[evlatests] C-band PDif compression, and more

[Rick Perley]

    The holography run used 3C84, which adds about 2.5K to the system 
temperature -- 10% of Tsys at most. 

    As feared, PDif compression is seen in a number of antennas, at the 
level of typically 1%.  If, as we believe, the compression does not 
affect the visibilities, an error of that magnitude to the flux scale 
will result from applying the PDif values.  (Note, however, that since 
for most antennas the 'compression' is much less than 1%, the overall 
effect will be less). 

    The compression was much more commonly observed in our second IF 
(tuned to 6.5 GHz) than in the first (tuned to 4.5 GHz). 

    Antennas with the most obvious effect are:

    1Band D, and 6B  for which the compression is ~1%.
    12B, 16B, and 27B, for which is is about 0.5%. 

    For most antennas, the effect is smaller than the measurement noise, 
or is swamped by real gain changes. 

    Other things found in these data:

    A) ea09 had steadily rising amplitudes throughout the hour-long 
run.  These are not due to amplifier gain change, as the PDif were steady. 
    B) 18D and 23B show significant gain variations between, and within, 
calibrator scans.  These were nicely corrected by the PDif values -- 
showing again that PDif is doing what it was designed to do.