Transcript Characterization of Instrumental Phase Stability on the SMA Interferometer

Characterization of Instrumental Phase Stability
on the SMA Interferometer
D. Y. Kuboa, T. R. Hunterb, R. D. Christensenc, P. I. Yamaguchic
aAcademia
Sinica, Institute for Astronomy & Astrophysics, Hilo, HI
bHarvard-Smithsonian Center for Astrophysics, Cambridge, MA
cSmithsonian Submillimeter Array, Hilo, HI
Abstract
1ST LO
SYNTH.
2
Hz)
+/- 5
G
(SKY
P
D
ORTEL
CRYOSTAT
FIBER-A
1550 nm (-0.5 dB)
1
IF-2 OUT (5 GHz)
1
VARIABLE ATTENUATOR IS SERVOED ONLY ONCE AT BEGINNING OF TRACK TO OPTIMIZE LO DRIVE TO YIG OSCILLATOR.
2
VARIABLE ATTENUATOR IS SERVOED ONLY ONCE TO OPTIMIZE IF DRIVE TO FIBER OPTIC TRANSMITTER (FOTx).
3
VARIABLE ATTENUATOR IS SERVOED ONLY ONCE TO OPTIMIZE LO DRIVE TO YIG OSCILLATOR.
4
VARIABLE ATTENUATOR IS SERVOED ONLY ONCE TO OPTIMIZE YIG LO DRIVE TO GUNN PLL.
5
VARIABLE ATTENUATOR IS SERVOED ONLY ONCE TO OPTIMIZE IF DRIVE TO MIXERS.
6
VARIABLE ATTENUATOR ARE SERVOED CONSTANTLY TO OPTIMIZE DRIVE TO SAMPLERS.
7
SECONDARY PHASE ROTATORS REMOVE RESIDUAL FREQUENCY OFFSETS FOR FRINGE STOPPING.
LO-2
+4 dBm
(nom)
1310 nm
from MRG
YIG-2 PLL
1310 nm
~109/200 MHz
1550 nm
(-0.5 dB)
+4 dBm
(nom)
1310 nm
TO RCVR-B SYSTEM
(NOT SHOWN)
IF-1 OUT (5 GHz)
LO-1
GPS
Reference
(1 PER ANTENNA IF
REQUIRED)
52 MHz
PLS
52
MHz
DDS
Direct Digital
Synthesizer
-3 dBm
(~9 + 100
(MHz)
SUM
x10 &
100 MHz
PLL
10 MHz
PLL
x2
FOTx
LASERTRON
coarse
tune
MRG YIG-1
PLL
fwd
NOTES:
FORx
5.5 to 8.5 GHz
FOTx
100 MHz
-30 dBm
-30 dBm
1310 nm
2
ORTEL
(-10.5 dB)
WDM
FOTx
ANT-8
CSO
JCMT
-5 dBm
(nom)
1550 nm
CONTROL BLDG
2ND STAGE
AMP
FIBER-C
1310 nm
ANTENNA
2ND STAGE
AMP
FORx
WDM
ISOLATOR
.
.
.
3
DIPLEXER
ISOLATOR
ANT-1
SUM
fwd
YIG-1
OSC.
+5 dBm (nom)
5+/-1.25
GHz
10-WAY PD
ORTEL
course
tune
200 MHz
BPF
52 MHz CLOCK
-7 dBm
(nom)
fine
tune
109 MHz
5+/-1.25
GHz
FAN
OUT
F(s)
4
HEMT
BPF
CHUNKS 21-24
(HARDWARE SHOWN
ONLY FOR RCVR-A)
ISOLATOR
/
SIS BIAS
2ND STAGE
(14K)
CHUNKS 17-20
x4
OPTICAL SPLITTER
horn
Fin - Fo =
200 MHz
DC
error
TO RCVR-B YIG-2
+20 dBm
(max)
RECEIVER SET A
(4 INSERTS, 1
SHOWN)
SHOWN)
CHUNKS 13-16
6
ANTENNA IF/LO Box
LENS
SIS
CHUNKS 9-12
IF
5
(oval)
3RD STAGE RECEIVER SET B (4
INSERTS, NONE
(4K)
4 CHUNKS / BLOCK
Fc = 153 MHz
(82 MHz EACH)
2nd DOWN
CONVERTER
FINA
L LO
(flat)
6
CHUNKS 4-8
+2 dBm
nom
4-Wy Sw
TO CORRELATOR
Ambient Load
Calibration
.
.
.
8-BITS
6
6 BLOCKS
Fc = 1 GHz
(328 MHz EACH)
4-Wy PD
DEMUX
TOTAL OF 6 X 4 = 24
CHUNKS PER
RECEIVER IF
16-BITS
DIFFERENTIAL
ECL / CHUNK
52 MHz
course
tune servo
2
.
.
.
SE TO
DIFF
CHUNKS 1-4
.
.
.
208 MHz
fixed multiplier
FORx
(1 PER RECEIVER INSERT
REQUIRED)
5
SAMPLER
2-BITS
1
TO OTHER GUNN
PLL & MULTIPLIER
ASSY'S
(oval)
4
GUNN PLL &
MULTIPLIER ASSY
TO OTHER GUNN
PLL & MULTIPLIER
ASSY'S
GUNN
OSC.
Fo
xN
~109 MHz REFERENCE
fwd
(flat)
3
Digital PLL Assy
fine
tune
7
6-WAY PD
DC
error
F(s)
1ST STAGE
(65K)
The Submillimeter Array (SMA) is a collaborative project of the Smithsonian
Astrophysical Observatory (SAO) and the Academia Sinica Institute of
Astronomy & Astrophysics (ASIAA) of Taiwan. The array consists of eight sixmeter diameter antennas with receivers operating from 180 to 700 GHz and a
digital correlator with 2 GHz of bandwidth. Located on Mauna Kea, Hawaii, the
primary elements of the SMA interferometer can be reconfigured across 24 pads
which provide baselines ranging from 8 to 500 meters. Each antenna cryostat
assembly houses up to 8 receiver inserts consisting of low-noise superconducting
(SIS) mixers. The inserts can be used in pairs for increased bandwidth and for
polarimetry observations. Local oscillators (LOs) are used in each antenna to
provide heterodyne mixing from the sky frequency to the 5 GHz intermediate
frequency (IF). A system block diagram for the IF/LO system is provided in Fig
1. Photos are provided in Fig 2. Instrumental phase stability measurements are
discussed herein.
109 MHz
error
~109 MHz
+11 dBm
(nom)
4-WAY PD
MIRROR 1
xM
6.2 to 8.3 GHz LO
harmonic
mixer
2ND LO
SYNTH.
1st DOWN
CONVERTER
fwd = LO
rvrs = {Fgunn - (M x LO)} = ~109 MHz
LO
POLARIZATION
SPLITTER
Introduction
3508
3836
6164
6492
6820
LOCATED ON OPTICS CAGE
200 MHz REFERENCE
Atmospheric water vapor causes significant undesired phase fluctuations for the
SMA interferometer, particularly in its highest frequency observing band of 690
GHz. One proposed solution to this atmospheric effect is to observe
simultaneously at two separate frequency bands of 230 and 690 GHz. Although
the phase fluctuations have a smaller magnitude at the lower frequency, they can
be measured more accurately and on shorter timescales due to the greater
sensitivity of the array to celestial point source calibrators at this frequency. In
theory, we can measure the atmospheric phase fluctuations in the 230 GHz band,
scale them appropriately with frequency, and apply them to the data in 690 band
during the post-observation calibration process. The ultimate limit to this
atmospheric phase calibration scheme will be set by the instrumental phase
stability of the IF and LO systems. We describe the methodology and initial
results of the phase stability characterization of the IF and LO systems.
1276
1194
806
724 MHz
3180 MHz
x6, 7 or 8
MULTIPLIER 6, 7, or 8 GHz
x10
List of Acronyms:
BPF = Band Pass Filter
FORx = Fiber Optic Receiver
FOTx = Fiber Optic Transmitter
MRG = Master Reference Generator
NOM = Nominal
PD = Power Divider
PLL = Phase Lock Loop
WDM = Wave Division Multiplexer
HP 8644B
fine
tune
F(s)
0.2 to 500 MHz
Fig 1 Smithsonian Submillimeter Array RF System Block Diagram
1. IF Functional Description and Impact on Phase Stability
3. Experimental Tests – Stability of Antenna YIG oscillator
The IF path which carries the 4 to 6 GHz signal begins within the antenna cryostat just after the SIS mixer (lower
left of Fig 1). This signal is leveled then optically modulated for transmission over fiber to the main control
building. The received optical signal is demodulated to electrical, leveled to accommodate for optical loss
variations, then passes through the 1st down conversion which subdivides the IF signal into six blocks. A 2nd down
conversion further subdivides the IF into 24 chunks each centered at 153 MHz and with a bandwidth of 82 MHz
(total bandwidth of 1968 MHz). Each of these 24 chunks are fine leveled to accommodate for channel slope and
sky variations then digitized using 2-bit samplers operating at 208 Msps (upper right of Fig 1). Phase movements or
drifts of the IF signal do not scale with LO frequency. I.e., a 1 degree drift of the IF-1 signal relative to the IF-2
signal appears as a 1 degree offset between the 230 GHz and 690 GHz signal. We have shown that the IF drift is
much less than 1 degree and is therefore negligible.
The long term phase stability of the antenna YIG-1 LO was monitored by sending the signal back to the control
building using the IF-2 channel (not shown in the figure). This returned antenna YIG-1 LO was mixed with the MRG
YIG-1 LO to produce a difference frequency of 200 MHz, which in turn was phase compared to the MRG 200 MHz
system reference using an HP 8508A vector voltmeter. The data shown in Fig 3 were taken during an actual science
track at 1 sample/second and was window averaged over 60 seconds to remove fast phase variations. These data
reveals a high level of stability of 0.10 degrees peak-to-peak and 0.025 degrees RMS over the first six hours. It is
significant to note that there is little or no correlation of the phase to the azimuth position which implies that the
azimuth fiber tension assembly is functioning as designed. Translated to the final LO frequency of 225 GHz the phase
number becomes 3.2 and 0.8 degrees peak-to-peak and RMS, respectively. Assuming a similar stability for the
antenna YIG-2 LO this number translated to 685 GHz becomes 8.4 and 2.1 degrees peak-to-peak and RMS,
respectively.
8.0
280
7.8
Antenna-4 AZ position, degrees
Antenna-4 YIG-1 round trip phase, degrees
340
7.6
7.4
7.2
220
160
100
40
-20
-80
7.0
0
60
120
180
240
300
360
420
-140
0
60
Elapsed minutes since 11:57:53 PM HST, April 16, 2006
120
180
240
300
360
420
Elapsed minutes since 11:57:53 PM HST, April 16, 2006
Fig 3 Antenna 4 YIG-1 round trip phase stability data (left) and corresponding azimuth position (right). YIG
tuning was changed at ~390 minutes. Peak phase drift seen over 6 hours was just over 0.10 degrees. RMS value
over this same interval was 0.025 degrees.
Fig 2 Antenna cabin photos. Left , cryostat with optics cage. Gunn oscillator and multiplier assemblies are
located on the top of the optics cage. Water vapor radiometer optics are just visible on upper left. Center, IF/LO
electronics showing IF-1/IF-2 (left) and YIG-1/YIG-2 (right) assemblies. Right, azimuth fiber optic tension
assembly. The yellow fibers pass through the center of the azimuth encoder and are pulled taught by a spring
loaded assembly.
2. LO Functional Description and Impact on Phase Stability
The LO system begins with a 10 MHz crystal oscillator phase locked to a GPS reference. A set of ~109 MHz and 200
MHz references are generated then optically modulated for transmission over fiber to the antenna. A common tunable
LO (5.5 – 8.5 GHz) denoted as MRG YIG-1 PLL in the lower right of Fig 1 is optically modulated and power divided
to each of the 8 antennas. Each antenna receives these LOs on fibers A and C and are demodulated to electrical then
leveled to accommodate for optical loss variations. The antenna YIG-1 PLL phase locks to the 200 MHz and LO-1 to
produce an output at either LO-1 +/- 200 MHz. Following the YIG-1 output is a harmonic mixer which produces
multiple harmonics (M) and mixes them with the Gunn oscillator output operating in the 80 to 120 GHz band. The
Gunn output is followed by a final fixed multiplier whose value is denoted as N. Table 1 provides example LO
frequencies for some typical tuning frequencies. It becomes obvious from the table that a small movement in the
antenna YIG phase translates to a non-trivial phase movement at the final LO. For example, a 1.0 degree phase
movement at the antenna YIG output translates to 32, 48, and 84 degrees at the final LO for 230, 345, and 690 GHz,
respectively, which is clearly too large because it will correspond to significant errors in the position measurement of
astronomical sources and to a reduction of the image fidelity.
Table 1. Example LO frequencies for 230, 345, and 690 GHz.
MRG YIG
6.825 GHz
6.877 GHz
7.947 GHz
Antenna YIG
7.025 GHz
7.077 GHz
8.147 GHz
M
16
16
14
Gunn
112.509 GHz
113.341 GHz
114.167 GHz
N
2
3
6
Final LO
225.018 GHz
340.023 GHz
685.002 GHz
M*N
32
48
84
Since the antenna YIG-1 and YIG-2 LOs are locked to
MRG LO-1 and LO-2 located in the control building, it
is necessary to characterize the phase stability between
these MRG LOs. This was accomplished by locking
MRG LO-1 to 7.000 GHz and LO-2 to 7.200 GHz and
mixing the two to produce a difference of 200 MHz.
This 200 MHz difference was then phase compared to
the MRG 200 MHz system reference. Fig 4 shows the
phase stability plot over a limit period of only one hour.
This test was conducted in the morning after a science
track and was affected by the air conditioning
compressor turning on at approximately 45 minutes.
Nevertheless a variation 0.30 degrees peak-to-peak and
0.058 degrees RMS was seen during the first 45
minutes. Assuming that each MRG YIG is drifting by
equal amounts we assign an RMS stability of 0.041
degrees to each.
MRG2 - MRG1 phase, degrees
4. Experimental Tests – Stability between MRG YIG-1 and MRG YIG-2
11.2
11.1
11.0
10.9
10.8
10.7
10.6
10.5
10.4
10.3
10.2
0
15
30
45
60
Time, seconds since ~8:10AM, April 22, 2006
Fig 4 Stability of MRG YIG-1 verses MRG YIG-2.
Note air conditioning compressor turn on at ~45
minutes.
5. Preliminary Conclusions
These early results have produced an antenna YIG stability of 0.025 degrees RMS. Adding to this number in RSS
(root sum squared) fashion 0.058 degrees RMS from the MRG YIG produces a stability of 0.063 degrees RMS.
Translating up to the final LOs of 225 and 685 GHz produces 2.02 and 5.29 degrees RMS of phase variation,
respectively. The phase variation seen between these two LOs and therefore between the two observed signals is
5.66 degrees RMS which corresponds to a contribution to the reduction in sensitivity of less than 1%.
Further tests are required to characterized the performance of the remaining 7 antennas. We would ultimately be
interested in staring at a 230 GHz and 690 GHz beacon over several hours to establish the phase stability of the
overall system working together.