[evlatests] PDif Issues at L-band

[Rick Perley]

    A test to try clarify the origin of the PDif compression problem 
(8-bit style) was done Saturday evening.  The test is rather 
complicated, and analysis is ongoing.   The array was in a bit of an odd 
state due to the power issues of that day. 

    The SY table data are in general quite good.  But some clear 
failures occurred and are noted below for completeness.  Note that we 
have only L-band data in this test. 

    ea08:  In IF 'A' only, both PSum and PDif alternate between two 
different states on a rapid (1 second or less) timescale. 
    ea10:  In IF 'A' only, PSum has two different states, similar to 
ea08.  PDif has many discrepant values, which look to me to be due to 
the rapid change of states seen in 'PSum'. 
    ea11:  The PDif values are all zero (with noise), while PSum looks 
fine.  Look like there is no switched power on any IF. 
    ea16:  The switched power in IFs 'A' and 'C' is about a factor of 
100 too low. 
    ea22:  The PSum data look perfectly normal, but the PDif values are 
bizarre, like nothing I've ever seen before!  There are equal numbers of 
positive and negative PDif values, but these are not due to noise.  
These are real signals, but just plain crazy, on all IFs. 
    ea28:  IFs C and D (LCP polarization) have zero-mean, noisy PDifs.  
The PSum data for these IFs are pretty good, but closer inspection shows 
they are much noisier than they should be.  RCP polarization (IFs A and 
C) are much better, but still not as good as they should be.  (This 
antenna has always shown this behavior at L-band). 

    Note that antennas 1, 2, 13, 15, 19, 23, 24 and 26 were out of the 
array.