15jan14

Intro

    Project p2030 reported a time variation in the total power of the winking cal while using alfa.

Winking cal on continuum source:

    on 21dec13 p2030 took 25 Hz winking cal data while tracking 3C138 (10 Jy at 1400 MHz).

Their setup was:

• alfa receiver, 172 MHz bw, cfr 1450
• hardware winking cal at 25Hz.
• 65usec sampling, 512 lags.
  
• Total power computed for the 1450 MHz band (172MHz)
  
• PolA and polB were averaged in the mock spectrometer.
  
• Data sampled for about 10 seconds on 3C138.
• Data then folded at the 40 millisecond period of the winking.

    The plot shows the folded cal data (.ps) (.pdf)

• Top: 10 seconds of winking cal data folded to .04 seconds (cal period).
• The vertical scale is degK. It is high since 3C138 is about 10 Jy at 1400 MHz.
• You can see the slope of the signal.
• Bottom: remove the median value from the calon and caloff
• This shows the differential change for during the calon and cal off
• The change is about .28 K or .21%

Winking cal with the alfa cover in place:

    On 14jan14 25 Hz winking cal data was taken  while the alfa cover was in place.
The setup was:

• 172MHz bw centered at 1450 MHz.
• 1 millisecond sampling, 2048 channels
• start on 1 sec tick so the samples are synchronized with the cal
  
• 40 samples per cal period
• Pol A and PolB were processed separately
• Data was analyzed for all 7 beams.

    The plots show the winking cal with the alfa cover in place (.ps) (.pdf):

• There are 7 pages (1 page per alfa beam).
• Top: cal on,off folded for 25 seconds.
• + shows polA
• * is polB
• The left half of the plot is cal on, the right half is calOff
• You can see the drift in the cal value.
  
• Bottom: remove the median value from the cal On and calOff plots
• The cal transition is occurring during the first samples of the transition (so they were ignored.).
• PolA and polB show the same variation
• The change is about .06K,-.2K  this is about .3 %

Summary:

• a time variation was seen in the total power of the winking cal.
• The variation is the same for all 7 beams of alfa
• The variations are also the same  for polA and polB
  
• this argues that it is not an alfa amplifier problem since you would expect them to vary a little
• it take up to 10 milliseconds for the cal to reach a fixed value when the cal is turned on.
• It also takes about 10 milliseconds for the cal to reach a constant value when the cal is turned off.
• The variation with time is about .28% of Tsys (with Tsys varying by a factor of 10).
  
• This argues that the problem is a gain and not a noise problem.
• At first i thought it might be an amplifier saturation problem where it took a long time for the amp to recover (like we had with the amps and the radars before the leds were installed on the amps).
• It's not this, since there is no saturation possible with the cover on.
• If the problem was a cal diode problem, then it would be a constant degK difference and the ratios to Tsys with 120K and 12K should be way different
  
• But fraction drift was the same for Tsys's that varied by a factor of 10. 
• So how would the cal know what the system temperature was going to be...
  

processing:x101/140114/alfawinkingcal.pro, 140113/alfawinkingcal.pro

---