[evlatests] L-band Performance for 13, 14 and 16

[Rick Perley]

    2.5 hours of test time Saturday morning were used to check system
performance at all bands.  The tests were:

    a) 0.4 second averaging continuum data, on a strong calibrator
and on nearby blank sky.   This is used to check sensitivity and
gain stability. 

    b) 3.3 second averaging in spectral line modes 1A, 1B, 1C and 1D,
providing 32 spectral channels over 25 MHz bandwidth.   These are
used to confirm bandpasses, delays, and sensitivity. 

    c) At L, C and X bands only, I observed the standard flux calibrator
Cygnus A, plus a nearby calibrator.  The goal here was to see if
reported antenna temperatures are reasonable given the (known?)
antenna efficiency.  (System non-linearities will also show up --
but never mind ...)

    Here I report only on L-band performance from the 'fast continuum'
data (test (a)). 

    1 - All three EVLA antennas worked well in all four IFs. 

    2 - With the exceptions noted below, the observed gain fluctuations seen
within the strong source observations were as good as the best VLA
antennas -- both in amplitude and phase. 

          Exceptions to above statement:

       A)  We saw a repeat for a peculiar amplitude drop/rise combination in
antenna 14 *only*, and on one scan *only*.  The observed correlation
(14 against 'x') drops quickly (within one second) to a level 10 to 15
percent lower than the initial state, then after 20 seconds, discontinuously
rises by 25%, then smoothly declines to the initial value.   This event was
seen in all four IFs, but only strongly in IFs B and D.  (But in the 
previous
incident, it was seen equally in all four IFs).   The event is sort-of 
mirrored
by the Tsys values, which rise for 15 seconds to values 1.5 times the
initial, then smoothly decline to initial values over about 30 
seconds.   (This
is IFs B and D -- for A anc D the Tsys rise is by only 1.5 degrees). 
    Remembering that the Tsys values are smoothed with about a ten-second
characteristic time,  while the correlation coefficient (which is smoothed
only over 0.4 seconds) is adjusted by the measured Tsys,
the observed phenomenon can be qualitatively explained thusly: 
    For reasons unknown,  the astronomical signal is 'diluted' by some
uncorrelated broadband noise.  This noise increases in strength , more
or less steadily, for 30 seconds, at which point is ends abruptly.  The
observed drop, rise, and decline of the observed visibilities can be
explained from the smoothing of the Tsys data utilized to modify the
observed  correlation coefficient. 
      
    B)  Antenna 13's reported Tsys, and the observed visiblities, 
occasionally
behave in a manner unlike any other antenna.  I'll spare you readers the
details, but will put together a small visual package to show later. 

    Back to overall observations:

    3)  Sensitivity.  The aipsweights give a good estimate of the
basic antenna sensitivity.  For all IFs, and all 3 EVLA antennas, the
sensitivity is significantly less than the best VLA antennas.  But
the factors vary between antennas and IFs.  Below is the ratio of the
aipsweight to that of the best VLA antenna.  The shown ratio can
be caused by a decrease in efficiency by the given factor, or an
increase in system temperature by the same factor (or a combination
of the two). 

    13AC      0.63
    14AC      0.80 in A, 0.67 in C
    16AC      0.67

    13BD      0.81 in B, 0.51 in C
    14BD      0.88
    16BD      0.82

    Perhaps more useful will be rankings, or a histogram.  I'm working
on these. 
    The noted drops in aipsweights are matched by increased noise in
the rms of the calibrator observations, and on the blank sky observations. 

    4) Closure.  I regret to report that we have significant closure between
antennas 13, 14, and 16, when they are calibrated against VLA antennas. 
These 3 baselines have amplitudes 8 to 9.5% too high for all IFs.  There 
is no
visible phase effect. 

    Spectral line results to follow.