NATIONAL RADIO ASTRONOMY OBSERVATORY
GREEN BANK, WEST VIRGINIA 24944
ELECTRONICS DIVISION INTERNAL REPORT No. 181
CRYOGENICALLY-COOLED C-BAND PIN DIODE SWITCH
GEORGE H. BEHRENS, JR,
OCTOBER 1977
NUMBER OF COPIES: 150
CRYOGENICALLY-COOLED C-BAND PIN DIODE SWITCH
George H. Behrens, Jr.
Introduction:
This report deals with a cryogenically-cooled, C-band, PIN diode switch
developed for radio astronomy applications. The various switching modes
available, principles of operation, operating characteristics and construction
details are discussed.
The switch was developed to satisfy the Dicke switching requirements for
a new law noise, dual band (25 cm and 6 cm), dual channel, cryogenic radiometer
under construction at NRAO. The design goal was to achieve a switch with mini-
mum loss, VSWR and noise contribution which could operate at cryogenic tempera-
tures over the frequency band from 4.5 - 5.1 GHz and provide the necessary
Dicke switching functions as dictated by the various types of astronomical
observations.
[1]The Dicke Radiometer:
In radio astronomy the signal levels received are of such law levels that
receiver gain instabilities can seriously limit the effectiveisensitivity of
the radiometer. This effect can be minimized by using the Dicke radiometer
as shown in Figure 1A. The receiver input is continuously switched between
the antenna feed horn and a comparison noise source at a frequency high enough
so that the gain has no time to change during one cycle.
The detected output V o of the Dicke radiometer is proportional to the
difference between the antenna noise temperature TA and the noise temperature
of the comparison noise source T :
V o = K (TA - Tc), (1)
[1] J. D. Kraus, Radio Astronomy. New York: McGraw-Hill, 1966.
2
and the minimum detectable signal is given by the following expression if
TA = T
C
AT
min = 2 T
RA/P.T(2)
where:
TR = receiver noise temperature, °K.
= predetection bandwidth, Hz.
T = integration time, sec.
In practice the comparison noise source is usually either a microwave
termination held at a fixed physical temperature or another antenna feed horn.
When a cryogenic receiver is used, the termination is usually held at the op-
erating temperature of the refrigerator (e.g., 20°K). Antenna temperatures
are usually around 20°K also, so TA T.21; T If there is a significant unbalance
C
between TA and TC' noise can be injected via a directional coupler, into the
colder channel to balance the system or the receiver gain can be reduced dur-
ing the higher temperature half of the switch cycle.
When a feed horn is used as the comparison noise source, it is usually
identical to the signal feed horn. Both horns are then offset laterally by
equal amounts from the focal point of the reflector insuring equal noise
temperatures for both feeds. This method has the advantage that fluctuations
in antenna temperature caused by atmospheric conditions are effectively can-
celled since the beams of both feeds see essentially the same area of the
atmosphere because their beams are usually only on the order of 2-3 half power
beam widths apart. However, when observations are made of extended sources,
it is advantageous to use the switch load mode of operation; otherwise, both
RF
IF
0SE
CTI
ON
SQU
ARE
LAW
DET
ECTO
R
TA
vIN=
K ( T
A ---
Tc
SYN
CH
RO
NO
US
DET
ECTO
RIN
TEG
RAT
OR
• .1■
1111111111w
wim
•■■■■
■■•)
AN
TEN
NA
FEE
DH
OR
N
CO
MPA
RIS
ON
NO
ISE
IS
OU
RC
E L
SW
ITC
H I
__D
RIV
ER
FIG
. lA
SIN
GLE
CH
AN
NEL
DIC
KE R
AD
IOM
ETER
INTE
GRA
TOR
ps..
..■4
1111
1
AD
DER
1V:D
EO•
AM
P
TRAN
SFE
RSW
ITCH
1.1■1.1
11110
SYN
CHRO
NO
US
DET
ECTO
R.1■
0411
10RF &
IF
SECT
ION
SIG
NAL
AN
T.
FEED
HO
RN
TA
Tc
SQU
ARE
LAW
DET
ECTO
R
SWIT
CHD
RIV
ER
SQU
ARE
LAW
DET
ECTO
R
SYN
CHRO
NO
US
DET
ECTO
R
VID
EOAM
Prim
emen
nem
s111011
CO
MPA
RIS
ON
AN
T.FE
ED
HO
RN
INTE
GRAT
OR
RF8II
FSE
CTIO
N
FIG
. 1B
DU
AL C
HAN
NEL D
ICKE R
ECEIV
ER U
SIN
G T
HE S
WIT
CH
BEAM
CO
NFIG
URATIO
N
beams will be on source and the output will be distorted. It is therefore
desirable in the design of a radiometer to incorporate a means to provide
both a beam switching mode and a load switching mode.
Dual Channel Operation:
In the single channel Dicke radiometer the signal power is observed only
half of the time. However, by switching the signal feed horn between two re-
ceivers and adding their outputs as shown in Figure 1B, the efficiency of the
radiometer is increased. It can be shown that the minimum detectable signal
noise temperature of a dual channel Dicke radiometer is reduced by a factor of
over a single channel radiometer. In order to achieve the necessary switch
action for a dual channel Dicke receiver using the beam switching mode a trans-
fer switch is required.
Dicke Switch Design Considerations:
One of the major disadvantages of a Dicke radiometer is the degradation
of system noise temperature due to the noise temperature of the Dicke switch.
The amount of noise added to the system by the loss mechanism of the switch is:
ATs
= (TP + T
R) - 1) (3)
where:
AT
•
noise added to system due to switch.
• physical temperature of switch.
= loss of switch.
TR = receiver input noise temperature.
11•11•11111,1
10
fD
rive
Mot
or, lo
ft'
hcin
eP
Gra
rnps
Cha
nnel
tipco
nver
ter
Ing1N
RC
EE
ME
NIE
MM
O:,
,,:,
,
$0
01
11
6::
.
Noi
seC
ontr
ibut
ion
Mea
sure
men
tsF
IG.
83A13038
CIV
O1
01
w39 /
w3g
Z
dad H
aLIM
S U
N3 IN
O8A
191,,d
C133d01
nis
ON
INill
3A110110N
1enis o
Nm
ini
3A1110V
ci VO
1100 SV
19Atig,
S4dVO
ssvdm
i dd
1:13111:1N
OIS
S3eiddnsS
IN 3
3AIIIT
MV
O
d IN V
Vd
01
830,180:4XU
ft
S 3-111GO
IN'M
S 3
001
0N
id 1.012dHSSN
O:18
aitild
01
00
39V3 1-1011PA
S
1104ei3dd00 "zo
03.1.V1d 0100
S4 dV
0 ON
1>4001930 Ad 2'2
01810313100
99
gacm
Anoo/n01
a J 3d01
afn01
8
Therefore, to minimize AT s , both the physical temperature T p and the loss L
of the switch should be kept to a minimum. In the 25/6 cm radiometer, -Co
minimize Tp the Dicke switch is mounted directly to the 20°K station of the
refrigerator as shown in Figure 2.
In planning the 25/6 cm radiometer, it was decided that a dual channel
Dicke radiometer be implemented. It was also decided that the Dicke switch,
besides providing the transfer switching function needed for dual channel op-
eration, should also incorporate a means to permit either load switching or
beam switching for either channel. Further, it was decided that the switch
be used to connect the L-band upconverters to the parametric amplifiers dur-
ing 25 cm operation. Dicke switching is not needed during 25 cm operation
because only line observations are performed and they are relatively un-
affected by gain instabilities.
TABLE 1
RF Paths thru Switch for Various Switching Modes
No. Switching Mode
Connections Made DuringSwitching Cycle
First HalfSwitching Cycle
Second HalfSwitching Cycle
,
I. Both Channels 2 4- A 1 4. ALoad Switching 5 ÷ B 3 ± B_
II. Both Channels 2 4- A 5 ± ABeam Switching 5 4. B 2 --->- B
III. Channel A Beam Switching 2 4- A 5 ÷ AChannel B Load Switching 5 4- B 3 .--''' B
IV. Channel A Load Switching 2 4- A 1 -÷ AChannel B Beam Switching 5 4. B 2 ÷ B
V.Both ChannelsLocked to Upconverters
6 4. A4 --->. B
No switching.
PIN
A D
IOD
ESW
ITCH
Para
mp
5
LOAD
TOM
IXER
Spa
ram
p
tro
4.5
— 4
,95
GH
zO
UTPU
T
f- L
—BAN
D.k
U/C
/\ip
conve
rters
Dua
l Mod
eA
1.0
ToT
rans
duce
r1.
40 G
Hz >
I
Ho V
ER
T,
IFE
ED
Pola
riza
tion
10.
HO
RIZ
e Po
lariz
atio
n
U/C
FEED
1>
A
4.4—
5,2
G H
zD
UAL
BEAM
FEED
FEED
LOAD A
L CRYO
GEN
IC D
EWAR jjl
Cor
n(Ene
nts
At 1
5°K
)
FIG
.4 B
lock
Dia
gram
Of T
he F
ront
End
Sec
tion
Of T
he 2
5/6c
m C
ryog
enic
Rad
iom
eter
10
The switch is a combination transfer switch and two SP4T switches which
provides various switching modes best described by referrring to Table 1 and
Figure 4. As shown in Table 1 there are five different switching modes avail-
able. The table shows the connections for the various modes for each half of
the switching cycle, e.g., in mode I (both channels load switching), Feed A
is connected to Paramp A and Feed B is connected to Paramp B during the first
half switching cycle. During the second half of the switching cycle, loads A
and B are connected to Paramps A and B, respectively.
Principles of Operation:
Referring to Figure 5, the operation of the switch in the Beam Switch Mode
can be explained as follows. Assume RF energy is propagating from Feeds A and
B towards junctions Jl an J2, respectively, and diodes D1 thru D8 are biased
as shown during the first half switching cycle. Next consider how the energy
from Feed A divides at Jl due to the impedance conditions at Jl as seen look-
ing towards JA and JB. Since D3 is forward biased, its low impedance (see Fig-
ure 6A) is transformed to a high impedance at Jl by the quarter wavelength
50 ohm transmission line according to the usual quarter wave transformer
equation:
.2/ (4)
ZINZo Zd
where:
0 characteristic impedance of line.
Zd = forward biased diode impedance.
However, the impedance at J1, as seen looking towards JA, is close to
50 ohms since D2 is reverse biased and presents little affect on the line
since its impedance is high (see Figure 6B). Also, diodes D1, D7 and D8 are
Z O
N c
° 1
44—
X/4
ZO
FF= L
O =
50S
/ -_-.
3-1
5012
ON
, O
FF
6cm
FEED
A
20°K
LOAD
A
25cm
FEED
25cm
FEED
HOR.
POL.
VER.
POL.
NO
TE:
1 )
Bia
s Con
ditio
ns
For
Sw
itch
ing D
iodes
Are
Encl
osed
By
Rec
tangle
s. B
ias
Conditio
ns
Of
Oth
er D
iodes
Rem
uin
s The
Sam
e Fo
r Both
Hal
ves
Of
Sw
itch
Cycl
e.
2)
Solid
And D
ott
ed R
ect
angle
s In
dic
ate
Bia
s Conditio
ns
During F
irst
And S
eco
nd H
alv
es
Of Sw
itch
Cycl
e,
Resp
ect
ively
.
3)
Solid
And D
ott
ed S
ignal Path
s In
dic
ate
Sig
nal Path
s D
uring F
irst
And S
eco
nd H
alv
es
Of Sw
itch
Cycl
e,
Resp
ect
ively
.
FEED
B6c
m
FIG
. 5
Sim
pli
fied
Sch
emat
ic O
f F
ront
End S
wit
ch O
per
atin
g I
n T
he
Bea
m S
wit
chin
g M
ode.
12
Ii I I
/ ii !
/
14‘' 3.8mm-01.- 3.8Mnrel L p Rp R p Lp 1 1,,e‘____4_4::,I I
150pH On On 150pH
Zo =50aRs 0.841
Zo= 50aL I 20pH
(a) Foward Bias
l it '
I 0 1
2 I I /1-2 1I.*. 4.4m m -01 14- 4.4mm-01I I l_p Rp Rp Lp
1 1
o 0-1-YY1 (YY1-----<) o200pH On OA 200pH
R 2 oicaL2CT T,14pF
0
( b) Reverse Bias
FIG.6 Room Temperature Equivalent Circuit OfHP5082-3141 PIN Diode
Zo= 501/ opH Zo= 50D,
Mode OOOOOOO .
1stHalf of sw. cycle
D1 .. FWD Fwd Fwd
Fwd FwdREV
Fwd FwdFwdREV
FWD REV FWD REV
Fwd Fwd FWD
D2 ..
D3 . •
D4 • • FWD Fwd Rod Fwd
D5 • • Fwd Fwd Fwd Fwd
Fwd FwdFWDREV FWD REV FWD REV FWD REVD6 • •
Fwd
Fwd Fwd Fwd
Fwd Fwd Fwd
Fwd
D7 • •
D8 • •
IVIII
1st 2nd
13
forward biased causing the impedance looking into each of these arms from JA
to be high. Therefore, all the energy entering Jl from Feed A is directed to
Paramp P.A. "A!' except for the relatively small losses absorbed in the high
impedance arms of junctions JA and Jl.
The energy arriving at J2 from Feed B can similarly be analyzed to show
that it is directed to P.A. "B". During the next half switching cycle the
bias conditions of D2, D3, D6 and D7 are reversed causing the RF energy to be
directed from Feed A and B to P.A. "B" and "A", respectively. The operation
of the switch in the other modes can be explained similarly by referring to
Table 2 which gives the bias conditions for the various switching modes.
TABLE 2
Bias Condition of Diodes for Different Switching Modes
Capital letters are used to indicate bias condition of diodes switching in that
particular mode. Bias conditions of diodes in a fixed bias state are italicized.
BIA
S 2
:CT1
C2
Sym
etri
c A
xes
Are
Sam
e A
sA
xis
Sho
wn
NO
TE:
BIA
S 8
All
Tra
nsm
issi
on L
ines
Are
Bal
ance
dS
oli
d D
iele
ctri
c S
trip
line
Const
ruct
edF
rom
lo
z. C
op
per
Cla
d D
uro
id 5
88
0
FIG
S?
Dia
gra
mati
c L
ay
ou
t O
f S
wit
ch
15
The complete schematic of the switch is shown in Figure 7. Table 4
gives the values of the various components and the manufacturers' part number.
The 50 ohm transmission line system used is balanced stripline con-
structed from photographically etched, copper clad, .0625" thick, teflon/
fiberglass. The ground planes are formed by the gold plated brass case (see
Figure 3). Channels 0.300" wide by .062" deep were milled in the switch case
to accommodate the stripline material. This type of construction was used to
maximize isolation between the various arms and achieve good mechanical sta-
bility. The ground plane spacing is 0.125" and the gold plated copper (1 oz)
center conductor is 0•100" wide for Zo = 50 ohms. A small lip was left along
the edge of all channels to improve mating surfaces and minimize radiation
leakage.
The switching diodes, HP 5082-3141, are silicon PIN diodes incased in a
50 ohm hermetic package which permits a continuous transition in the 50 strip-
line circuit. This stripline package concept, according to Hewlett-Packard,
overcomes the limitations in insertion loss, isolation and bandwidth that are
imposed by the package parasitics of other discrete devices.
The photograph in Figure 3 shows the physical layout of the various
components.
16
TABLE 4
Component Data
Components Value Function Manufacturer's P/N
C i 3.3 pF Blocking capacitor ATC-100A-3R3-P-50
C 2 56 pF Bypass capacitor ATC-100A-560-K-P-50
C 3 0.05" x .075" Capacitive tap ••••• •■■■•
L ___ Bias Injection coil Pionics 9 1/2 T-47
LPF ___ EMI suppression filter Erie #1250-003
Dl-D8 ___ PIN diode HP 5082-3141
Striplinematerial c = 2.23 R/T Duroid 5880
2, 1 Zo = 50 0 Transmission line56 = .27"
162 Zo = 50 0 Transmission line= .573
2, 3 Zo = 200 0 Inductive tuning stub ••••
= .700"
2'4 Zo = 200 Q Capacitive tuning stub •■•■■
56 = 0.195
2, 5 Zo = 41 Q Impedance transformer56 = .36"
Zo = 41 Q
Impedance transformer= .55"
1•11■1
FEED A
FEED BLOAD A
.. aft
.0•1111..4•■••
.0 ammo. • awrow■ II0
...................a • • IP
•■■
• • •
U/C A• asies:a •
• •
•••■ • ...40 ••• • r ..jr•.. • .• •
RECEIVER PASS BAND
1111■10
LOAD B FEED BU/C B
• • • 111, IP • • • a. I • • IP • •Er •
6---RECEIVER PASSBAND-1 FEED A
17
4.0 4.2 4.4 4,6 4.8 5,0 5.2 5.4 5.6 5.8 6.0
FREQ.— GHzFIG. 8 I NS E RTION LOSS PARAMP "A"—÷ INDIVIDUAL PORTS T= 300°K , I=20mA/DIODE
-0 0.5
.„„
°N.
111111...1
6.02,0- I I I I I I I I I
4.0 4.2 4.4 4.6 4.8 5.0 5.2 5,4 5.6 5.8FREQ.— GHz
FIG. 9 INSERTION LOSS PARAMP "B" INDIVIDUAL PORTS T=300°K, I=20mA/DIODE
_.-_—_—__. .......
LOAD A
I ....- ..10-17 • l'—f—dr.Z.
41\_ FEED Al
_
U /C A NZ-FEED B
•
18
1.5 -4.0
14—RECEIVER PASSBAND—olI1I I 1 I I I I I I I
4.2 4.4 4.6 4.8 5.0 5.2 5.4 5.6
FIG. 10 INSERTION LOSS PARAMP "A" INDIVIDUAL PORTST= 18°K, I= 150mA/DIODE
5.8 6.0
03
cocn0 0.5-J
0
cr I .0
cn
1.54.0
U/C BLOAD A
01111•1•11
1lo--RECEIVER PASSBAND---.1II
I I1 I I I 1 I I I 4.2 4.4 4.6 4.8 5.0 5.2 5.4 5.6 5.8 6.0
• dir . .... .... .....auld "I'd dr. ar.
r
— • FIE E D A
FEED B
FIG. II INSERTION LOSS PARAMP "B" INDIVIDUAL PORTST =18°K, I = 150mA / DIODE
19
TABLE 5
Measured Noise Contribution due to Dicke Switch
Feed A Feed B°K °K
Channel A
Continuum, 4.245-5.040 12.8 14.9
4.6 GHz, 3% bandwidth 10.2 13.3
4.75 Gllz, 3% bandwidth 15.2 16.9
4.870 GHz, 3% bandwidth 11.8 13.0
4.920 GHz, 3% bandwidth 13.0 10.0
Average .................. 12.55..... 13.5
Channel B
4.245 - 5.040 17.4 14.3
4.6 GHz, 3% bandwidth 13.7 9.6
4.75 GHz, 3% bandwidth 18.9 16.76
4.870 GHz, 3% bandwidth 17.8 12.2
4.920 GHz, 3% bandwidth 17.5 14.7
Average . • OOOOOOOOOO • 17.0 13.3
A summary of the measured switch characteristics are given in Tables 6 and 7.
20
TABLE 6
Switch Characteristics over 4.4 - 5.1 GHz Band
Channel A Channel B
Maximum insertion loss (dB) 0.75 0.75
Maximum noise contribution (°K) 16.9 18.9
Average noise contribution (°K) 13.0 15.2
Maximum VSWR 1.5:1 1.5:1
Minimum isolation (dB) 30 30
TABLE 7
Switch Characteristics over 4.0 - 6.0 GHz Band
Channel A Channel B
Maximum insertion loss (dB) 1.1 2.0
Noise contribution Not measured Not measured
Maximum VSWR < 2.0:1 < 2.0:1
Minimum isolation (dB) 30 30
21
Performance:
Measurements were made to determine switch insertion loss, isolation,
reflection coefficient and noise contribution at both 18°K and room tem-
perature conditions. Results of the insertion loss measurements are shown
in Figures 8 thru 11. At 18°K the switch has a maximum loss of 0.75 dB over
the operating band of the receiver. Isolation measurements, made in the 4.0-
6.0 GHz band, revealed greater than 30 dB isolation between ports. VSWR mea-
surements indicate a maximum VSWR of 1.5:1 in the pass band of the receiver
and a maximum of 2.0:1 from 4.0 - 6.0 GHz. Noise contribution due to the
switch was measured at various frequencies across the band and are as shown
in Table 5.
Discussion of Results:
The theoretical noise contribution AT of the switch can be calculated
from equation (3). Inserting the measured value of switch loss, 0.75 dB, and
an assumed physical temperature of 18°K (temperature at which refrigerator
was operating) into equation (3) gives a AT s = 7.8°K. However, the average
measured value of AT is 14.1°K (average of values given in Table 5) and is
6.3°K greater than that calculated by equation (3). This discrepancy can be
attributed to three known sources of error: (1) impedance mismatch, (2) shot
noise, and (3) uncertainty in the value of physical temperature used in equa-
tion (2). The temperature could be significantly higher than 18°K due to
diode junction heating caused by the bias current.
The first source of error, impedance mismatch, could account for 0.7°K
of the 6.3°K discrepancy. This comes about due to the termination on the par-
amp input circulator which can be considered an 18°K noise source. Assuming
22
the switch has a VSWR of 1.5:1, as measured, 0.7°K of the 18°K can be re-
flected at the switch and returned to the paramp as an increase in noise
temperature.
The remaining 5.6°K of noise can be attributed to short noise and
inaccuracy in the known physical temperature Tp , however, an analysis to
determine the magnitude of these two sources of error has not been made.