[evlatests] EVLA L-band Behavior

[Rick Perley]

    I had early test time today, so ran the EVLA through its paces.  All 
antennas,
all bands.   Only one bad item to report:

    *  The EVLA antennas don't respond to pointing requests -- at least 
not for me. 
I began the run with a check of pointing, including referenced 
pointing.  No solutions
were returned from any EVLA antenna, at any band.  Upon closer 
inspection, it
is clear that the EVLA antennas don't move to the requested (half-power) 
positions.
They certainly did this yesterday.  All VLA antennas behaved as expected. 

    Other than that, all is working very well indeed.  I have analyzed 
only the L-band
data carefully so far, and from this:

    Amplitudes and phases are marvellously stable (exception noted 
below), with
gain fluctuations at least as good as VLA antennas.  [To put numbers to 
this, the
amplitude gain fluctuations are less than 0.5%, peak, for most IFs, 
phase fluctuations
are a small fraction of one degree about the general trend].  Superb!

    The single exception is antenna 13, IFs B and D.   There was a large 
(> 100 degree)
phase jump on IFs B and D, at 17:19:13 IAT.  The change was not 
instantaneous --
a stable phase was regained only after two records (3.3 second averaging 
was used). 
Nearly all the change occured in the first (and the amplitude was 
severely depleted for
that one record).  About 20 degrees more change was seen in the 2nd. 
   
    Careful inspection of the amplitude stability shows that we are 
getting 3% drops
in antenna 13 amplitude, on IFs B and D only.  When this happens, it 
occurs on every
3rd integration -- that's exactly 10 seconds separation.  But the effect 
is not seen all
the time:  every 3rd record was weak by 3% in amplitude over 20 seconds,
followed by 20 seconds where all was good, then 20 seconds when every 3rd
record was down, etc. 

    Gotta be a L302-2 issue?

    Bandpasses were normal, and delays are fine on all IFs. 

    I'm playing with Eric's new histogram fitting program (UVHGM) to 
estimate
*real* sensitivities.  Early indications are that 13, 14 and 16 are less 
sensitive than
VLA-only baselines by ~20%, as indicated by the aips weights. 

    I'll report on the other bands, as I can process the data.