[evlatests] A couple of global switched power issues

[Rick Perley]

    In trolling through the 8 bands of data taken last week in the 'flux 
densities' run, a couple of probable-global characteristics become 
clear.  ('probable' because I've gone through 5 of the eight bands so 
far, and the issues below are in all five). 

    1) ea17 is giving no useful switched power in IF 'C'.  There are 
non-zero numbers, but they are not reflecting changes in gain which are 
easily visible in the other polarization.  IF 'D' is fine. 

    2) ea20 provides good PSum and PDif most of the time in LCP.  But on 
occasion, the values become noisier by about an order of magnitude.  The 
duration of this noisy behavior is typically an hour, following which 
there is an hour or more of perfectly good data.  Curiously, this 
happens in both IF 'C' and 'D', but not at the same time!  The problem 
is qualitatively the same, but the two IFs appear to be independent. 

    I add one more extreme curiosity (which Ken claims 'cannot 
happen').  ea04, in IFs A and C, and in subband 3 only (AIPS counting), 
gave zero mean (i.e., only noise) PDif values.  All other subbands gave 
excellent data.  This occurred at Ka band only.  It occurred uniformly 
throughout the entire 30 hour run (despite it being run in four 
separated pieces!).  This might make it a predictable failure ... 
somebody might take a look.