Intro:

     On 19apr11 project a2010 (alfalfa) reported some instabilities in alfa beam1B.
Data from projects a2048 (ages), a2010 (alfalfa), and a2130 (galfacts) was examined to see what was going on:

18apr11: a2048  (ages) (19:30 to 23:30) (top)

• 1200 secs of data is taken while the telescope sits (drift mode) .
  
• The telescope then moves to a new position and repeats the drift.
  
• there were 10 1200 second drifts.
• The mock spectrometer with 1 second integrations and 172 MHz bandwidths were used.
• the 1450 band was used to compute the power.
  

• Page 1 total power vs time (minutes).
  
• Top polA, bottom polB
• Each color is a different alfa Beam. The vertical scale is Tsys (with offsets for display).
• the dotted lines show the start of each 1200 second drift. The gap before each start is the time to move to the new position. During this move no data is taken.
• Each 1200 second scan has been separately normalized to the median value.
  
• beam 1b (red) around 200 minutes there is a large change in the total power.
• Page 2: Overplot each 1200 second drift for beams 0 and 1.
• top to bottom: beam0a,beam0b, beam1a, beam1b
• bottom: beam1b shows 3 drifts that vary by 2 to 6% of Tsys. All other beams are relatively stable.

19apr11: a2010 (0:00 to 2:30) (top)

• Drift scans for 600 seconds.
• wapps 100 MHz bw centered at 1385 MHz.

• Page 1: total power vs time for the 12 600 second drifts.
• Top polA, bottom polB
• Each beam is plotted with a separate color. They have been offset vertically for display
• Each beam,pol total power data is normalized to the median value of the  600*13 samples.
• The dotted vertical lines show the start of each 600 sec drift.
• Beam 1b (bottom red) shows variation in the total power.
  
• Page 2:
  
• Top: overplot beam0 a,b and beam 1 B
• Bm1b (green)  has a 2% Tsys variation (after removing the Tsys drift)
• beam0 (black,red) show a  ripple with an amplitude < .5% and period of around 400 seconds.
• Bottom: beam0a tp over plotted with temperature variation in Gregorian dome room.
• The blue line shows yblue= (tempGregorianDomeRoom - linearFit)*.1
• The variation in the temperature varies with a period of about 25 minutes. This is much slower than the bm0 variation so it is probably not coming from the dome room temp variation (still need to check the downstairs temp variation).

A2130 data:

• beam 1b jumps around in the a2130 cal deflection data. The largest jumps are 20apr11.
  

Summary: (top)

• alfa beam 1b showed variations in total power output 18-20 apr11 (and probably after). bm0,bm2-5 were ok.
  
• a2010 showed 2% tsys variation
• a2130 showed 5% variation in the calDeflection
  
• The changes were  seen using the wapps and the mock spectrometer so the problem is not in the backend.
• the changes take about 100 - 200 secs to vary up to 2% of Tsys.
• The cal deflection change of 5% shows that it is  a variation in Gain and not Tsys
• Calon-calOff subtract out Tsys (since the calon,caloff was much faster (.2 secs) than the variation (hundreds of secs)).
  
• I'm assuming the cal diode output is constant.
• a 400 second variation in bm0 a,b was also seen with the a2010 wappdata.
  
• Amplitude < .5% sys
• The Gregorian room temperature ripple was much slower than this so it is not a temperature related gain change from the dome.
• need to check the temperature variation of the clock room where the wapp, fiber receivers are located.