1370 hipass filter ripple in polA

Background:

    lband wide has a filter bank installed after the dewar. There are 9 filters switchable filters:
  1    1370 hipass (ext Input unit. Use with filter 5)
  2    1120-1220
  3    1230-1470
  4    1550-1820
  5    1100-1800
  6    1360-1470
  7    1320 hipass
  8    1180-1280
  9    1280 hipass (ext Output unit. Use with filter 5)

Filters 2-8 can be or'd together (as long as the frequencies don't overlap).
Filters 1,9 are inline switchable before and after  filters 2-8. They always need one of the 2-8 filters inserted to complete the path (usually filter 5).
the lbfb  command can be used to insert one or more filters in the rf path.

the problem:

•  project a2703 reported ripples in polA of lbw on 27oct12.
  
• The period was about 10MHz.
• they used the interim correlator
• The ripple showed up in all 4 sbc of polA. This ruled out a bad interim correlator board.
• They also used the 1100-1800 bandpass filter as well as the 1370 hipass filter.
  
•  I looked at the ages data for 29oct12. This run used lbw.
• No ripple was seen.
• They used the wapps.
• They only used the 1100-1800 bandpass filter.
  

a2703 on/off ripple in spectral bandpass.

• Each page has the data for an on/off position switch pair. these varied between 240 and 300 secs on.
• Black is pola, red is polB
  
• The top frame shows the spectral density function for the first 25 MHz band (they had 2 offset in freq).
• the 2nd frame shows the median total power vs time for the 1 second samples of the on source
• the 3rd frame has the median total power vs time for the 1 second samples of the off source.
• The first on/off had no ripple.
• all the other on;off patterns show a large ripple in polA +/- 5% tsys up to < +/- 20% Tsys.
• PolA also showed a variation in the total power with time (up to +/- 2% Tsys). PolB was relatively stable.

30oct12 testing the 1370 hipass filter looking at total pwr vs time.

• lbw was used.
  
• the telescope was stationary.
• the interim correlator was used to take the data.
• 4 25 MHz bands were taken centered at: 1385,1400,1420,1440 MHz.
• The problem was seen in all 4 bands. 
• The total power in the 1385 MHz band was plotted vs time.
  

• Top: total power vs time using circular polarization.
• The circular is created after the dewar.
• At 200 seconds the 1370 filter was switched out. the variation in polA stopped.
• Bottom: total power vs time using linear polarization (the feed is natively linear).
• PolA is still the pol that is  drifting.
• This means that the problem is after the hybrid (ie. it's not in the dewar),.
  
• At 500, 600, and 760 secs datataking was stopped then restarted.

• total power vs time is plotted (linear polarization).
• the1370hp filter was in the path for all of these test
• Top: 1370hp filter using normal switches.
• Wiggling the filter input, output cables
• between green lines input cables were wiggled
• between blue lines output cables were wiggled.
• wiggling the cables made no difference.
• At 340 secs
•  datataking was stopped (a source was drifting through)
• the command lbfb 1 , lbfb 1 5 was reissued.
• at 360 datataking was started and polA was drifting.
• 2nd: continue monitoring with  normal switch positions for 1370hp
• 50 secs. stop.. a source had drifted through.
• issue commands, lbfb1 , lbfb 3, lbfb 1 5
• 80 secs: start datataking. polA is drifting again.
• 3rd: connect the 1370 hp filter to the switches normally used by the 1290 hp filter.
• datataking was stopped, started 5 times.
  
• while stopped, different filters were switched in, out..
  
• the total power was stable. this points to the switched that 1370hp was using, not the filter itself.
• Bottom: 1370hp back to using its normal rf switches..
  
• PolA is drifting again.

01nov12: install new rf switches for the 1370 highpass filter.

• Top: total power vs time with 1370hp filter
  
• PolA drifts by about .5% for the 400 seconds. polB was stable
• at 500 secs
• datataking was stopped.
• the cmds lbfb 1 , lbfb 5, lbfb 1 5 , then datataking was restarted. polA and B were stable.
• bottom check the stability of the other switches.
• datataking was setup to measure power at 1385 MHz with the 1100-1800 filter (filter 5)
• Before each scan (dashed blue line)
• filter 5 was swapped out
• filter N was inserted (N=1,9)
• filter 5 was switched back in
• This was looking to see if any of the other filters interfered when they were switched in , then out.
• The dark blue line saw polA vary.
• This was filter 6 (1360->1470 bandpass) being switched in and then out.
  

Summary

• a2703 saw ripples in their on/off band passes when the 1370 hipass filter was installed.
• Other users who did not use the 1370 hp filter did not have a problem.
• Variations in total power always  occurred in polA (whether linear or circular pol was used).
  
• since the hybrid for circular is after the dewar, the problem was not in the dewar.
• banging on the 1370hp input,output cable made no difference.
• switching the 1370hp filter to use the 1290hp filter rf switches showed no variation in  the power.
  
• Installing replacement rf switches for 1370hp made the total power variation diminish to about .5% over 400 secs
• when switching in/out other filters, the 1360-1470 bandpass filter seemed to affect the total power (even though it was not switched in).
• We may want to keep an eye on the lbw filter bank. The switches may be getting old...(>15 years?).
• We should also make sure that the voltage from the switch driver to the switches is correct.