

Butterworth or Chebyshev polynomials. In fact, if the stated. Either (21) or (22) may be used for the synthesis
moduli of these functions (17) are constrained to be small process to be described later.
over a passband, then since (4) holds, ] r, I and I J7.[ must
THE CHEBYSHEVEQUIRIPPLE SPECIFICATION
also be small. The situation is even more advantageous than
outlined by this simple consideration because for the Butter- In order to specify the Chevyshev function, it is convenient
worth and Chebyshev cases it will be shown that the zeros firstly to rewrite (12) and (13) in terms of cos 28, tan O, and
of cot O; instead of t =j tan 0, giving

A. Be

()C. A,

are those of I’. and r., separately. Hence the Butterworth An-l(–cos 20) j tan 0S. .2(cos 269
. (23)
case represents an exact synthesis procedure, and the
Chebyshev case gives an almost exact result. The only devia-
[ j tan OGl(cos 20) Z.–I( – Cos 20) 1
tions in the latter case from equal-ripple performance of A ~ A. B.
and A ~ [see (1)] occur in the ripple levels, which are not ()Co A.
quite equal in amplitude, and in the exact location of the
edge of the passband. In practice, the deviations are ex- Z.-,(COS 20) j cot OS.-*( – Cos 20)
—
— (24)
tremely small and may be safely neglected. [ j cOt eCn_l( – cos 2e) Zn-l(cos 20) 1
Analysis shows that the function I’,/Te takes the general
form Note that there are two main differences between (23) and
(24). The ~ and C polynomials are multiplied by j tan 0 for
r. tPn-l(t’)
—= (18) (23) andj cot o for (24), and the coefficients of the odd parts
j“, (1 – P)”-’ of the respective ~, ~, and ~ polynomials differ only in
LEVY AND LIND : SYNTHESIS OF DIRECTIONAL COUPLERS 83

sign. For the symmetrical networks considered, B and ~ THE SYNTHESISPROCEDURE

are either both even or both odd, so that (B.— CJ differs
The function I’JTe has been specified for the Butterworth
from (Be– CJ only by a multiplying factor ~ tan?f). The F+
case by (21) and for the Chebyshev case by @9), These func-
function of (17) is then given by
tions may be realized by a cascade of (n– 1) double-length
unit elements with n single-length open-circuited shunt stubs
F+=; :+; at the junctions, with the exception of certain extreme specif-
[1 o
ications were negative element values occur. This is because
1 cos 20 the specification functions, in general, always lead to a
. ~ jK tan 13Pn-1 —
– realization consisting of (2n —2) single-length unit elements
. () Cos 26.
with one or more stubs at the junctions, but we are inter-
Cos 20 ested only in the special and unique case of a cascacle of
+ ; jK cot OPm.I — double-length unit elements with shunt stubs, the realization
() cos 20.
of which is more restricted than the general case. This re-
.
sfz~
p ()
~—n—————
‘1
Cos 20

Cos 20.
(25)
striction has not been expressed in analytical form, and it
is doubtful whether such a form exists. In practice, it has
been found that negative element values occur only for un-
where K is a constant. Now consider the function reasonable performance specifications, i.e., a small number
of branch guides with tight coupling and large bandwidth.
PL=1+]F+12
The standard Darlington synthesis procedure is employed.
From (4) the modulus squared of the even-mode reflection

=1+
K2#_;

()?

xc
(26)
coefficients is given in terms of \ I’JTe ] as

(1 – x’)
(31)
where

x = Cos 20, Xc = cos 2ec, (27) r. itself is determined by choosing only left half t-plane poles,
and the symmetry of the network is assured because the
and K, as in the Butterworth case, controls the coupling, numerator of (31) is the square of an odd function in t.
and in this case also the ripple level. Equation (26) has the The factorization of the denominator of (31) must be car-
generic form of the insertion loss for a cascade of (n– 1) unit ried out numerically, i.e., by digital computation, since there
elements and one short-circuited shunt stub of commensu- is no closed form solution. Having found r., the input im-
rate electrical length 20, as described by Riblet 141and Carlin pedance is determined from the formula
and Kohler. [s] They each showed that in order to give
l+r. Ae+Be
Chebyshev equal ripple response with (n – 2) ripples in the ze=— (32)
band – x.< ~< x., P._l(x/xJ should be set equal to l–re=ce+A.

from which the even-mode transfer matrix parameters are

determined using the fact that Ae is an even polynomial, and
B. and C. are odd polynomials in t.
The first step in the synthesis of the driving point im-
(28)
4 \ .&c/
pedance Z, is the extraction of a series open-circuited stub
of characteristic impedance al of such value that a double-
where TJx) is the Chebyshev polynomial of the first kind len.gth unit element of uniform characteristic impedance bl
of degree n. Referring to (25) it is seen that may then be extracted. It is easily shown that th; condition
for extraction of a double-length unit element from a driving
r,
— = jK tan OPn_I z (29) point impedance Z(t) is that both the open- and short-
T. () xc circuited impedances derived from Z(t) have a derivative
with respect to t which is zero at t= + 1. Using this fact, it
where
is simple to show that when the above condition is not
satisfied, it is necessary only to perform first the extraction
(30)
x. of a series open-circuited stub of characteristic impedance
‘=(+%)”(-)
al = - Z.:(l) = – Z,;(l) (33)
and it is obvious that I’JTO has the same zeros in the r-plane
as I’,/Te (excluding the origin and infinity). Thus (29) and where the prime denotes differentiation with respect to t.
(30) represent the specification functions to give nearly equi- In terms of the transfer matrix parameters it is worth re-
ripple reflection coefficient and isolation—not exactly equal membering that
ripple because the values of xl 1 and AZ given by (1) do not
contain the factors T, and To, which, however, have moduli z..=;> z,.=;. (34)
very close to unity over the passband.
84 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, FEBRUARY 1968

TABLE I
ELEMENT VALUES FOR BUTTERWORTH BRANCH-GUIDE COUPLERS FOR n = 3 TO 9 BRANCHES

B.E. B84wn ‘081DE

B7WWH OUIDE 11.?.UT’CANCES II MAIN LINS lWITTAXCES IWITMWEX

Vsm b a4 a5
az a3 a4 a5 1 b3 b4 ‘2 a3
IIal —.
b2 .
—.
‘,-3 m-,

m 2+3 17.3 ,.46 1.02 0. W33 0.0193 OJJJ!69 0.0620

lEm%rx-
19.5 1.19 1.01 0.0501 u,l W8 1.0039
15 lIJ.6 1.18 1.U1 o. M96 0.1s24 1.0128 15 12.5 1.44 1.03 “. Ix& 0.0350 0. ‘b242 “.,, ,,
10 9.5? 1.17 1.03 0. 16~3 0,3450 1.0445 10 7.70 1.41 1.06 “.01,8 0.0546 0.1564 “.*IJ9, 1,0049 1.0393 1.0653

L1-.l
8 7.6o 1.16 1.CX! (1. 2~6 0.4607 1,0757 8 5.77 1.40 1.08 0.0152 0.0830 0,2(%3 0.2817 1.OWO 1.(3528 1.1126
6 5.60 1,15 1.07 0.’2.587 0,6448 6 4.OC 1,38 ,a, ~ “.”796 0.,”72 “.z@6 “.&w5 1.0170 1.1oLI9 1.2316
1.1342
5 4.70 1.14 1,09 0.3078 0.7874 1.1S33 5 3.11 1.37 1.15 0.0320 0.1!22’2 0.3454 “.5,$3 1.(E41 1.1444 1.3392
0.3553 II. 9974 1.9573 4 2.26 1.36 1.19 “.0344 0.1395 0.4367 0.7w1 1.0352 1.2146 1.5239
4 3.70 1.14 1.1’2
3 2.69 1.14 1.$9 0.4149 1,3432 1.3n4 3 1.46 1.35 1.25 0.0265 0.1593 0. 5942+ ,. “7”9 1.0537, 1.3391 1.8861

!.5 2.2~ 1.13 1.22 0.4513 1.6188 1.4693 2.5 1.10 1.34 1.30 0.02’71 0.1700 0.7276 1.4363 1.8682 1.4426 2.2214
2 1.7& 1.13 I .29 ().494, p. “3*5 2 WC 1.33 1.39 0.0272 0.1810 a. 9427 2.1,98 1.cS93 1.6820 2.8004 _
1, 59%
— n =8
—
T
n-4 20 16.9 1.49 ?.02 U. DU17 0.0115 0.0332 u. 092 1.0002 t.0014 1. W39.1. M15
20 ,g. o 1.29 ,.0, “. “253 o.q~~ 1.0033 1,W53 15 ,2., 1.47 1.04 0. C032 0.”21” 0, “598 00 “97, 1.0007 1.0c48 1.0137 1.019
15 ,4. e 1.28 1.02 “. @1514 0.,355 1.0o79 1.0176 10 7.33 1.44 1,06 “. 0(%’2 0.0392 U.11C4 0.1819 1.0027 1.0188 1.0532’ l.m4
10 9.09 1.26 1.84 0.0’821 IJ.2Y34 1.0281 1.0632
8 5.43 1.43 1.03 O.wal 0.0505 0.1440 ().2443 1.W49 1.0342 1.c975. I.137
8 ~. ,5 1.25 1.05 0.1 W4 0.3357 1.0486 1,1101
6 3.68 1.41 1.13 fJ. IXL)6 0.0648 0.1935 0.3517 1.0095 1.0662 1.1919:1.274
6 3.2° 1.25 l.I38 0.1328 0,4646 1.0875 1.2o28
3 2.33 1.40 1.16 0.0119 u. U733 0.2291 0.4447 1.0136 1.0951 1.2817/ 1.4w
5 4.2ti 1.23 1.12 0.1498 (3.5635 1.12C4 1.2850 4 2.00 1.39 1.!2” “.”,33 0.2785 0,60,5
0.”823 1.0201 1.1419 1.4357! 1.648
4 3.31 1.= 1.15 0,1688 0.7823 1.17o2 1.4165 3 1.25 1.38 1.29 “.0,45 0.@13 “.3548 0.92, “ 1.0311 1.2245 1.7354 2.147

Dl-
3 2.39 1.21 1.20 0.1895 0.9466 1.2510 1.6488 1.37 1.33 0.0149
?.5 .920 U. 0952 0.4129 1.2401 1,0399 1,29282.01052.639
‘.5 1.09 1.21 1.23 0.200” 1.1377 1.3125 1.8422 2 1.35 1.34 0 .0149
.681 0.0979 0.4978 1,8466 1.0530 t .39772 .4811 3.55o
1 .W 1,28 “.2, ”2 , .4y74 . .—
2 l.m 1.3293 2.1411
. n-g
n-s
m 16.5 I .52 1.0’2 0. mm a. W.%7 U. 0224 0. LY+36 0,0539 1. owl 1. IXYJ8 1.13028 l.o@f
* 18.4 1.36 1.02 V.UW?2 0.0302 0.0749 1.wl3 1.0c46 15 ,,.7 1.50 ,.W 9.0017 0.0124 00 CJW7 0.0781 0. “ga , . “W3 , . ““~ ,ofJ, ”rJ ,.”, &

IIII
Is 13.5 t .34 1.0s 0.8233 u. 092 0.1348 1.0C44 1.0136 10 6.95 1.47 1.08 % (X)33 0.0235 0.oj’34 “.1455 “.,8,2 , .“”,4 1.(),,6 ,. “396 , . “7., f
w 8.63 1.32 1.05 0.0426 “.l&5 “.2542 1.0163 1.0573 8 5.21 1.45 1.09 0. C!C43 0. 03(!4 u. 0976 0.1934 “.245, ,.fj”q ,.0214 , .~yl ,.,33(
8 6.70 1,31 1.07 0.0542 0.2172 L).3411 1.0286 I.1013 6 3.33 1.44 1.14 0.0057 0.0392 0.1285 0..2725 0,36”, , .“”~z .,. @/, ~ ,. ,44,5 1 .z69,
6 11.76 $.50 *.*” 0.06s7 u. 2922 IJ. 4857 1.WQ3 1.1828 5 2.52 1.43 1.17 0.0065 0.c441 0.149o 0.3361 0.465” 1.(x)76 ,ofJfjfJ6 ,..2,~~ , .t~

5 3.9’2 1.29 1.12 0. U/w 0.3460 0.6048 1.0727 1.26% 4 1.75 1.42 1.24 0.0073 0.0492 U. 1?50 0.4441 0.6517 1.0113 .1. C9w 1.3285 1.6495
4 2.91 1.28 1.16 0.0857 0.42”” “.7931 1.1036 1.3922 3 7.09 1.40 1.28 u. 0080 (3. 0537 0.2098 0.6485 1. “653 ,.”,7g ,.1446 ,. fins P., 6&
3 2.01 1.27 1.22 0.0941 0.5318 1.1383 1.1541 1.6326 ?.5 .900 1.39 1.32 3.0032 0.0553 0.2328 u. 8428 1.518o 1.0232 1.1s91 1.7505 2.691$
2.5 1.63 1.26 1.24 CI.U5T6 0.6148 1,4457 1.1526 1.83(J4 2 . 5?!0 I .3a 1.35 0.LW83 0.0557 0. 26N 1.1952 2.45$Ig ,.”3,3 , .25.7P 2. “8.j3 3..683$
2 1.18 t .26 1.33 0.0999 0.7323 1.%16 1.247u 2.1417

m
.-6

TrHEF3T
1.ow7 1.W33 1.0034

1.uu24 1.u115 1.(x85

10 P,.12 1.371.”6 0.”224 “.1”48 “.2”% 1.U091 1.u431 1.0694
1.0162 1.u769 t.lz46
6 b.33 1.341.11 0.03640.17740.3957 1.u3u2 1.1453 1.2332

ffithti
1.0423 1.207> 1.34S0
1.610 1.3079 !.5311
3 1.73 1.31 1.22 (),0493 0.28740.9162 1. IJ917 1.4876 1.8.322
~5 1.31 1.31 1.28 ().llm 0.3179 1.1642 1.1134 t.6.yM 2.2029
1.1493 1.87392.7433

TABLE 11
ELEMENT VALUES FOR CHEBYSHEV BRANCH-GUIDE COUPLERS FOR n =3 TO 8 BRANCHES

~ .E. wMNCH WIDE rWT7ANCES ‘;111 LTNS 1WITT,4NCES

RR,wcn 083DE IWITMNCES
“~ ‘y Cesb
& ~m. V3NF ~,la21a31a4’ ‘y”
IS2 COUP COUP DIR . Vsm 4
al ‘2 “3 a4 bl “% b3 b4
(DB) (DB) [DB)
n=3 II =4
.
2U 19.9 36.8 1.WC 0.0509 0.0993 1.0039
15
10
,JJ. $1 35.9
9.86 34.4
1 .W1
1. cob
0.091 t
0.1656
0.1795
“.3385
1.0130
1.0451
20
13
10
19.9
14.$1
9.97
58.6
57.4
55.5
T .000
1 .“00
1 .“””
0.0258
o. cJ465 “.1344
“. “843 “.25,3
0.0747
RTzr
1.0287 1.0634

I---L
.1 8 7.’35 33.5 1.007 0.2124 0. JJ512 1. 077U .1 8 7. F6 54.4 1.001 0.1073 0.3329 1.0498 1.1106
6 5.24 32.5 1.of5 0.2761 LI.6301 1.1369 0.1371 0.46”7
6 5.$26 53.1 1.001 1 .0s99 1.2(!41
5 4.83 31.9 1.021 0.3174 0.7686 1.1874
5 b.-6 52.3 1.002 0.1550 “.55139 1.1240 1.2873
4 3.83 31.2 1.03U 0.3681 0.9729 1.2641
JJ 3.B6 51.4 1.003 0.1753 0.7032 1.1758 1.4207
3 2.33 30.4 1.o45 0.4330 1.31o3 1.3899
3 2.-6 5f.1.3 1.W4 0.1978 o.g41g 1.26I36 1.6575
20 19.5 24.5 1 .0(!4 0.0531 (L Llg48 1.W41 20 19.5 40.3 1. 0“” 0.0275 “. “73” 1. w25 1.0053
15 7JJ.5 23.5 1. W9 0.0957 “, 171j3 1,0134 15 14.5 39.1 1 .Oul 0.0497 0.,312 1.0085 1.0777
10 ~. 4. 2,.9 , ,024 “.,76, “m~,~~ 1 .m71 10 q. 117 37.1 1.0U4 “.”g,2 0.24140 1.0307 1,u640
1,2 8 7.38 21.0 1.040 0.2280 U.4207 1.0808 1.2 8 7.115 35.9 1 .W7 0.1171 0.3233 1.0535 1.1121

k
6j .34 19.9 q . U73 0.3w8 0.5830 1.1454 34.6 1.u13 0.1513 “.4474
6 5.43 1.0977 1.2082
54 .32 19.3 1.1OP U.3499 0.7087 1.2011
5 b.42 33.7 1.019 0.1726 “.5432 1.1357 1.2945
43 .30 18.5 1.150 u.413u 0.8958 1.288o 4 3.fil 32.8 I. 027 0.1975 0.6851 1.19U4 7.4301
32 .29 17.7 1.236 “,~””? 7.2,56 1.4391
3 2.4? 31.6 1.c43 0.2268 o.g235 1.2926 1.6859
20 1“.9 16.9 1.013 0.0573 0, “864 1 .W43 .
20 ,E.’q 29.2 T. ““3 0.0306 “. “699 1.0027 1.0053
I .3 15 13.8 15.9 1.029 0 .1043 0.1532 1.0142
7.075 “. j g6p “. 2786 15 13.9 28.0 1 .W7 0.0559 “. 125” 1.0093 1.0179
lC Q.65 14.2 1 .q 5C4
—
10 8.76 25.9 1.020 “., U1./ “.23, ” 1.03411 1.065’2
1.3 8 6.72 24.7 1.033 0.1365 0.3043 1.0606 1.1149
S& .67 23.2 1. “~~ “.,~fi “.4193 1.1126 1.2158
53 .66 22.3 1.084 u. 2097 “. 5086 1.1585 1.3076
42 .66 21.2 1.124 0. 2a59 “. @~Q 1.2313 1.4588
31 .77 20.0 1.197 0 ,2932 (). 87”5 1.3577 1.7379
201 &f! 21.0 1.011 “. “339 O. c,546 1.w31 1.0054
15 12.9 19.6 1.025 0 . @5’5 “.,143 1.u181
1,010?
I -~ 10 7 .77 t7.4 1.067 “ .13y (j . mj~ 1 .C4D4 1.0669
85 .63 16.1 1.110 0 .1726 “ .2683 1.0722 1.1190
63 .61 14.4 t .202 0 .2377 “ .3@ 1.1373 1.2261
52 .66 13.5 1.2* 0 .2&#3 Q.4335 1.196o t.w47
—
LEVY AND LIND : SYNTHESIS OF DIRECTIONAL COUPLERS 85

TABLE II (Confd)

~ B. E. n3ANcH 0uID3 mTTANCES VAIN LIKE 1,*.TITANCES

F?2 C.:P
23)
mm?
m
DIP .
1~~)
w n .2
1 ‘2 ‘3 a4
b
1
b
2
I b3 b4

-w
n-5 n.?
—
11.0”1311.00461 I 20 19.0 65.4 1. o“” “. 0C48 “.”>,6 “.”455 “.”573 1, W05 1.0025 1.0051

ITxEEEr
‘Q:
15 lti.9 79. u I.OW 0.0240 0.0502 0.1335 8. LX145 1.0157 15 13.9 C3.6 1 .“()” “.0”9” “.”394 “.”815 “., ”22 1.W18 1.0”89 1.0177
10 9.:7 76.8 1.uW 0.0439 0.1666 0.2514 1.0168 1.U577 10 8.87 60.6 1 .LmJ 0,0176 0.0735 0,1512 0.1906 1 .W74 1.0347 1.0686
8 7.87 75.51 .UW 0.0360 v.e175 0.3371 1.0295 1.1021 ,3 8 6.@6 58. e 1. WI c.0232 0.0955 0.1999 0.2561 1.0136 1.0636 1.1263
1.0342 1.1917 6 4.85 56.6 1 .Uul 0.0306 0.1255 0.2784 “.371 1 1.0265 1.l#L6 I .252 I
5 4:87 72:91 :ow 0.08000.34730.5976 1.0753 1.2720 5 3,86 55.3 1,092 0.0350 0.1451 0.3417 0,4734 1. 0382? 1.1837 1.3?56
1.1077 1.4049 412.68 53.71.””310.”39!? 0.1692 “.4411 0.6513 !.”572 1,2776,.5%3
3 70.51.000 0.0935] 0.537011.1277 3 1.92 51.7 l.w4 ().0446 0.2w1 0.6257 1.0330 1.0906 1.4$82 2.0512

+t+tli
2.:7 1.161J~ 1.644c
-.
2LI 19. > 55.9 1.WU 0.0143 0.WJ2 0.0721 1. LXJ14 1.0047 .1 49.0 1.OW 0.u064 0.0233 0.0440 0 fJ536 1.uw7 1.w2R 1.0052
15 14.5 54.5 l.wO 0,0261 o.wOl 0.1294 1 .W!49 1.U159 15 13.0
=l-’+Tfl~Lu~-
47 I 1.u”7 0.01P3 “.c427 0.07L77 0.0g50 1.0s25 1,0099 1.0181
10 9.49 52.1 1.001 0.0484 0.1665 0.2427 1.0183 l.waa 10 7.93 43.9 1.0ti3 0.u246 0.0802 0.1454 u.1757 1.oIu:? 1.0397 1.0716

1M-Ill YMK“
.2 8 7.11- W.8 l.wl 0.c621 u.2178 u.3249 1. W24 1.10451 4 G 5.89/ 42.0 1.005 0.0331 0.1CJ48 0.1924 0.2354 1.0191 1.0738 1.1335
6 5.46 49,1 t.w2 0.07980,29450.4619 1.0601 1,19761 6 3.89 39.6 1.010 0.0448 0.1396 u.2698 0.3417 1. fJ382 1.1480 1.2722
5 4.46 48.2 1.003 U.WU3 0.35040.5736 1.0842 1.2828 5 2.91 38.1 1.014 0.0523 0.1631 0.3343 0.4388 1.0563 1.2199 1.4124
4 3.JJ6 47.o 1.w5 CI.1OI7 0.4289 u.757(J 1.1215 1.4229 4 f.97 36.4 1.022 0.0611 0.1936 0.4402 0.6138 1.0864 1.3453 1.6709
3 2.47 45.6 t.lxx O.11* 0.5515 1.0%0 1.1639 1.6810 3 1.08 ?4.1 1.o37 0.0716 0.2364 0.6520 1.u148 1.1427 1.5974 2.2422
+
W 19.9 41.1 1.001 0.0165 0.2498 0.c685 1.w16 1.w48 20 16.8 35.5 1.002 0.0095 0. IJ252 0.0416 0.0485 1.W1O 1.0032 1.0053
15 13.9 39.6 1.002 0.0303 0.0894 0.1224 1.0v56 1.0163 15 11.7 33.4 1.006 o.wf?6 v.c461 0.0738 0.0852 1.0037 1.0115 1.0188
10 2.82 37.1 1,005 0.0572 0.1653 0.2280 1.0213 1. 1%08 10 6.36 29.8 1.017 0.0392 0. LY365 0.1352 0.1555 1.0158 1.0473 1.0762
.3 8 6.79 35.7 1.0G9 0.0744 0.2165 0.3c41 1.0382 1.1089 5 6 4,53 27.7 1.c30 0.0545 0.1132 0.1787 0.2073 1.0305 1.0900 1,1450
6 0.76 33.9 1.016 0.0975 0.2941 0.4314 1.0720 $.2086 6 2.56 25.1 1.059 0.07”79 0.1516 0.2527 0.3021 1.0638 1.1876 1.3064
5 3,73 32o9 1.023 0.1119 0.35200.5380 1,1022 1.3014 5 1.64 23.4 1.088 0.W48 0.1784 0.3179 0.3929 1.0970 1.2878 1.4787
4 ,2.76 31.6 1.034 0.1287 0.4351 13.711u 1.150$ 1.4574 4 .820 21,3 1.144 0.1181 0.2149 0.4336 0.5692 1.157o 1.4759 1,8195

3 1.79 30.1 1.055 0.1481 0.5706 1.c443 1.2337 1.7551 3 .235 18.3 1.274 0.1567 0.2708 0.7094 1.0471 1.2886 1.9200 2.7o32

Jtttlm
— —.
20 19.0 30.0 1.UC4 0.02u2 u.u487 0.0634 1.0019 7 #x149 20 15.1 23.4 1.o11 0.0160 0.0259 0.0377 0,042 1.0016 1.0037 1.W55
15 12.9 !X.4 1.008 0.0377 0.0871 0.1124 1, 00+58 1,0168 15 9. 8s 21.1 1.027 0.0326 0.0464 0.0659 0.0725 1.0c61 1.0137 1.0198

10 7.80 25,8 1.023 0.0735 0.16w 0.2066 1.0266 1.0639 ,6 Iv 4,67 17.0 1. M’7 v.u748 0.0827 0.1179 0.1288 1.0280 1.0591 1.0836

ELE1.ua.
1.0485 5 2.6g 14.5 1.160 0.1112 0.1034 0.1544 0.1698
8 5.74 24.2 1.039 0.0980 0. J2c91 0.2738 1.1158

Ill 6 3.7o 22.2 l.o73 0,1333 0,2843 0.3870 1.0943 1.2267 6 1.02 11.4 1.348 (3,1796 0.1268 0.2202 0.2491
—

mm
5 2.71 2f.1 1.1c6 0.1575 13.3U16 0.4837 1.1369 1.3336
n-8
4 1.75 j9.6 1.163 0.1829 0.42720.6466 1.21s35 1.5216
1.0025 I .0051 20 1-.1 58.7 1.WO 0.0036 IJ.0154 u. 0336 LI. U479 1. W(I4 1.0019 1. VU43 1.0036
15 11.6 19.0 1,030 0.0511 0.0805 0,c989 1.w28 1.0176 15 13.0 %.7 1.WO 0.0070 0.0286 0.0605 0.0850 1 .(XJ14 1. W69 1.0154 1.0199
.5 10 6.36 16.0 1.086 IJ.1053 0.1434 0.1776 1,0358 1.0683 10 7.95 53.2 1.001 0.0144 0.0545 0.1123 LI.157 1 .fn%l 1.0283 1.0622 1. 07%
8 11.29 14.3 1.148 0.1465 0.1835 0.2326 1.0671 1.1265 4 8 j.93 51.1 1. W2 0.0195 0.0715 0.1478 0.!2111 1.0116 1.0533 1.1171 1.1508
6 2.32 12.1 1.293 0.2150 0.2424 0.3266 1.1375 1.2576 6 3.+! 48.6 1.003 0.0265 0.0947 0. 2U35 0.3072 1.0237 1.1v81 1.2416 1.3144
5 2.97 47.0 1.005 U.U31O 0.1”95 0.2472 lJ .3953 1.0352 1.1613 1.3682 1.4853

n-6 4 2.C3 46.6 1.008 (1.0362 0.1275 0.3148 u. 5575 1.0546 1.2539 1. 6u23 1,8121

20 19.5 72. o l.oo0 0,cYY75 0.0325 0.0605 1.0008 ! .W35 1. 0U54

i=M---Hl 3 1.1? 42.5
43.1
1.014
1.001
0.0422
U.005710.
0.1498 0.4392
01m10.033210.
0.9372
OWOl
1.0912 1. 43B4
11.00 W[l.0W311.00K11
2.1211 2.5798
:U057

H
?5 lh.5 70.1 1.OW 0.0138 0.0587 u.1085 1.W27 1.0120 1.01/36 1.002 U.0113 lJ.033u 0.0595 0.077

Mw
15 11.7 40.8 1.0023 1.0085 1.0165 1.0204
10 9.50 67.5 1.OW 0.W260 0.1031 0.2027 1.0105 1.@52 1.0703 10 6.62 38.0 1. W7 0.0242 0.0634 0.1099 IJ.1418 1.0101 1. U359 1.c#5 1.0839
.2 8 7.49 66.0 1.OW .0336 0.1401 0.2705 1.0189 1.0812 1.1270 5 8 4.60 35.7 1.013 0.0338 u.u836 0.1450 ().1828 1.0198 1. C691 1.1318 1.1618
6 5.49 64.1 1.cW 0,c432 (3.185’3 0.3829 1.0356 1.1553 1.2465 6 2.65 31.9 I. u27 11.w83 U.lllg o. 2“24 “.2785 1. fJ421 1.1457 1. 2&o 1.3493
5 4.49 63.fJ 1.001 0.c488 0.21570.4759 1.0503 1.2233 1.3599 5 1.74 30.1 1.040 0.0585 “.1309 “.2504 L), 36471 1.0644 1.2242 1. 443* 1.5558
4 3.49 61.7 1.001 O;0~7 0.2551 0.6249 1.0733 1.3351 1.5546
I 4 1.9°3127.8\I.~1 10. 0VOI0.15Wlu.33u9 ]0.53Wl [email protected] .761~11.97%
3 2.50 60.0 1.002 0,0603 0,3098 0.9059 1.1121 1.5400 1.9368 3 , ‘jti 24.7 1.124 IJ.09s5 0.lij;8 “.;;1; 0.9853 1,1890 1.6934 2.5549 3.@31
—
20 1 .9 53.4 1.000 0.w89 0.0334 0.0583 1.0009 1.w36 ,1.0055 20 1;.1 29,5 1 .“0(% “.”1”3 “.”2”” “.”3,6 “.0387 ,.w,, l.”W~ 1.””42 l.W~
15 13.9 51,8 1.(3w 0.0165 v.06c4 0.1c41 1.0032 1.012611.0188 15 9.95 27. IJ 1 .c114 0.0212 0.0363 o.cl%l 0.0672 1.0042 1.01u9 1.0181 t.o?ll
10 8.95 49.0 1.C+M 0.0317 0.1118 0.1936 1.0127 1.0481 1.0717 10 4. ~~, 22.6 1.047 “.”492 0.c&j7 O.IUW “.1z04 ,.”, % I.”- 1.”781 $.wv9
m--t-t-t
.3 8 6.83 47.4 1.002 0.2414 0.1456 0.25’79 1.0230 1.0872 1.1303 68 2.’[1 20.0 1. U87 u. 0730 “.Ll@82 lJ.134, “.,&j, ,.”@, ,.”%4 1.15% 1.181’2

L. I_LL
6 4.81 43.5 1.004 0.0542 0.1940 0.3656 1.0442 1.1690 1.2560 6 1.U” 17.7 1.186 0.1152 “.j125 0.,892 “. 236.3 j.”y7 ,.215” 1.3517 1.4129
5 3.82 44.3 1.W6 0. L!619 0.2278 0.4561 1.0632 1.2456 1.3773 5 . ‘,77 )4.6 1.294 0.1508 “.1252 “. 236 0.3162 I.IU@ 1.344u 1.5747 1.6816
4 2. !33 42.8 1.o1o 0.0705 0.2728 0.6038 1.W35 1.3745 1.5895 4 .306 12.1 1.499 “.2024 “.,31U “. 3262 u. 4776 , .2&” , .5~,~ 2.”28~ e.243b

k%l%ka-
3 1.87 41.0 1.016

17
—
~ 18.1 39.7 1.001
15 1~.o 37.9 1.003 0.0215 0.1%21 0. W74 1.W41 1.0135 1.0191

10 7,87 34.9 1.0043 0.c424 0.1153 0.1797 1.016’7 1.0526 1.0738

.4 33.2 1.014 0.0566 U.1507 0.2387 1.0308 1.0966 1.1359
8 s.63
31.1 1.026 0,0765 0.2027 0.3387 1.C607 1.1915 1.2723
6 3.82
5 2.M 29.7 1.038 0.8895 0.2401 c.4248 1.0886 1,2832 1.4078

w%El---
4 1.88 28.1 f.059 0.1W3 0.2922 0.5702

Mltlt
3 1.W 26.1 1.099 0.1253 0.3739 0.8714
—

15 11.7 26.4 1.013 0.0308 0.c623 0.0880

10 6.48 23.2 1.038 0.0640 0.1143 0.1597 !.0241 1.0591 1.0773
1.5 1.c458 1.11u8 1.145u

e
B s.63 21.2 1.w56 0.W* 0.1489 0.2106

E&El--
6 2.46 18.8 1.129 0.1272 0.1977 0.2982
5 1.% 17.3 1.199 ~.t559 0.2332 ~.3757
—

1.6 15 So& 16.1 1.031 0.03w 0.0561 0.0749

10 b.68 12.4 1.159 0.1136 0.w36 0.1303
8 ?.59 10.3 1.286 0.1676 0.1114 0.1682
—
86 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, FEBRUARY 1968

— RESULTS OF SYNTHESIS es WR

---- PREVIOU9

FOR 8 BRANCHES
RESULT ‘z]
to 3 =NO OF BRANCHES

1.10 5 z MI DBAND COUPLING (dB]

t?
VSWR

1.30

1,25
t NO. OF BRANCHES = 3 456 789

‘:& 20 30

>= ,2
40 50

DIRE CTIVITY
60

(d B)
70

~I!2
[.20

Fig. 5. VS WR versus direetivity curves for Chebysbev

couplers for bandwidth of 1.2.
1.15

1.10

1.05

1.00
1.0 1.1 1.2 1,3 1,4 .-
Age 10 20 30 40 50 60

OIRECTIVITY (dB)
m
~.ao

G=’4
Fig. 3. VSWR as a function of frequency for Butterworth
branch-guide couplers.
Fig. 6. VSWR versus direetivity curves for Chebyshev
couplers for bandwidth of 1.4.

VSWR
@=No. OF BRANCHES
i Vsw?
3= MIOBA?A0 COUPLIN9 (dB)

1(
I 04
A
— RESULTS OF SYNTHESIS
,2 - @ @ @=NIJ OF BRANCHES
S = MI DBAND COUPLING (dB]
‘--PREVIOUS RESULTS@ FOR [0 6
5 BRANCHES
1.021- \
I )46 q2.9 II 10 .9
‘\
.<&=$
hbb ~~ , ,@ ,0 15

I I I 20
I 00 :\kL, 20 20
20 30 40 50 60 70 60 10
DIRECTIVITY (dB) 10 13 20 25 30
&=l,l
OIRECTIVITY (dB)
Ag* &., G
xg2
Fig. 4. VS WR versus direetivity curves for Chebyshev
couplers for bandwidth of 1.1, Fig. 7. VSWR versus direetivity curves for Chebyshev
couplers for bandwidth of 1.6.

DISCUSSION OF COMPUTED RESULTS

ity shows similar maximally flat behavior. The performance
Results have been computed for a large number of cases of a 3 dB coupler having 8 branch guides analyzed from
of practical interest, and the element value are presented in previously published results [’1 is also shown for comparison
Tables I and II (pp. 84 and 85). This information is mainly as the dotted line in Fig. 3. It has been shown by analysis
restricted to cases where the directivity of the couplers is that the reason that the latter is inferior to the present result
greater than 20 dB, It has been established that the results for n= 8 is that the zeros of the F+ functions [see (17)] of
obtained from the synthesis procedure result in a consider- this previous result are clustered about the midband fre-
able improvement over existing designs. It is seen also that quency rather than being exactly coincident with it.
the new information includes specifications which have pre- The performance of Chebyshev designs is displayed in
viously not been presented, and it is shown possible to design Figs. 4 through 7 for a large number of center frequency
branch-guide couplers with good VSWR and high directiv- coupling values and for various numbers of branch guides.
ity over large bandwidths, i.e., of the order of one octave or Each figure represents a case of constant bandwidth. The
more. curves show the maximum ripple VS WR and the minimum
Performance curves for 3 dB couplers having 3 to 9 ripple directivity for a given number of branch guides as the
branches and Butterworth characteristic are shown in Fig. center frequency coupling is varied. The latter values are
3. In these curves VSWR has been plotted, and the directiv- indicated in decibels by the labelled points on the curves.
LEVY AND LIND : SYNTHESIS OF DIRECTIONAL COUPLERS 87

These figures present an overall picture of performance and EXPERIMENTAL RESULTS

are useful in selecting a design for a given specification. Several designs for Butterworth and Chebyshev char-
The element values for either Butterworth or Chebyshev acteristics have been constructed both in X-band waveguide
designs are obtained, respectively, from Tables I and II. It and in stripline. Due account was taken of corrections to be
is evident that it is now feasible to design branch-guide applied to the branch and main arm lengths and impedances
couplers to give excellent VSWR and directivity over much due to discontinuities, as explained in previous publica-
broader bandwidths than hitherto thought possible. A com- tions. [’l, [21In the waveguide designs, each coupler was con-
parison with previous designs[zl is shown as the dotted curve structed in the form of a metal block with slots forming the
in Fig. 4. It is seen that here an even greater improvement is branch guides. The block was inserted between two wave-
attained than in the Butterworth case. guides having a section of one side of the their broad walls,
equal to the length of the block, removed. The arrangement
EXPLANATION OF THB TABLES suffered from practical limitations, e.g., the difficulty clf ob-
Table I gives the values of branch and main line immit- taining good contacts, and the residual VSWR measured
tances for Butterworth specifications for 3 to 9 branch using a block without slots was approximately 1.04. Results
guides with couplings from 2 to 20 dB. Since the couplers for couplers constructed by milled block or similar manu-
are symmetrical, the immittance values tabulated are given facturing techniques might be expected to be somewhat
as far as the center of the coupler. The bandwidth factor superior to those presented here.
&O/&Z corresponding to points where the directivity falls A sketch of a 3-branch Butterworth 10 dB waveguide
to 20 dB is given for each case and also the band-edge cou- coupler is given in Fig. 8, and the measured results are shown
pling [B.E. COUP (dB)] and the VSWR at these points. in comparison with the theoretical curves in Fig. 9. Table I
The performance is symmetrical with respect to electrical shows that the directivity falls to 20 dB for XgO/&Z= 1.17,
length (3= 21rl/& on each side of the center guide wave- where the band-edge coupling is 9.58 dB and the VSWR is
length &O, and the bandwidth is expressed as the ratio 1.03. With a midband frequency of 9.5 GHz, the band-edge
19/00= ~gO/ho2, where A,z is the guide wavelength at the high- frequencies are approximately 8.7 and 10.4 GHz, as in-
frequency end of the band. dicated in Fig. 9. The results show good agreement with
For design purposes generally, it would be more useful to theory within the practical limitations. The error of 0,4 dB
work in terms of the ratio of the guide wavelengths at the in midband coupling was not repeated in other designs,, e.g.,
band edges, i.e., the ratio a 4-branch 6 dB coupler for Butterworth response (Fig. 10)
gave almost perfect agreement between measured and
X.1 0, theoretical coupling, and good agreement for directivity and
—.—.
VSWR (Fig. 11). The worst deviation from theory occurs at
Xgz 01
the high-frequency end of the band where the ratio of the
Since we have waveguide dimensions to the guide wavelength is becoming
large. Deviations caused by the use of a narrowband ap-
02–6’ 0=00-01 (35) proximation for the equivalent circuit of the T-junction dis-
continuities may become considerable here.
it follows that the required ratio X~l/&g is related to the A 3-branch coupler constructed in stripline for a Butter-
ratio k~@~l as given in the tables by worth characteristic having 3 dB midband coupling is shown
in Fig. 12. Originally designed for a midband frequency of
Ago 1.5 GHz, it actually centered at 1.525 GHz, and the theo-
retical curves of Fig. 13 are drawn accordingly. A similar
02 Xgl c
—=_ . (36) coupler designed for Chebyshev performance also gave ex-
01 AQ2 Ago “ cellent agreement between theory and experiment for VSWR,
2–C
9’ directivity, and midband coupling, but showed similar
coupling deviations at the band edges. In these designs there
Thus X,O/k,,= 1.4 represents a band-edge guide wavelength is a large difference in width between the inner and outer
ratio hQ1/A~Z of 2.333, i.e., greater than one octave. The branch lines, and the length correction is different from the
figures for A,l/&, corresponding to A,O/k,, equal to 1.5 and two cases, However, a mean length correction is taken in
1,6 are 3.00 and 4.00, respectively. Similarly, Table 11 practice. This and other factors could be responsible for
gives design information for Chebyshev characteristics for deviations from strict theoretical performances at the band
n= 3 to 8 branches. Here the bandwidth factor k~@OZ is a edges.
parameter, and the maximum VSWR ripple, the worst An example of a broadband 5-branch waveguide coupler
directivity ripple within the band, and the band-edge designed for Chebyshev VSWR and directivity characteris-
coupling are given. tics is shown in Figs. 14 and 15. Here the midband coupling
Some of the theoretical results presented in the tables is 8 dB, and for &o/hoz = 1.4 Table II gives a directivity
would not yield useful designs for a variety of possible ripple of 24.2 dB, a VSWR ripple level of 1.039, and a
reasons, e.g., difficult line admittance values or excessive coupling which drops to 5.’74 dB at the band edges. The
coupling variation across the band, but it is thought that slightly nonequal ripple performance is clearly indicated.
the tables cover most cases of practical interest. The experimental results agree quite closely with the theory.
88 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, FEBRUARY 1968

.460 TYP 4i7 TYP

–3-E!+-++ COUPLING (dB)

’75 80 8.5 90 9.5 100 10.5

FREQUENCY (GHz)

Fig. 8. Cross section of Butterworth 8-branch 10 dB coupler in DIR ECTIWTY (d B)

WR90 (WG16) waveguide.

COUPLING (dB)

i
10.0
S@
90 - 7.5 8,0 S.5 9.0 9,5 10.0 10.5

FREQUENCY (G Hz)
8.0
80 8.5 9.0 9.5 10.0 I 0.5 11.0
FREQUENCY (G Hz)
VSWR
OIRECTIVITY (dB)
0----0 EXPERIMENT
),
50 — THEORY

u
40 ,\

/’
30

* .-0-” ‘b
20

10
8,0 8.5 9.0 9.5 I 0.0 10.5 11.0
75 80 8.5 9.0 95 100 105
FREQUENCY (G Hz)

FREQUENCY (G Hz)

VSWR c----a EXperiment

Fig. 11. Theoretical curves and experimental results for the
— THEORY
coupler of Fig. 10.
1.15 ?
I

~.091 TYP
.296 TYP
11 I
J-
so S,5 9.0 9.5 I 0.0 I 0.5 11.0 .478TYP
FREQUENCY (GHz)

T
Fig. 9. Theoretical curves and experimental results for the ..?~r
coupler of Fig. 8.
1“

1-
1.922 TYP

.400TYP

.029 TYP

* ‘ ,;”:Y;
.307

11 1 1
Fig. 12. Plan view of Butterworth 3-branch 3 dB coupler in air-
.0s4 TYP spaced stripline, with ground-plane spacing 0.3125 inch and strip
thickness 0.0625 inch.
“393‘Yp-+--l 1--+”323
Fig. 10. Cross section of Butterworth 4-branch 6 dB coupler in
WR90 (WG16) waveguide.
LEVY AND LIND : SYNTHESIS OF DIRECTIONAL COUPLERS 89

COUPLING (dB)
A t t t
.400 TYP .419TYP .446TYP
3.5

-3-EY=-+-EEE
‘%
‘-x.
3.0
2 e,
2.5

1.30 1.40 L50 1,60 1,70 1,80

DIRECTIVITY (dB) FREQuENCY (GHzI
4 - F.117
.040 TYP -- .
. --.088
TYP
.410 TYP +-+ ~+.377 TYP

Fig. 14. Cross section of Chebyshev 5-branch 8 dB coupler in WR90

(WG16) waveguide.

1.30 1.40 1.50 1,60 [.70 1.80

FREQuENCY (Q Hz)

VSWR
+–-+ Experiment CONCLUSIONS
125 , — THEORY P
1!
/
The new synthesis technique for branch guide directional
L20 !’
119
\
couplers gives superior results compared with previous ap-
1.15 proximate methods both for Butterworth and Chebyshev
1.10
VSWR and directivity characteristics. The coupling shows
monotonic variation with frequency for the symmetrical
1,05
couplers described, in common with previous theories. The
1.00
1,30 1.40 1.50 1,60 I.m 1.80 results for practical branch-guide couplers designed using
FREQUENCY (Q Hz) the new theory give good agreement with the computed re-
sults, and the design of couplers having a bandwidth greater
Fig. 13. Theoretical curves and experimental results for the coupler
of Fig. 12. than one octave appears to be feasible. The design pro-
cedure is simplified by the publication of tables of branch
COUPLING (dB)
guide and main line immittance values covering most eases
of practical interest.
Branch-guide couplers in stripline are widely used in
microwave miniature circuits, and the broadband de~igns
:~ given here could be utilized for such applications.
5 11 , , 1,
7.5 8.0 8.5 9,0 9.5 100 I 0.5 II o
FREQUENCY (G Hz)
ACKNOWLEDGMENT
OIRECTIVITY (dB)

The authors wish to thank Dr. J. D. Rhodes for discus-

50 t
sions and valuable comments during the course of this
40 F-,
,/ ,A
,\ work.
/’ \ --d’ ,P
30 ‘.= --- ‘\..%-
p<242 lLJ\, /.9’
REFERENCES
20

,o,~ 85 9.0 9.5 100 10.5 I 1.0

[11R. Levy, “Directional couplers, “ in Advances in Microwaves,
1, L. Young, Ed. New York: Academic Press, 1966, pp. 155-161.
vol.

FREQUENCY (GHz) [’1 L. Young, “Synchronous branch-guide directional couplers for

low and high power applications,” IRE Trans. Microwave Tfzeory and
VSWR c----a EXPERIMENT Techniques, vol. MTT-10, pp. 459–475, November 1962.
[3] J. Reed and G. J. Wheeler, “A method of analysis Of Symmetri-
A _ THEORY
cal four-port net works,” IRE Trans. Micro wave Theory and Techniques,
VOI. MT’T-4, pp. 246–252, October 1956.
[41H. J. Riblet, “The application of a new class of equal-ripple func-
tions to some familiar transmission-line problems,” IEEE Trans.
7.5 8.0 85 9.0 9.5 100 105 110 Microwave Theory and Techniques, vol. MTT-12, pp. 415-421, July
FREQUENCY (GHz) 1964,
[51H. J. Carlin and W, Kohler, “Direct synthesis of band-pass trans-
Fig. 15. Theoretical curves and experimental results for the coupler mission line structures,” IEEE Trans. Microwave Theory and Tech-
of Fig. 14. rziques, vol. MTT- 13, pp. 283-297, May 1965.