Intro
The xband receiver on the p12m telescope
was monitored for 23 hours on 01jun21. The setup was:
- we used the cal noise diode rather than the sky as the
input signal
- we disconnected the cables that connect to the lna amp
(up at the horn).
- the 30enr noise diode -> 6db splitter -> cables
disconnected above.
- So we were driving polA,polB with a 30-6= 24enr noise
source
- We need to measure the noise diode setup more
accurately.
- This will tell us if the wandering around of the signal
is from the if/lo or the lna's.
- take data :16:20:28 0121 to 11:09 02jun21 AST
- telescope sitting at Az=0 el=75.. although this
should make no difference since we were not looking at the
sky.
- xband receiver using 8050 to 9200 MHz
- 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, 4096 channels/pol, 42 KHz channel width
- spectra dumped once a second
- We used the same processing as when on the sky (even
though there wasn't any rfi).
- When computing the total power, frequency channels with
rfi were rejected:
- compute rms/mean for each freq channels.
- do a linear fit to the rms vs freq and reject any
channels with fit residuals > 3 sigma.
- the rfi rejection was not very helpful. We had 19
hours of data with the mean value varying with temp. The
rms/chan was large so clipping at 3 sigma left a lot of
rfi.
- the total power for each 1 second spectra was then
computed over the good channels (scaling by the channels
used).
- The first and last bands were only partially filled with
rf power. They were cutoff by the p12m rf bandpass filter.
In these cases only the part of the band with power was
used for the computation.
- The total power vs time for each frequency band was
normalized to the average of the first 1000 points
taken (so the units become Tsys).
Total power vs time, temp, and ra
The total power vs time for the
entire 23 hour session was plotted:
The
total power vs hour of day with temp (.ps)
(.pdf)
- The x axis is hours since the start of 01jun21 (so hr 30
is 6am 22may21)
- Top
- over plot total power vs hour of day for the 7
polA bands
- there were no jumps in the total power.
- Middle:
- over plot total power vs hour of day for the 7
polB bands.
- Bottom: pedestal temperature
- The pedestal temperature (behind the postamp chassis)
was recorded once a minute.
Total
power vs temperature (.ps) (.pdf)
- Top: median (over the 7 freq bands) total power vs
hour of day.
- The green curve is the temperature over plotted on the
curve (scaled to fit)
- Bottom: linear fit total power vs temp.
- solid lines: total power vs temp
- dashed lines: linear fit
- black polA, red polB
- The coef:
- tpA=3.71 - .0294*tempF
- tpB=3l76 - .0298*TempF
- The total power units are Tsys.
- So the electronic gain must be changing by 2.9 and 3.0
% per deg F
blowup total power vs time
total power vs time blowup
(.pdf)
- The total power vs time is plotted in 1 hour blocks.
- polA 1st and 3rd frames
- polB 2nd and 4th frames.
- The 7 bands for each pol are over plotted.
- Each 1 hour plot is normalized to unity.
- There are no total power jumps.
- There is no frequency dependent wandering of total power.
Summary:
- The electronic gain dependence with temperature :
- -2.9%/degF polA
- -3.0%/degF polB
- This variation is larger than what we had while on
the sky (2.8%,1.7%)
- The increase is probably caused by the temperature
variation of the noise diode (it is not temperature
compensated).
- There are no total power jumps
- the only variation in total power is correlated with the
temperature.
- This shows that the previous jumps and
variation of power with time is coming from the lna module
or feed.
processing: x101/210601/xbmontp.pro