After the fact
- 09apr21: two fiber patch cables in the pedestal
(postamp chassis to fiber tx) were replaced.They were
causing the levels to shift.PolB was much worse than polA.
This probably caused the slow variation in polB after the
galaxy passed (see total power stability).
Intro
The xband receiver on the p12m telescope
was monitored for 4 hours on 24mar21. The setup was:
- 16:00 to 20:00 AST
- telescope sitting at Az=344.1, El=52.4 deg
- 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, 8192 channels/pol, 21Khz channel width
- spectra dumped once a second
- The program looped 4 times. Within each loop the
proram did:
- 11 300 second scans
- 1 calon,calOff (30sec on, 30 sec off)
- There was a gap of about 30 seconds around 17:00 (bug in
my program).
- There was no rain during this observation.
- Each 300 second scan was processed separately
- An rms/chan was computed for each set of 300 1 second
spectra.
- The expected value of this should be deltaT/T=
1/sqrt(bw*integrationTime) =.007
- Rfi was rejected by doing a linear fit to the rms and
excluding channels with residuals > 3sigma
- 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.
Total power stability
Total power vs time and right ascension:
The total power vs time and ra is
plotted for the entire 4 hour session.
The
total power vs time and ra for the 4 hours (.pdf)
- Each band was plotted separately.
- The total power was scaled by the median value in 19 to 20
hours . This will be called Tsys.
- Each 172 Band was plotted with an offset for display. 8132
at the bottom, 9116 band at the top
- the dashed blue line shows the tsys level for each band
(defined as the average value 19 to 20 hours)
- Black is polA, red is PolB
- The large variation around 17:00 is the galactic anti
center passing through the beam (the dec was 52.72 deg).
- Page 1: plots the total power vs Hour of day (AST)
- Page 2: plots the total power vs Right
ascension. Since the telescope was sitting, the sky
would drift through the beam
- the declination remained constant at 53.72 deg.
- Looking at the haslam et al sky survey showed that the
increase at 17 to 18 hours ast (or 3.5 to 4.5 hours RA) is
the galactic anticenter).
Things to take note of:
- If things were stable (and no external sources), both
traces should lie close to the blue line.
- You can see a change of about 10 to 15% Tsys from
16:00 to 20:00. This could be an electronic gain change from
temperature.
- The first and last bands seem to have more offsets. This
is probably from the band being only partially filled with
rf power.
- For some reason polB had a much larger variation 4 to 6
hours than polA.
- There are a number of negative going spikes.
- Saturation might cause this. It is probably an
electronic gain dropout.
Blowup of each 300 second scan 19 to 20 hrs ast:
A higher resolution plot was made by
plotting the total power vs time and the spectrum of the
total power for each 300 second scan from 19 to 20 hrs ast
(this was after the galaxy had passed through the beam.
Total power vs time and
spectra of total power for 300 second scans (.pdf)
- The spectra was computed as: spcTpA= abs(fft(tpA-
mean(tpa)))^2
- black is polA , red is polB
- Top: total power vs time (300 seconds).
- the 7 bands are over plotted with offsets for display
(bottom low band, top high band).
- Bottom: the spectrum of the total power for the 300
seconds of total power
- the 7 bands are again plotted with offsets for display
- the vertical scale is in arbitrary units ( scaled if for
display).
Things to take note of:
- You can see the negative going spikes
- The first and last scan seen to have more junk, but this
may be from 1/2 the band not having rf power.
- PolB consistently tends to have more oscillation/jumps
than polA
processing: x101/210324/xbmontp.pro
Using the cal to compute Tsys across
the band.
A cal is installed via the coupler before
the 1st amp. 30 second calON, calOffs were done every 60
minutes during the 4 hour period.
We have not yet done a hot/cold load test to get the actual
temperature of the cal across the band. Felix used the diode,
cable loss, and coupler properties to come up withy 45Kelvins.
This was used (across the entire band) when computing Tsys.
When we get more accurate measurement of the cal, we'll redo
the tsys measurements.
Combing the 7 freq bands
There were 7 172 mhz frequency bands
covering xband. They were spaced by 164 MHz . To compute tsys:
- for each 172 MHz band compute a single value to
convert spectrometer Counts to Kelvins
- compute the average calon,caloff spectra for the 30
seconds of data
- Compute the total power from each average spectra
- exclude the edges of the bandpass.
- also exclude 45% of the first and last band (where the
rf power did not fill the band)l
- tpCalDeflection=tpCalOn - tpCalOff
- CalsPerTsys=tpCalOff/(tpCalDeflection) (since calOff
should be Tsys).
- convertSpcCntsToK= CalInKelvin/(calsPerTsys)
- The spcCalOff(f) was then converted to kelvins after
removing the IF bandpass
- spcTsys(f) = (spcCalOn -
spcCalOff)/spcCalOff * convertSpcCntsToK
- the tsys vs freq was then interpolated to a fixed
spacing (25KHz) covering the entire band 8050 to 9200 MHz
- in the spectra overlap region, 45% of the overlap
(3.6MHz) was ignored (to get rid of the filter dropoff).
The plot shows Tsys vs
frequency for the 4 calon,offs taken during the 4 hour
period (.pdf):
- --> remember this is using 45 K for the cal across the
entire band..
- black is polA, red is polB
- the vertical scale is kelvins
- The blue plots at the bottom show the where the 172MHz
mock bandpasses are located.
- You can see that the first and last band are cutoff (by
the p12m xband rf filter).
- There are 4 traces corresponding to the 4 calon/off
measurements spaced by 1 hour in time.
- they were measured around 17.2,18.2,19.2,20.2 hours ast
- The top trace was taken when the galactic anti center was
in the beam (so the Tsys increased by the power from the
galaxy).
- The bottom 3 traces were taken after the galaxy.
- Unfortunately we don't have a calonoff around 16:00 hrs
(before the galaxy). If we did then it would probably show
the same tsys as later in the evening (since the
elecronice gain affects the sky and cal values the same.)
Things to take note of:
- we saw the galaxy :)
- Outside the galaxy Tsys was 100 to 140K (using the
Tcal=45K)
- there is a large ripple in the spectra.
- at lower frequency it is worse in polB
- at higher frequency the two pols are closer together.
- the purple dotted lines fall close to the peaks of the
ripple. It is spaced by 64.11 MHz
- a standing wave ripple (from a reflection) with a
64.11 MHz period would be a distance of 2.34m in air.
processing: x101/210324/xbmoncal.pro
Looking at the rmsbychannel for the
cal off scans.
The rms by freq channel was computed for
each of the 300 second scans after 19:00 hours (the galaxy had
passed). The computed value was then divided by the average of
the frequency channel so the radiometer equation could be
used:
- deltaT/T = 1./sqrt(channelBandwidth*IntegrationTm)
- The channel bandwidth was 172.032/8192. = 21KHz.
- the integration time for each record was 1 sec.
- the ExpectedRms is .0069 (Tsys)
The plots show the rms
by channel for each 300 scan scan after 19:00 hours (.pdf)
- each frame is the rms/mean by channel vs freq
for a 300 second scan.
- Black is polA , red is polB
- The blue line is the expected rms: .0069
- the white gaps every 164 MHz were areas that were
excluded because of the bandpass falloff.
- the two pols are close to the expected values. polB is
noisier than polA.
- the spikes around 8200 are most likely rfi.