[evlatests] No runs, no drips, almost no errors

[Rick Perley]

    I ran a short (30 minute) 'fast continuum' file this afternoon, 
looking for
amplitude drops, phase spins, and other assorted phenomena we have been
struggling with over the past few weeks.

    The run consisted of five different sources (A,B,C,D and E), 
observed twice each, in
the following manner:

    A,A,B,B,C,C,D,D,E,E

    The idea was to see how the phase behaved when changing sources, and 
when
changing scan, but not source. 

    First the good news:

    1) No amplitude dropouts were seen. 

    2) No phase spins were seen. 

    3) The ~6 seconds of 'dead' data, at the end of every scan, has been 
eliminated.
(Bryan got it right the first time!)

    4) The phase returned to the proper value upon every new observation 
-- no
jumps were seen.  The phase behavior of 14 and 16 are exactly as 
expected when
compared to the neighbor VLA antennas. 

    Then the bad news:

    1) A new phenomenon is found:  When changing source, the first 30 to 60
seconds of data from the EVLA antennas, following the arrival of the 
antennas
on source, are bad.  (In other words, all the VLA antennas are fringing with
each other 30 to 60 seconds before the EVLA antennas fringe).  Bryan's
new scripts may have something to do with this ...

    2) Antenna 16D is bi-stable:  about 1/3 of the data points are at an 
amplitude
level about 60% of the normal.  The distribution is 'clean', in the 
sense that
the amplitudes are either normal, or reduced by that factor.  There are no
intermediate values. 
    Note that 16A, which used to have two amplitude states, is now quite 
normal. 

    Sensitivity:

    Ken reminds me that the amplitude levels have not been re-set, so 
any conclusions
about sensitivity aren't worth the time to either write here, or read.  
When the levels
are re-set, we'll repeat the sensitivity tests at all bands.