[evlatests] Tests July 7

[Barry Clark]

First I tried driving antenna 10 with the Executor.  Things worked better
than last time.  Sending the phase commands to the L7s were in sync with
what the Modcomps were doing, near as I could tell by watching LEDs.  Might
be off by one or two heartbeats - will need to investigate that later.
Works satisfactorily at L band and C band, driving both IFs AC and BD.
Trying other bands would have required more typing than I was willing to
do at the time.  The Walsh function modulation did not work.  Giving the 
seed to the CMP did not result in the Walsh function going out on the waveguide.
There were occasional glitches where the phases became thoroughly messed up
for ten seconds.  This is believed to be due to missing interupts that
Hichem is trying to find.  Summary:  satisfactory progress but we aren't 
there yet.

At this point the electronikers had finished setting up the new digitizer
in antenna 16, and we (Mark, Vivek, and I) tried to use it.  Something was 
drastically wrong, especially with IF C.  We believe that the delays are
vastly different than they used to be.  We kept going to narrower and 
narrower bandwidths looking for the delays.  At narrow bandwidths, we were 
getting the full fringe amplitude expected, but we had a lot of trouble 
deciding on what the delay really was.  Turns out that we were very slightly 
overdriving the digitizer (fewer than 1% of the captured values were 
above 3/4 of full range).  In this state, for reasons unknown to me, 
amplitudes were very unstable, changing by a factor of two over a few 
minutes time.  This was confusing the 'delays' delay solver, which caused 
it to report contradictory results.  The 'D10' delay solver didn't seem
to work any better, again for reasons unknown to me.  (We might be having 
overflow problems in the FIR, I guess - I'm inadequately informed about the
effects of Mike's recent firmware changes.)  All this points out again that 
getting levels set by the Executor (using the T304 detectors) is a priority
(getting the necessary software into the T304 has been delayed because
the hardware has been a bit unstable, and proliferates in versions).
Getting the D351 deformatter set up under executor control is also a 
priority.  We ran out of time before getting delays set properly, but we 
suspect that something may have changed by many tens, if not hundreds, of 
nanosecs.

When the electronikers finished with 14, it came up and appeared to work 
well.  Naturally, we ignored it in favor of the misbehaving child.

If this sounds like not much for six hours of effort, blame the fog of war.