[evlatests] Broken/Malfunctioning Switched Power -- A summary

[Rick Perley]

    The switched power monitoring system works very well (within certain 
bounds that Vivek and I will describe in today's meeting) for nearly all 
antennas at all bands. 

    I have reviewed the behavior of this system from the 'flux 
densities' run, taken January 19/20, and give below the (very short!) 
list of those antennas/bands where the monitoring system is clearly not 
working at all.  In drawing up this list, I have left out those cases 
where apparent malfunctions are very likely due to the well-known switch 
problems. 

    1) Global problem:  ea17 in LCP was providing extremely noisy data.  
The PDif values vary over a ratio of nearly 2:1 at all bands, and in 
both IFs B and D.  The same variations are seen in PSum, but they are 
proportionally much smaller -- so the derived Tsys also varies by nearly 
a factor of two.  The character of the variations is pure noise -- there 
is no trend on any timescale. 

    2) Band-dependent problems: 

    a)  ea28 in LCP at L-band.  The system works well for about half the 
time.  For the other half, the values are many times higher, with 
enormous variations in power.  The system stays in each state for 
typically an hour or two, then reverts to the other state.  *** This 
situation has never changed since the switched power system was 
installed on this antenna***. 

    b) ea6 in LCP at S-band.  Here the PDif (only) changes smoothly over 
a few percent, over timescales of seconds.  The PSum looks stable.  This 
change is not visible in the visibilities.  Once again, this situation 
has been in place since the receiver was put into service. 

    c) ea25 in both polarizations at K-band.  The PDif values are just 
noise.  PSum looks fine.  It acts like the diode is not switching. 

    If these issues can be addressed, then I have no hesitation in 
recommending that users apply their (suitably edited) switched power 
tables.  'Suitably edited' means that there are transient events, most 
of which accompany band changes or occurance of RFI, which need to be 
reviewed and possibly purged prior to any smoothing or application.  The 
switched power system is very sensitive to external RFI which lies 
within the subband.  Blind application of the PDif values is usually not 
appropriate in this situation, as the visibility data are in general 
differently (and usually much less) affected. 

    I'd like to emphasize here that the PDif system is *very effective* 
at removing the gain losses caused by the band switch problems.