links
Intro
setup
Processing the data
Plotting the results
Summary

---

Intro

      Experiments are being performed to see if we can monitor variable sources who have flux levels on the order of Janskies. The current sefd of the 12meter is about  4000 Jy (so this is going to be tough :). The new cooled wide band receiver will help since it's expected system temperature will improve by a factor of 2 or 3.
 
    The current variable source tests are using:

• ra,dec mapping , dec,ra mapping of the sources
• 1degx1deg (10 amin bm) uses 13 strips each 6 bms long.
• The actual integration on the source is on the order of 1/100 (13*6) .. or maybe a little better if you include the hpbw data.
  
• crosses on the source
• This uses az,el strips of 6bms.. You spend about 1/6 of the time on source (excluding the time to move from az to el strip).

    If we used on,off position switching then the on source integration time could be about 50%. Possible trouble with on.off position switching are:

• The off position (which tracks the same az,el as the on) will see a different portion of the sky. This may cause a small error when computing (on-off)
• This could become worse if you use a  longer on,off time.
• There is no pointing error measurement. It could cause the source flux to decrease if the error is appreciable.
• Doing crosses before, after the on, off could be used to correct for any pointing errors (if we can see the sources in a few crosses).

Problems with the current warm receiver:

    We've seen that the current  12meter (warm receiver) output level changes with temperature.

• more info:  11nov21: totpwr vs time and temp (26 hours of cal on,off)

    Since the cal diode is temperature stabilized, the 25Hz hardware winking cal  can be used to correct for the electronic gain variations with temperature.

• more info:
  
• 17jan22: correcting for gain variations with the winking cal
• 16mar22: correcting for gain variations with winking cal.. achievable rms/mean
• One drawback with the current (warm receiver) cal is that it is 20 to 30K. If both calOn and calOff are used then Tsys takes a hit  of 1/2 the cal temp.

The ra,dec mapping tests are using the hardware winking cal. The crosses are not using the winking cal.

Look at on,off position switching stability using the hardware winking.

   Some tests were run using  on,off position switching with the  hardware winking cal running during the on and off position. We wanted to see
if it was an acceptable alternative to the  radec mapping or cross patterns when monitoring total power from weak sources. It will also give us some insights into how good (or bad) the current warm receiver is. 

---

Setup:

• The xband receiver was used for the tests.
• 7 x 172.032 MHz bands were recorded.
• The band center freq were:
•  8219, 8363, 8505, 8647, 8789, 8931,9073 MHz
• 25 Hz hardware winking cal was run during the on,and off position  acquisitions.
• The mock spectrometer setup was:
• 172.032 MHz bw, 1024 channels, 2millisecond integrations, 16 bit spectra, polA,B recorded.
• 5 minute on,off position switching was done with the off position tracking the same az,el as the on position.
• 3C274 was used for the tests.
• chris salter has 26Jy for the xband flux
• jpl antenna calibration note has 46Jy at 8420Mhz.
  

---

Processing the data:

• The routine masgetscanhwc() was used to input the winking cal data for the on and off position scans
• masgethwcal() input the file, separated the calon,off, and integrate to 16 millisecond calon, 16millisecs caloff
• 1 sample at the start and end of the 20 ms duration was dropped  so 20ms calon -> 16ms integration calon.
• a 300 sec scan gave 300/.04ms = 7500 sample for calOn and 7500 for cal off.
  
• The rms/mean spectra from the cal on and off cycles were used to flag channels with rfi
• linear fits were done to the rms/mean spectra. Any freq channel with a residual  >3sigma was flagged as rfi.
• The average cal deflection spectra (casOnSpectra - calOffSpectra) was normalized and then used to flatten the calOn,caloff spectra. This removes the if/lo bandpass shape.
• the total power cal difference for each 40ms cal cycle was computed using the good freq channels.
• A 3rd order fit to the totPwr  cal deflection was then done to measure the electronic gain variation.
  
• The gain correction was applied to calOn,caloff totpwr time series and then it was converted to kelvins using the cal values.
• The average calon,caloff spectra was also computed after correcting for the electronic gain variation of each 16ms spectra.
• the resulting data for the onPos and off position was then used to plot some results (chkscani_onoffhwc.pro in x101/220803/)
  

---

Plotting the results:

    There are two sets of plots:

•  set 1: 7 pages (1 for each frequency band) for a scan (either position on or position off)
  
• black lines are polA, red lines are polB
  
• The data has been smoothed to 25 samples (.4 secs of data, 1 second duration)
  
• top: fractional change of cal deflection vs time
• The  calDeflection is CalOn-calOff for each 40ms cal cycle.
  
• This change is assumed to be the electronic gain variation (since the cal diode is temperature stabilized)
  
• The left most 300 seconds  has the CalDeflection for the On position data,
• the rightmost 300 seconds has the CalDeflection for the off position data.
• The green lines have the 3rd order fit to the calDeflection (for onposition, offposition)
• Note that the on position and off position scans have been normalized separately.
  
• 2nd frame: The calDeflection/3rdOrder fit is plotted for on source, off source.
• The fit will be used to correct to the on and off scan total powers.
  
• 3rd frame: fractional change in total power vs time for the calOff  data (cal on was ignored).
  
• black,red polA,B after the gain variation correction.
• dark,light blue (polA,B) before the correction.
• The black,red curves are not completely flattened by the correction.
• This will occur if the input power before the cal (horn,sky) is changing during the scan ( eg clouds)
• bottom frame: srcDeflection vs time  using the calOff data
• total power (Position On - position off)  smoothed to 25 cal cycles (.4 secs data).  
  
• The vertical scale in  degK
• The variation across the 300 seconds can come from the Position On,or  position Off data.
• black, red are polA,polB after the gain correction
• dark,light blue are polA,polB without the gain correction
  
• set 2: Source Deflection averaged over each band vs frequency for all patterns for the  day
• * is polA, square is polB
• Top frame is with the electronic gain correction
• bottom frame is without the electronic gain correction
• I used the mean cal deflection from the off position to convert to deg K.
• Normally you would have onPos, offPos, then calDeflection when not using a winking cal.
  
• The rms of the source deflection for the 300 secs is plotted as the error bars.
• I offset polB x axis by 4 MHz so it wouldn't overlay polA error bars.
  
• The error bars about each point are the 1 sigma deviation for the 300 second src Deflection.
  
• The table of observations:

• DATE	SRC	on,off patterns (first set of plots)	AvgSrcDeflVsCfr (2nd set of plots)
  220803	3C274	onoff_1.pdf onoff_2.pdf	AvgTpsrcDeflectionVscfr.pdf
  220819	3C274	onoff_1.pdf onoff_2.pdf	AvgTpsrcDeflectionVscfr.pdf

  
  

Notes on plots

• 220803,220819: total power low for 1st  minute of the first on position scan.
  
• The start of the on Position for the first pattern of the day has increasing gain for about 1 minute.
• this occurs in all freq bands
  
• The start of the off does not show this.
  
• After gain correction,  the total power In this one minute is decreasing with time.
  
• At first this looked like the electronic gain correction was being overdone.. but
  
• tsys could be changing (maybe  clouds).. although it is a bit suspicious that both days have it only at the start.
• There could be a pointing error at the start, so we are not exactly on the peak of the beam (not sure why the 2nd pattern didn't have it).
• the electronic gain correction assumes the cal diode power is constant. If it changes then any electronic gain correction  would over or under compensate the correction.
  
• The source deflections changes by about  .7 kelvins (avg source deflection 1.2 K) over the 300 sec on,off position.
  
• Not sure why it showed up in both days at the start.
• The telescope optical gain will change with elevation. This has not been considered since the elevation change during a scan is less than 3 deg in elevation.
  
• 220803:
  
• First set of plots:
• pat 1:
  
• Src deflection.
  
• PolA,B difference is large for the uncorrected data.
• polA,B difference small for the corrected data.
• Uncorrected poB source deflection becomes negative (--> Off source stronger than on source)
• Pat 2:
• uncorrected source deflection large value and large variation.
  
• 2nd set of plots:
• corrected values:
• polA,B values close
• max difference 1st, 2nd pattern ab out .3K at lowest and highest freq band.
• uncorrected values
•  PolA,B difference
• .3K pat 1, .5K pat 2 polA,B difference.
•  constant difference with freq for each pattern.
  
• Pattern 1 , 2 average values
  
• .75K average pattern 1
  
• 4.0K average pattern 2
  
• 220819:
  
• First set of plots:
• Pat1:
  
• there are two bumps in the total power off after correction, seen in all freq bands.
  
• This is probably clouds during the off scan
• Pat2:
  
• lots of variation in total power on and off position. probably clouds.
• The total power in the  off has 2 bumps. seen in all freq. probably clouds.
• 2nd set of plots:
• PolA,B difference
• corrected:
  
• pat 1: .1 K difference across all freq bands.
  
• pat 2:  difference increases to .2 K above 8600MHz
• Uncorrected:
• pat 1: .1 K , above 8600MHz increases to .35K
• pat2:  polA varies dramatically above 8400MHz. src deflection goes negative.