lbw tests done on 10jan19 showed that the signal in polA subband 2 was changing in value relative to the other 3 bands.
Since all 4 bands had the same rf input band, this problem pointed to a problem at or after the if2 mixers (where the single rf band gets split into 4 separate copies).

    On 11jan19 data was taken while the cables for polA subband 2 (behind if/lo rack 4) were tightened.
The setup was:

• lbw, inear pol, 1407 Mhz, 1100-1720 Mhz rf filter
• 750 Mhz if1 mixer
• copies of the same rf band were then sent to interim cor bands 1-4
• interim cor, 25Mhz bw, 1 sec sampling.

    I started the datataking running, and then started tightening the iflo cables for polA band 2 (behind rack 4).
The plots show the results of the test (.ps) (.pdf):

• Top polA, bottom PolB
• the colors are different subbands: black band1, red band2, green band3 , blue band4.
• I normalized each trace to the median value of the first 100 points.
• Around 340 seconds, subband 2 (red) jumped up by about 5%.
  
• I was tightening all of the subband2 cables behind the rack during this time.
• I'm not sure which cable was loose.
• If  the loose  cable was part of the if/lo then it will have affected all measurements using band2
• If the loose cable was the one that went to the interim correlator (last in the sequence of tightening) then it would have only affected data taken with the interim correlator.

SUMMARY:

• The signal from subband 2 , polA  in the downstairs if/lo was drifting in value.
• After tightening the subband2, polA cables behind rack4, the signal jumped up by about 5%.
• This would affect all observations polA band2, unless the cable was the bottom if/lo PolA subband2 cable that went to the interim correlator. In this case it would only affect interim correlator observations.
  

processing: x101/190111/chkiflocables.pro