[evlatests] L-Band Sensitivity -- some remarkable results!

[Rick Perley]

    I've had a close look at real sensitivity at L-band.  First, some 
description of
the observing run:

    The observations were of 3C84, for calibration, and a piece of random
blank sky, about 5 degrees north.  Most of the time was on the blank sky. 
Data were taken in mode '4', with BW= 12.5 MHz, giving 15 channels, each
of 790 kHz, for the AA, BB, CC, and DD correlations. 

    Two observations at L-band were made, the first when the source was
at 35 degrees elevation, the 2nd when it was at 47 degrees. 

    Antenna stability etc. were excellent, as described earlier. 

    We now have two excellent tools to measure sensitivity:  The AIPS
program ANBPL derives the antenna-based weights from the calibrated
baseline-based weights.  These can be plotted in a convenient form.
    The program UVHGM can now solve for the widths of a histogram
of real and/or imaginary parts of calibrated data.  This gives the true
sensitivity, as opposed to the antenna weights, which are derived from
the switched power. 

    Some conclusions:

    1) The actual thermal noise quantified by UVHGM is an excellent
match to the antenna weights provided by ANBPL.   I become
increasingly confident that the weights are good estimate of actual
sensitivity.  A more quantitative analysis is yet to be done, but the
early indications are very good.

    2) The antenna weights for VLA antennas rose by 30% between
35 and 47 degrees elevation, corresponding to a ~14% decrease in
system temperature.  For the EVLA antennas, the corresponding
increase in system temperature was only ~4%. 

    3) There is a dramatic difference in EVLA sensitivity between
1460 and 1380 MHz.  At 1460 MHz, the EVLA antennas are near
the bottom of the sensitivity distribution.  But at 1380 MHz, they
are near the top!  Indeed, antenna 13, in IFs A and C (at 1460 Mhz)
is the least sensitive antenna in the array.  But in IFs B and D, at
1380 MHz, 13 is the nearly the best at 47 degrees elevation, and
is certainly the best at 35 degrees elevation (due to the elevation
dependency, or rather lack thereof, noted above).    These conclusions,
based on antenna weights, have been confirmed with actual
noise measurements through UVHGM. 

    Overall, a dramatically better picture than yesterday, and better
(at L-band) than I've seen on any test dataset.