[evlatests] EVLA tests, Aug 8, 2005

[Ken Sowinski]

                       EVLA Tests, August 8 2005

It has been a week since we tried anything semi-serious so the goal
today was to see what state antenna 16 is in now.  When I began the
day last week IF C was the only broken IF; today IF C was the only
working IF.

L302-2 did not lock at any of the bands tried.  Jim proposed to take
up a replacement module on Tuesday.  IF A did not work because there
is no DTS module.  

That leaves IFC.  I was able to make it work at L and C bands, but not
at X band.  Delays were far off because IF C was not working last time
delays were adjusted.  I fixed that and looked at the phase switching
problem which had not magically fixed itself since last week.

The phase in IF C was stable at L and X band.  It is conceivable
that it was winding so fast at X band that the antenna looked 
incoherent.  That is supported by the large RMS seen on D10 while
looking for X band fringes.

The fact that phase switching does not work suggests a timing error.
The intent today was to try to estimate that timing error.  Phase
switching was turned off at the correlator, a simple square wave
pattern (seed=32; 32 on, 32 off) was used in the L302 and the 
integration time was set to 8 WG cycles allowing 4 integrations in
each state.  If timing were correct we would see a pattern that is
eight records long, 4 with direct phase and 4 with the phase offset
by 180 degrees; all of the same amplitude.  What was seen was that
the first record of each phase state had the amplitude degraded by
about 1/8th.  I offset the timing by one WG cycle (to -(32sec + 1 WG
cycle) and the degraded record changed to the last record of each
phase state and again by about 1/8th.  This behavior suggests that 
the phase switching events are occurring in the middle of a WG cycle.
The 12.5 per cent amplitude reduction is consistent with a shift of
half a WG cycle when the integration time is eight WG cycles.

offset    ampl   ampl at phase transition
======    ====   ========================
-32+1WGC  .24      .15
-32       .24      .21
-32-1WGC  .24      .21
-32-2WGC  .24      .15

Either we have something very subtle wrong in the L302 or the phase
offsets are being strobed into the DDS in the middle of a WG cycle.
If there really is something wrong it is hard to see how it would
I asked the L305 to resynchronize and that did not help (or hurt)
matters.  


It seemed to me that the T304 power level setting command is not working
right.  I discussed this with Pete and he demonstrated that the MIB does
the right thing when commanded by hand.  I was able to demonstrate that
the executor sets the target correctly, at least at X band.  Is it possible
that it does not work at other bands, or that the target command arrives
before power levels have stablilized after a band switch?