[evlatests] P band

[Barry Clark]

First problem as that the deformatters (D351s) had lost track of time.
Some of them thought it was July, some thought September.  In either
case, when they were getting time-tagged commands for October, they were
saying "Not going to wait that long", and rejecting them.  So the 
resampler gain commands were not getting through.  Hichem said they
should be rebooted.

I don't know when the D351s lost track of time (Rich might be able to
tell from the monitor data), but since that time, the requantizers
have been set to the default for the other bands, 30.  This has very
bad effects on the system - in the worst case a factor of several in
sensitivity when the gain was set to the current command of 15.

So I had the Executor run a script that rebooted the D351s.  I suspect that
there may be a command delivery problem from the Executor.  Of the 
couple of dozen I looked at, one did not get the word to reboot, and,
after the reboot, two others did not get the command to reset their
FIRgain when the script was restarted.

IF A and C for all antennas had about the same RMS on the link that
they did at other bands, as one might expect, since they are under level
setting control.  This seems to be mostly in the 8 to 12 range, which I
believe is a little higher than what we had in mind when we set this up,
but never mind, it seems to work pretty well.

Judging the effect of changing things on fringe power (and thus sensitivity)
was sometimes difficult.  Fringes were sometimes rather random.  I didn't
trace this down to see if it was problem antennas or interference bursts,
or whatever.  But when the fringes were well behaved, they agreed pretty
well with the the T5s, that best sensitivity was obtained when the T5 
ALC voltage was in a pleasant part of its range (about -3v).

For IFs A and C, the T5s, and in those cases where the fringes were well
enough behaved to see what was happening, the fringe power, were happiest
with a requantizer gain of 8.  A couple of cases got a few percent better
sensitivity at a gain of 6 - 14 A and 26 A.  One case was very unhappy
with a gain of 8 and wanted more like 12 - 21 C.  This could most easily
be explained by extra power somewhere in the band, outside the normal
P band area.

IFs B and D are set from attenuator values in a table.  Most of them are
set to produce lower link RMSs than the AC, by about five db - 14, 17,
18, 23, and 26.  21 and 24 have much the same link power as the AC sides,
or perhaps only 1 db lower.

So what to do.  First, I plan to fiddle the tabulated attenuations
to get AC and BD behaving more nearly the same, then I propose to set all
the requantizer gains to 8.  As I argued yesterday, I think the same
number is right for 4 band, but it is too late to test that.  Somebody
remind me to check in a year when next we hand the dipoles.

This is a bit much for today, I'll tackle the problem again next week.