Oscillation in lband wide polarization B
10sep01
While checking out the lband narrow baseline instabilities
i noticed a separate oscillation in the total power of lband wide. The
figures
show
the oscillations.
-
Figure 1 top is total power dumps of 25Mhz bw at 1 second intervals. The
four traces are 1375,1387,1400, and 1413. Black is polA and red is polB.
Each frequency has been offset by .01 for display purposes. The Y axis
is Tsys/Tsysavg -1. There is an oscillation of .5% Tsys in
polA and polB that has a period of about 100 seconds (the temperature dependent
oscillation of lbn was 10 minutes). The hybrid is being used for this observation.
The telescope was tracking blank sky.
-
Figure 1 bottom is the same configuration except that the hybrid is not
being used. Now there is only an oscillation in polB that is twice the
amplitude of that in the figure above. This clearly shows that the problem
is at or before the hybrid in channel B.
-
Figure 2 top is data taken in 17jul01 with lbw and the hybrid in. There
is lots of interference but the 100 second oscillation is clearly visible.
-
Figure 2 bottom goes back farther to data taken 05may01 with lbw and the
hybrid in. The 100 second oscillation may be there but is it not as obvious.
So it looks like we've had a 1% oscillation in polB
of lband wide for at least the last two months (probably longer)
without anyone noticing it. This probably occurred since most people use
lband narrow. Another problem is that the oscillation may be stable enough
so that people who only looked at on/off-1 might not have seen it (or at
least it would have been smaller).
We need to come up with some daily diagnostics that
can spot these problems sooner.
processing: x101/010908/lbwosc.pro
01apr02
The ripple
in lbw channel B looks like like it is still there. The data is from
a1520 doing an OH experiment at 1665. The plot is 2.5 Mhz total power.
The data are 1 second dumps for 300 seconds then a cal on,off. This is
then repeated. The 300 second are continuous in time and then there is
a jump till the next 300 seconds starts (the vertical jumps are the cals).
processing: x101/020401/lbwripple.pro
02may02
The ripple
in lbw polB was measured with the hybrid out at 1405 Mhz using a 25Mhz
total power band. The telescope was sitting still with the sky driftin
gby. 500 seconds of data was taken with the cables straight through, and
then 500 seconds of data was taken with the cables switched at the output
of the dewar. The 1 second total power measurements are plotted vs
time at P(t)/median(P) - 1.
The top figure shows the ripple in polB (red) versus time. The cables are
in the normal position.
The bottom figure has the ripple in the polA (white) as it enters the correlator.
The dewar output cables were switched so that the "polA' signal is now
coming from the polB amplifier in the dewar.
Since the oscillation switched with the cable switch,
the oscillation is coming from before the dewar output cables (either in
the dewar or the bias box).
processing:x101/020502/doit.pro
07may02
5
second cal on,offs where done tracking blank sky at 1405 Mhz using
lbw with the hybrid out. The cal was computed using the 390 Khz bandwidth
as :(calon-caloff)/caloff. In the plot, polA is white and polB is red.
The za changed from 14 to 8 degrees while this data was taken. A linear
fit was subtracted to get rid of the Tsys change vs za. The rms (in units
of Tsys) was:
| polA |
.0014 |
| polB |
.0021 |
| expected (sqrt(2)/sqrt(390e3*5) |
.0010 |
So polB is 50% worse than polA and pol A is about 50% higher than expected.
The ripple in polB has a period of about 100 seconds so the 5 second on,offs
could vary depending on which part of the cycle they were taken on. The
normal 10 second cal on,offs would have more of a problem.
processing: x101/020507/docal.pro
home_~phil