[evlatests] phase flippiing is not in the correlator

[Ken Sowinski]

We at last have a clear demonstration that the 180 degree
phase flipping long reported by Rick has its origin in the
antennas rather than the correlator.

Last week Rick provided a file in aid of studying the mapping
problems which provoked a phase change of 180 degrees at every
scan in half the antennas.  80 second scans were used so that
we were able to see the problem in a reasonable amount of time.
The observation was at L band (SSLO was 988.0 MHz) and the LO
offset fundamental was 10 kHz.  Multiples for each antenna were
chosen based on the slot occupied by the station board which 
processed data from that antenna.  As reported in the past 
antennas form two groups based on the parity of the multiplier,
one group flips and the other does not.

ID   Loc  Mult
--   ---  ----
ea19 r1s0  1
ea01 r1s1  2
ea02 r1s2  3
ea09 r1s3  4
ea25 r2s0  5
ea28 r2s1  6
ea24 r2s2  7
ea03 r2s3  8
ea18 r3s0  9
ea23 r3s1 10
ea04 r3s2 11
ea08 r3s2 12 (but no L band receiver!)

The file was modififed so that half way through the antennas in
rack 1 were no longer sent L302 commands to try to distinguish
between an L302 and a correlator problem.  The results were clear
and also allow us to determine which antennas are flipping.

1.  Baselines between antennas in rack 1 exhibited no phase flips
after L302 commands were disabled.

2.  Baselines between all antennas in rack 1 and others with even 
parity exhibited phase flips after L302 commands were disabled.  This 
is the evidence that antennas in even slots (odd multiples) are the 
ones which flip.

3.  Baselines between all antennas in rack 1 and others with odd
parity all exhibited no phase flips after L302 commands were disabled.


The correlator is exonerated, but we are left with two mysteries.
1.  What is the L302 doing?  2.  Why is it doing anything?  We
exepect it to do nothing when commanded to the frequency that it
is already set to.