Intro

    Pulsar observations at xband are being done with the 12meter telescope. The pulsars being used and the setups being used are:

• B0329+54 (200mJy at lband)
• sample time: 500 usecs
• 3600 second scan
• B0833-45 (1000mJy at lband) ..vela
• sample time: 200 usecs
• 300 second scans
• The mock spectrometers record the data with setups of:
• 7x172.032 MHz bands
• cfr each band:
  
• 8219.84 8360.96 8502.08 8643.20 8784.32 8925.44 9066.56 MHz.
• bands are spaced by 141.2 MHz
• channel width: 1.344 MHz
• 105 channels between band centers
• Prior to each pulsar observation a winking cal scan is done
• hardware winking cal (25hz)
• 60 seconds of data
• cal data taken while tracking the pulsar.

Processing the data:

• 6% of the band is ignored when processing the cal and psr total power.
• The cal scan:
• input 60 seconds of cal data
• compute total power for each sample
• compute the mean value of the total power for polA, and polB
• This should be between the calon and the cal off
• compute the median value for all points above the mean value (this is the  cal on value)
• compute the median value fro all poitns below the mean value (this is the cal off value (tsys))
• This lets us compute the calON, calOff without worrying about the samples taken during the cal transistion (the alternative would be to fold the data at the cal period)
• Compute the conversion factor from mock spectrometer counts to cal units
• calDeflMockCnts=(calOn - calOff)
• sclToCAl= 1/calDeflectionMockCnts
• The scaling to cal units is then  applied the psr data so the result is in cal units.
• we currently do not know the cal value vs freq in degK (we're having a difficult time getting the resources to put the absorber in front of the horn on the 12meter)
• When we finally get the cal vs freq in degK, we will be able to convert things to kelvins
• The cal diodes are temperate controlled so the values will be stable with time.
• we can't compare the values from different freq bands (since the cal changes with freq)
• But we can compare the values from the same band vs time.
  
• PSR data
• Input the data averaging the spectra by row (this is about 1 seconds)
• compute the total power for each 1 second averaged spectra
• Plot the total power vs time in cal units. This is Tsys in calUnits (typically around 3.)
  
• If more than one pulsar is taken on a day
• Compute the average total power by freq band for each pulsar
• plot the ratio of this total powers.
• this will show if there is anything strange in the Tsys values for the two obs.
• say clouds or instabilities.
• Normalize the total power vs time for each band to the its median value and plot it vs time.
• this blows up the vertical scale and lets us see variations on the order of a fraction of a %.

---

Plotting the results:

    The table shows plots of the results, and  a link describing what happened on that observation

date(ast)	psr	plots	detected	notes	moreInfo
211217	B0329+54	.ps   .pdf	no	clouds/rain? polB drifts 1% after this compression from sband rf ben's writeup	separate writeup.
211217	B0833-45		yes
211231	B0329+54	.ps   .pdf	yes	rain 1/2 way through jumps 25%, slow recovery	ben's plots .pdf
220103				50 ohm load  now on sband rf
220103	B0329+54	.ps   .pdf	no	polA shows freq dependence of  tp, 2% jump midway thru. polB looks ok	ben's plots .pdf
220105	B0329+54		no	rain at end	more info/plots
220106am	B0833-45		some bands	2 frq Bands had low power, probably incorrect setup
220108am	B0833-45	.ps  .pdf	yes	totPwr vs time looks ok.  polA 8360 band has some variation (+/= 1%)	ben's plots .pdf

---

more info

---

220105/06  B0329+54, B0833-45

==> this was a bug in the setfreq routine. it was occasionally selecting the wrong 500 Mhz filter for if1->if2. fixed 9sep21

• Note: B038-45 was taken 2am ast 220106 but the plot labels it as 220105 (that was pst:).
  
• total power vs time  (.ps) (.pdf)
• Page 1:
  
• Top: B0329+54
• slight rise in power at end of 3600 sec observation. Probably clouds/rain.
• pulsar was not detected by ben.
• Middle: B0833-45
• taken a few hours after B0329+54.
  
• B0833-45 band1 polA,b and band 7 (polB) offscale ..
  
• Bottom: Tsys ratio B0329+54 / B0833-45
  
• B0833-45: Tsys in freq band 1 polA,b and freq band 7 (polB) very high (power levels very low).
• Page 2: variation of Tsys with time
• The total Power was normalized to the median value in each band,pol
• top: B0329+54
  
• polA colors vary with band
• some variation of total power with freq at beginning
• polB: color purple all bands
• At the end of the observation the Tsys increases by about 4%. probably clouds or rain.
• Bottom: B0833-45
• polA colors vary with band.
• Black is the first band that had vary low power levels
• polB: purple
• two of the bands that diverge were from the first and last freq bands that had low power.
• Band passes for B0833-45 (.ps) (.pdf)
• Plot the average bandpasses  vs frequency for the 60 second winking cal scan
• Units are spectrometer counts.
• Black : polA
• red: PolB
• 1st band (8219 MHz band) polA,B and last band (polB) and low power levels.
  
• i've seen this kind of problem previously (13dec21). first and last bands had problems.
  
•  Back on 13dec22 i found:
• After the ifsetup xb,
• The 325 MHz output levels for the first and last bands were low in power (in some pols) looking at the front panel outputs.
  
•  the lites in the if/lo panels were all correct
• The front panel outputs of the 1500MHz filters were ok
• The mixer band select lites were selected correctly.
  
• I checked the synthesizer outputs:
• i switched to port 4 (external outputs) and the freq and amplitude of the synth looked ok for the bad bands.
  
• I don't think the synth port switch is bad since its 1 output  feeds both polA and polB (splitter after the switch). The pulsar obs showed polA ok and polB bad  for the last band..
  
• So either 1 of the 2  switches into the mixer or 1 of the two switched out of the mixer could be having problems.
• I saw this problem happen often if i did an ifsetup and then followed it with a setfreq request .
• By putting a delay of 1 second between the ifsetup and setfreq the problem would go away (on 13dec21).
  
• After the 05jan22 problem i upped the delay at the end of the ifsetup from  1 to 3 seconds to see if this would help.
  
• Need to do more trouble shooting on why the switching is not always occurring correctly.
• The level to move the switch may be low, or the switch may be taking too long to switch
  

---