[evlatests] Insensitivity of PDif compression to total power

[Rick Perley]

    One more benefit from the test last night:

    It can be conclusively shown that the 'PDif Compression' problem is 
*absolutely independent* of the input digital power. 

    In general, from this test, the PDif compression, when going onto 
Cygnus A, is quite small:  typically no more than 5%.  Some antennas 
have absolutely zero compression.  These include:  ea03R, ea07R, ea07L, 
ea08L, ea16L, ea20R, ea24R, ea27R and ea27L.  Many others are only 1 to 
2 percent.  The only antenna-IF with significant compression is ea02L, 
for which the PDif values are extremely low and variable. 

    Since the observed Cygnus A at two very different power levels in 
this experiment (varied by a factor slightly more than three), I could 
compare the PDif compression values in the low-power state to those in 
the high-power state.  **They are exactly the same**.  PDif compression 
is unrelated to power measured on the station boards. 

    The earlier (early September) observations -- indicated quite 
clearly that the PDif compression issue is only notable in the digital 
PSum power, and doesn't affect the cross-correlations.  (A safe 
statement is that any compression in the cross-correlations is limited 
to less than about 3%).