[evlatests] Driving VLA antennas

[Barry Clark]

Spent the usual (large) amount of time confusing myself.  I was operating
in subarray 2, which is using Fluke set 2.  This is OK most of the time
since the Fluke sets are usually set to the same frequency.  When 
subarray 1 switched to P band, however, Fluke set 1 changed frequency.
The Executor controlled antennas used Fluke set 1, and therefore gave
no fringes with the rest of subarray 2, but fine fringes with each other.
Except for IF C from antenna 9, which appears to have a stuck relay, and
was using Fluke set 1, and therefore fringed with the Executor controlled
antennas but not with the rest of subarry 2.  These symptoms were *very*
confusing.

At one point the CMP stopped sending commands to the antennas.  Since 
Hichem was at lunch at the time, the only thing I could think of to do
was reboot it.  When I did, it didn't come back.  When Hichem came back
from lunch, it was up and read.  I suspect it was trying to mount some
nfs file that isn't there, and took many minutes to time out.  Somebody
should have a look and see if that was what happened.

Then the machine ran very well for about an hour.  The Walsh function
patterns appeared to stay in sync with the rest of the VLA, and the 
antennas generally behaved themselves quite usably.  

There were occasional glitches - bad amplitudes and phases - with the 
Executor controlled antennas.  Amplitudes were down by 30-50%, phases 
pretty random.  These sometimes occurred in a single 1 2/3 sec integration,
usually in two adjacent, once in six adjacent.  These are pretty hard
to explain.  An error in walsh function timing should result in no
phase change or in a 180d phase change.  An error in fringe rate or 
phase timing should result in a systematically increasing phase difference
depending on amplitude reduction.  The actual behavior corresponded to
neither pattern.

I was working only at C band, so I have not yet really tested setups at 
other bands.