[evlatests] June 30 tests, part 1

[Ken Sowinski]

                        EVLA Tests, June 30, 2005

Both antennas were in use for GMs pol tests in the morning.
It took a while to get it all organized but eventually we
learned that 14C was dead (no SD) and everything else seemed
to be working.  It seems that the funny L band behavior is
still there but George will have a more careful look.  14C
was dead because the output attenuator of the downconverter
was misset and we were overdriving the sampler.

Trying to diagnose 14C and to prepare for testing I found that
the 14C downconverter MIB was not responding.  Ditto for 16A.
Both were reset by cycling power just as software time began.
Perhaps we want to enable the watchdog for the downconverter
MIBs until they become more stable just to avoid the misery
of having to reset them so often.  Jim Jackson suggests that
we not look hard for software problems until all the HW is in
a stable state; I interpret that to be over a month from now.

The deformatter has been reporgrammed so that much less signal is
needed out of the downconverter to the DTS module.  Jim Jackson
recommends 24 dB of attenuation now.  This seems to be wrong for L
band at antenna 16 only 8 dB seems fine there.  After the experience
of continually having to set levels as we switch bands I echo Barry's
suggestion that we implement a set power level command in the
downconverter fairly soon.  Not all power detectors are reliable
enough yet to let this work, but some are and it will ease testing
considerably and will be essential to 'routine' observing.

At 18:05:10 IAT a long track was done on 3C84 at L band and ended
at 19:29:40.  The phase wind at antenna 16 was clearly there and
was modulated over time.  George reports from his tests this morning
that both 14 and 16 show this behavior but, unfortunately, not always
at the same LO settings.  It might be interesting to make a set of
observations at a sequence of X band frequencies to see if we can see
this at another band if we try hard enough.

Delays were checked and corrections were needed at C band for 14, 
and at X band for both antennas.  

After setting the delays I looked at the continuum closure errors
reported by the Modcomp ANTSOL at X band.  IF A is noticably worse
than C, probably because the VLA T3 does more violence to the IF A
pass band than it does to IF C in the baseband conversion process.
       IF A       IF C
     Re    Im   Re   Im
14   15%   30%   4%   6%
16    9    30    8    6

A few minutes of 50 MHz line data was taken to see if there are any
especially noticable birdies.  Both these scans only lasted a few
minutes.
20:24:30  Start K band, mode 2AC
20:45:00  Start X band

Just for fun I measured delay errors at 50 MHz in both line and
continuum modes.  They are quite different, perhaps because of
peculiarities of either the VLA or EVLA bandpasses.  Using antenna 6
as a reference antenna the continuum delay errors were essentially
zero.  In line mode the apparant delay error between 14A and 16A was
20 ns.
     IFA  IFC
14   +10   -4
16    -4   -4