gain variation with temperature
15sep01
On 07sep01 blank sky was tracked rise to set
using the cband receiver. On 08sep01 another blank piece of sky was tracked
with the lband narrow and lband wide receivers. Both of these observations
were at night. Multiple 25 Mhz bandwidths (per receiver) were sampled at
1 second intervals using the correlator. A temperature sensor in the gregorian
turret room was sampling the air temperature every 10 seconds while the
measurements were being taken. The temperature sensor is situated in the
center of the room between the IF/LO racks.
The
figures show the total power (for each 25 Mhz band) versus time. 4
bands with two polarizations each were measured simultaneously. The processing
of the time series took 2500 second blocks at a time and computed
tp[i]/median(tp). A second order baseline was used to remove the zenith
angle dependence.
Figure 1 has 4 25 Mhz bands at cband. Pol A and PolB are plotted in the
same color. The temperature is varying by about .7 degF while the gain
changes by about .8%. This occurs in all frequency bands and both polarizations.
Figure 2 plots 4 25 Mhz bands using lband narrow. There are discrete jumps
that correlate with the temperature cycle (this is a separate problem..
see
A1459 problems). If you undo the discrete jumps then there is
a .6% gain variation per deg F. Again the variation is in all frequency
bands and both polarizations.
Figure 3 plots polA of lband wide with the hybrid out (pol B has a separate
100 second oscillation ..see
lbw 1% oscillation). The gain variation is .5% per .5 degF. It is in
all frequency bands. PolB probably has the same oscillation but it is difficult
to see with its 100 sec 1% oscillation.
All 3 receivers show a gain variation with the temperature
cycle. It ranges from .5% to 1% per degF.
The same cband variation had been measured back
in aug2000 .
The temperature dependence of lbw
was measured on 12jun01 after an sband transmitter run and no gain
variation with temperature was seen. The difference may be that the component
that was varying had not yet reached equilibrium with the temperature that
the temperature sensor was measuring.
lband narrow has a smaller variation than the other two. It is more difficult
to measure it because of the jumps.
This problem is a common problem with multiple receivers so it is probably
in the IF/LO. It is definitely something in the turret room since it correlates
with the turret room temperature. It has been there for at least a year.
It must be a component that is common to both polarizations. Things that
come to mind are:
-
The 1st lo synthesizer. Although the mixer output should not be sensitive
to a small change in the lo levels.
-
The zeeman box polarization switch. This is an active component that allows
rapid switching of the two polarizations. It's initial purpose was for
zeeman experiments (until the correlation method was implemented). It is
now only used for diagnostic purposes.
I've called this a 1% gain variation but i really don't know that. It could
be the noise power of a device that is changing.
processing: x101/010908/gaintempdepend.pro
home_~phil