[evlatests] L-Band sensitivity on 13, 14 and 16

[Rick Perley]

    We really do have a serious sensitivity problem at L-band on these 
antennas.

    The Sunday night data were taken on 3C286 and blank sky, with 0.4 
seconds
averaging, in continuum, with default frequencies.  Despite the odd 
phase wind
reported earlier, and the lack of fringes between VLA and EVLA for IFs A 
and C,
the data were easy to edit and calibrate.    The elevation was very 
high, about
80 degrees. 

    The aips weights again indicated a serious lack of sensitivity for EVLA
antennas.  That this represents the truth was shown by fitting histograms to
the black sky crossed-hand data, real and imaginary parts. 

    VLA antennas numbered 1 through 10 gave a histogram with 1-sigma
width of 72 mJy for both frequencies (1465 and 1385 MHz).

    The three EVLA antennas gave a width of 103 and 92 mJy for 1465 and
1385, respectively.   I have adjusted this number for the EVLA-EVLA
excess ('closure') caused by the mis-matched bandpasses between the EVLA
and VLA antennas.   All three EVLA baselines are equally poor.  This seems
to be a real EVLA issue, not just a rogue antenna. 

    So where is the problem?   Bob Hayward and I did efficiency and Tsys
tests at the front end on antenna 13 which clearly showed that G/T should
be about the same as VLA -- perhaps at most reduced by 10%. 

    I list some possible explanations:  a) High Tsys.  b) Low efficiency.
    c) Bad pointing.  d) Defocussed.  e) Subreflector seriously out of 
position.  

    c, d, and e hardly seem likely at this frequency. 

    If we could believe the reported Tsys, then we can discriminate between
a) and b) by observations of the change in antenna temperature when going
onto a standard calibrator.  For these observations, the calibrator was 
3C286,
with flux density about 15 Jy -- which should contribute about 1.3K to the
system temperature. 
    For antenna 13, the increment observed is indeed about 1.3K, but the
reported system temperature is 50K!  If the efficiency is right, then 
Tsys is
nearly twice what it should be.  If Tsys is wrong (because the Tcals are
incorrect), then the efficiency is half what it should be.  Either way, 
we've
got trouble.
    For antenna 14, the increment is a paltry 0.8K, while the Tsys is about
what we expect -- 30K.  So here the situation is reversed from 13, but
the result is the same -- G/T is greatly reduced. 
    For antenna 16, the increment is slightly over 1 K, with a reported 
system
temperature near 35 in RCP, and 30 in LCP.  This one is closer to what we
expect. 
    For the VLA antennas, the observed increment varies, but is nearly
always proportional to the reported system temperature -- a strong 
indication
that the Tcals are incorrect.  For those antennas whose reported Tsys is
about 35K, the increment is exactly as expected -- about 1.4K. 

    I have (separately) taken L-band data at 14 frequencies between 1275 and
1750 MHz, in spectral line, using Cyg A, blank sky, and two standard
calibrators, to check the variation with sensitivity with frequency.  Chris
Carilli's student Ran Wang is reducing these data for me, and will give 
results
when done.