update:19apr21 the dropouts were
created by the sband portion of the signal overdriving the
system.
Intro
Looking at the 12meter square law detected
signal on the oscilloscope in the control room showed that
there were dropouts in the total power signal. On 12apr21 high
time resolution data was taken with the mock spectrometer. The
setup was:
- telescope was sitting at az=259.49, el=41.28 (
ra=22:31:xx dec: 4:37:44.5)
- xband receiver
- Mock spectrometer used to record the data.
- 7 172 MHz bands centered at:
- 8132,8296,8460,8624,8788,8952,9116
- Each band spaced by 164 MHz
- 2 pols, 1024 channels/pol, 168 Khz channel width
- spectra dumped once a millisecond
- 300 seconds of data was taken.
- The total power was computed for each 172MHz bw,
1millisecond sample. PolA and PolB were kept separate.
- The spectra of the total power was also computed for each
172 MHz band (3000000 points). The computation was:
- spc=abs(fft(tpband - mean(tpband)) keeping the 1st half
(since it was real data.). I didn't square the result.
Plotting the results
The plots show the total power
time series and the total power spectra (.ps) (.pdf)
- Page 1: Total power vs time for the 8460 MHz band.
- black is polA, red is polB
- The total power was divided by the median value to make
the vertical scale units Tsys.
- PolB is a lot noisier than polA
- An offset of +/- .2 was added for display.
- Top: 1st second of data
- Bottom:last second of 300 second scan
- The 200 ms up,down shifting is probably external rfi.
- The green dotted lines show a comb with a spacing
of .08 seconds (12.5 Hz)
- every position goes negative
- every other position then overshoots positive. this
has a spacing of .16 seconds (6.25 Hz).
- The blue dashed line has a spacing of .64 Hz (8 times
the .08).
- the bottom plot shows that the periodicity remain stable
within our 1/300 second resolution.
- Page 2: overplot the total power vs time for the 1st
second for all 7 172MHz bands.
- Top: polA
- Bottom:polB
- This shows that the dropouts were occurring across the
enter 1GHz bandwidth.
- Page 3: spectra of the total power.
- Black is polA, red is polB
- I've offset polB for display.
- Top: 0 to 500 Hz bw
- There is a finer comb around multiples of 100 Hz. The
frequency spacing corresponds to the .64 second spacing
in the time domain.
- bottom: blowup 0 to 130 Hz.
- The green dashed lines are a 6.25 Hz comb (from the
.16 second positive going spikes)
- the blue dashed lines are 6.25*2=12.5 Hz (from
the .08 second spacing)