[evlatests] PDif Compression at P-band

[Rick Perley]

     This simply summarizes what I'll say in the upcoming meeting.

     I've previously reported that this phenomenon (decrease in PDif 
when observing a strong source) was mostly absent at P-band -- as 
opposed to its strong presence at other bands.

     The test I ran last Thursday (3C295 and Cygnus A) provided good 
data for a closer look.

     Results:

     PDif compression is present at this band, but at a lower level than 
on other bands.  Cygnus A approximately quadruples the system 
temperature.  At that level:

     *  The median PDif compression is .98 (i.e. a 2% drop in apparent 
gain).
     *  90% of all antenna/IFs available for the test (46) showed an 
effect between 0.94 and 1.01.
     *  40% of all antenna/IFs available showed no compression/expansion 
to within the 1% accuracy of the measurement.
     *  Five antenna-IFs showed an unusually high effect:

         - 4Y, at 0.87
         - 26Y at 0.92
         - 27X at 1.04 (i.e., expansion)
         - 15Y at 1.08
         - 22Y, which was so variable in its power that an accurate 
measure was not possible.  It looks like about 1.5 (major expansion).

     Note that:

     * There is no apparent correlation between those discrepant 
antennas and the list of misbehaving antennas reported earlier. That is 
-- most of the antennas in the list above showed normal sensitivity, 
bandpass, amplitudes, etc.

     Bottom Line:

     This is not a large enough effect to be concerned about.  We have 
shown previously that the effect of 'PDif Compression' on the actual 
visibilities is a factor of a few less than the values suggested by PDif 
itself.  Hence, the median 2% compression observed will have essentially 
no effect on the measured visibilities.

     Perhaps more interesting is what causes some apparently normal 
antennas to have large PDif compression/expansion...