[evlatests] Switched Power Functionality Check

[Rick Perley]

    The 'stress test' script being developed by Emmanuel provides us an 
easy way to check whether the switched power is operating normally.  We 
ran the test yesterday evening, and I've reviewed the results. 
    In general, the systems are working quite well.  I summarize the 
results in three categories:

    A)  Critical -- these are antennas/bands where the system is not 
working at all. 

    B) Worrisome -- working, but not well, or not up to standards. 

    C) Curiosities -- other issues or oddities, which are probably not 
important. 

    In this test, ea03 and ea09 were out of the array. 
----------------------------------------------------------------------------------------------------------

    A)  Critical Items

    1)  ea23 and ea25 at K-band.  There is zero-mean PDif on all four 
IFs.  There is perfectly good PSum values, so it seems clear that the 
noise diode is either not working, or is not being switched. 

    2) ea02 at C-band.  All four IFs provided zero-mean (only noise) 
PDifs.  This antenna gave no fringes (i.e. -- just noise), but was 
supplying appropriate PSum. 

    3)  ea14 at X-band.  Exactly the same situation as ea02 at C-band.  
(No fringe amplitudes, no PDif, but normal PSum). 

    4) ea24 at Ku-band.  Exactly the same situation as the two items 
above. 

    B) Worrisome Items

    A longer list:

    1)  L-band:   ea19, on all four IFs,  has PDifs which are 1/4 the 
value that they should be.  Worse -- the variance in the PDif values is 
about 100 times too high.  (The normal variance is about 2%, but for 
this antenna, the typical maximum values are twice the minimum). 
                       ea28, in LCP, has PDifs which change from being 
relatively good to relatively bad on timescales of minutes.  During the 
'bad' periods, the variance in the values becomes 5 to 10 times higher 
than normal. 

    2) S-band

              ea06 is showing ~5% changes in the PDif, changing smoothly 
on timescales of about 1 minute.  The LCP and RCP both vary -- but are 
exactly mirror images of each other!  These variations are not seen in 
the PSum, so this cannot be a real system gain change.  It looks to me 
that something is causing the injected noise power to change. 

             ea15 is showing 10% discontinuous jumps in PDif and PSum.  
All four IFs behave identically.  These are real gain changes, which 
last for a few seconds to tens of seconds before changing to a new state. 

             ea19 shows smooth, sinusoidal changes of about 6% amplitude 
in PDif in RCP only, with period ~ 75 seconds.  No changes are seen in 
total power, so this has something to do with the injected noise cal. 

    3) C-band

             ea16 is giving a variable and 'ratty' PDif (ratio of max to 
min is 1.25, rather than 1.02) on LCP only.   For much of the time the 
PDifs were fine, but for short (30 second) periods, they got very 
ratty.  During these periods, the PSums also got messy -- but the 
deviations were in the opposite sense, so that the claimed system 
temperatures were non-physical (less than 18K for some samples). 

    4) Ka-band

             ea13 has a PDif only 1/10 the normal in RCP, but is normal 
in LCP.  The derived system temperature is 400K.   The fringe amplitudes 
are down by a factor of 3, so it's all quite consistent:  We have a very 
noisy receiver in RCP. 

    C) Curiosities, Oddities, etc.  (Most of these are issues which are 
'fixed' by application of the PDif values.  The issue is why these 
things are happening at all ...)

     1) L-band:  ea25C and ea28B both give occasional odd PDif values.  
These are not likely to be RFI -induced, as the D and A channels of the 
same polarization are tuned to the same frequency, but do not show these 
problems. 

    2) Ku-band:  ea07 in RCP, ea13 in RCP, and ea25 in LCP are all 
showing significant (~15%) changes in both PDif and PSum on times scales 
of tens of seconds.  These are generally smooth changes, and likely 
reflect real gain changes -- which shouldn't be happening.  In addition, 
ea07 in LCP has a smooth decline in PDif and PSum, by 10%, over the 
short (2minute) scan duration.  Why would our amplifier gains do this? 

    3) K-band:  ea20, on LCP, shows a V-shaped smooth change in gain 
(both PSum and PDif), over a 2-minute timescale, of amplitude near 10%. 

    4) Ka-band:  ea20, in LCP, shows a rapid rise, of 25%, in both PSum 
and PDif, over the 2 minutes on-source.  The behavior appears to be a 
continuation of the rise in gain at the end of the K-band scan -- 
reported in item (3) above.  

    5) Q-band:  ea14 in LCP shows a steady decline of 10% in PDif and 
PSum.  ea24 in RCP shows a similar rise. 
         ea20 has very weak fringes, but a normal system temperature.  
Likely a bad, or missed referenced pointing?