[evlatests] Switched Power Accuracy

[Rick Perley]

    I may have been premature in my denunciations this morning of the 
accuracy of the switched power correction scheme ...

    Taking X-band for example, I have two calibrators, 3C286 and 
J2007+4029.  After standard point-source calibration, I noted that flux 
ratio between these sources was different, by up to a few percent, with 
different antennas.  A simple list of which antennas have the ratio high 
vs those who have it low showed a clear division by antenna distance 
from the center -- i.e.:  I've been misled by a resolution effect.  And 
indeed, 3C286 is partially resolved at this band, in this configuration. 

    Using the standard 3C286 model helped reduce the apparent 
discrepancy, but didn't eliminate it.  We could therefore have an 
inaccurate model, or a real switched power problem, or something else... 

    A better test should use bona fide point sources.  Another test, 
which should be done with reasonably high priority, is to track one, or 
perhaps two sources, for an extended period of time, looking for 
periodicities in the switched/total power, and other 'odd, unexpected' 
phenomena, preferably at a band where external effects like RFI are 
minimized...