R 6915641
R 6915641
R 6915641
MICROSTRIP CIRCULATORS
by
DOCTOR OF PHILOSOPHY
Approved:
1969
ii
TABLE OF CONTENTS
Page
INTRODUCTION 1
LITERATURE REVIEW 3
Microstrip Parameters 36
Ferrite Parameters 47
CONCLUSIONS 68
BIBLIOGRAPHY ' 70
ACKNOWLEDGMENTS 75
APPENDIX 76
LIST OF SYMBOLS
vector laplacian
K off-diagonal component of g
N demagnetizing factor
g tensor permeability
y on-diagonal component of ^
i
vi
LIST OF FIGURES
Page
INTRODUCTION
from the incident port to the next adjacent port and isolating
interest here.
port junction was made and it was shown that not all ports of
transmission line.
LITERATURE REVIEW
In 1955, Carlin (14) was the first to show that any loss
diameter and the applied bias field were varied until all
culator action.
lator design in his brief reports. Chait and Curry (16) pub
solved.
disk acts as a mirror image of the other and only one disk
need be considered.
to assure circulation.
Ferrite
Cylin'ders
electric field were made and it was shown that most of the
they will travel around the disk with the same propagation
longer travel around the disk with the same velocity and the
most often varied are the external magnetic field and the fer
netic field there are two frequencies which will show circula
In his two papers Bosma laid much of the ground work for
textbooks (17, 19, 26, 29, 32, 34, 40, 41, 42, 44).
For the TEM mode to exist the top conductor, of finite width,
bottom conductor is more than three times the width of the top
il
ferent configurations.
is only good for wide strips, but gives good results in this
microstrip design.
mann (23) gave results that generally agree with those of (6).
Kostrize (31), on the other hand was concerned with the use
microstrip.
papers which give easy to use charts for obtaining the shape
(27, 35).
away.
in Figures 1 and 3.
Conductor
\Pattern
Ferrite
Cylinder
Dielectric
Substrate
(la) 2^ E"
(lb) V X H = -jwEo EE
(Ic) V • E = 0
(Id) V ' gH = 0
where
y -j< 0
(le) g = ^ y 0
0 0 1
2
(2a) (spz p 3p "*• p29cj)2 k'iEg 0
where
2 2
(2b) = w^EoMoSf^—^-^)=
This equation has solutions, depending on the number of
= Jn'kp)(a+ns:"* + .
«P =
(4c) = 0
«P = -/
Uo ( j] )
Mb') /^5^
= j
Uo ( ^ )
+ a.,e-j"*(j^_,(kp, - (1 - I»}
24
{4c') = 0
(5c) Port #3 = 0 = 0
.(5d) Elsewhere = 0
(Gb) • 2Ji(kR)
^ - 5='
(») = i ((1 +
ditions
rite disk and the magnetic field vector is parallel to, and in
magnetic bias field 3^ and g will have the same value. Ports
II and III will see voltages which are 180° out of phase with
the input voltage and one half the value of the voltage at
port I.
have its own resonant frequency which will depend on the sat
will arrive at ports II and III out of phase and will add to
applied bias field is less than the field required for reso
device any port can be chosen as the input port with similar
tions.
(11) R = 1.84/w/ESoPoWeg2'
Fay and Comstock (22) give curves which are useful for obtain
ferrite radius.
the procedures used. These, for the most part, are based on
ber of tasks. For best results the purpose for which the
P
(12) Isol = 10 llogiop^ldb.,
in
f0 = 4000 MHz
6 = 20%
Microstrip Parameters
(13) Zo = ^
defined by
(14) C = ^
($2 - 0 1 )
Since microstrip will not support a pure TEM mode the char
characteristic impedance Zq by
I
(15) Z, = Zo
ing factor which takes into account the fact that the top con
tor loss, dielectric loss and radiation loss. The first two
ergy is in the region between the conductors and that TEM mode
(17) a = —^^
^ 2ZoW
38
(18) ^ /yTtanS
it was found that the smallest was the least lossy. This was
the ground plane and the holder of the single pole ceramic
tions.
40
flattened tab
^ 1/8 " long
#5
r^n \
\ .\
m
conductor
manner which has not yet been explained fully. The first
tor.
(19) = -kpj|(kpPle"i*
Hp = - 1 J,(kpP)e-]4
where k' =
and y = jwoe^^^eo.
(2°) ®p =
dependence.
-3pertirbea<4
(21) w - Wo =
of cylinder
gram.
could be matched.
Ferrite Parameters
met and after the microstrip parameters have been chosen the
tion 4nMg, the line width AH, and the disk geometry. These
regard.
ciated with the strip-line model have been adopted, some with
(22)
the internal field equal zero. This then gives the first in
(23) = N(4ïïMo), Mo = Mg
gauss.)
are common.
0
•f-1
4J
(TJ
•1-1
-AH*
thickness.
0 = phase angle of
using
(25) 01 = .
I
This allows the fractional frequency splitting, 6' to be
evaluated from
(27) ^ - 2.466'
(28) Waff =. 1 -
^eff
(29)
eff q + ygff(1 - q)
be a large correction.
(30) k = w v/Seffy'gffUoeo
modes '
ferrite with 47rMo - 450 gauss the disk diameter was calcu
be approximately
(32) = N(4ttMo).
magnet pole piece must have at least 20 per cent larger diam
bias field changes the admittance of the ferrite and the in
desirable region is between the low field loss region and the
netic bias.
values which gave minimum insertion loss was much more nar
technique.
!|
in
M
0)
m
f = 3700
= 4100
= 450
K = 390
340
•H
mittance plot.
the port match. For instance the conductor geometry need not
Ferrite Disk
Conductor
system. In the final run when the ports are matched this
VSWR
meter
detector
slotted
line
detectors
coupler
that port.
o
•H
Isol
VSWR
12
Ins. Loss
.1.10
3.75 4.25
Frequency GH
CONCLUSIONS
mission system.
outlined here.
BIBLIOGRAPHY
ACKNOWLEDGMENTS •
done while the author was working for Collins Radio Company.
for the many helpful discussions they held during this period.
\
76
APPENDIX
value. >
0 ^ 0
-3pertiîrbea"'î^p "
, _ top numerator
" ° 2 /// GE& • E°dT denominator
volume z z
k R 2ïï k R 2ïï
(2A) / / x{j'x(x))'d<j>dx + f / l[ji(x)]:d*dx}
k r 0 k r 0 ^
p P
kpR kpR
~ "Td^^ / x (j1(x))^dx + / i(ji(x))^dx}.
^ k r k r
P P
77
and
k^R k„R
"j • P P
(2A') -235{j( / x(jo(x)}^dx + / x(J2(x)]^dx} ,
k r . k r
P P
^ - 1]1.
2
k R
(6a ) 2e / / / | e l ^ d ^ d z p d p = ^ / x[ J i (x))^dx.
0 0 0 Z Kp g
78
(6A') [4 x 10-®)
2 ~ i T R[Jo
^ - - (k^R)Ji
• p (kpR))
+ (Ji(kpR))2}
50