[evlatests] Widened L-band tuning for VLA antennas

[Rick Perley]

    All VLA antennas have now had their low-pass and stop-band filters 
-- originally installed in very early times to stop saturation from a 
strong Forest Service link frequency -- removed. 
    A test was run yesterday to measure system performance.  The test 
was run in spectral line mode, at 16 frequencies from 1210 to 1960 
MHz.    All VLA antennas can now tune above 1730 MHz (the previous lower 
limit).  Some conclusions from the test:

    1) The previously-reported incoherency between the VLA and EVLA is 
now known to be due to the executor's (mistaken) belief that the L6 
synthesizers cannot bet set above 3090 MHz.   Hence, the VLA antenna 
were fringing nicely at another (unrequested) frequency.  My earlier 
tests (on antennas 5 and 10), taken under Modcomp control ( which 
apparently suffered no such misconceptions) clearly show that 4010 MHz 
is a legal and working frequency.  Ken is currently checking the 
documentation to determine what the upper limit is. 

    2) VLA sensitivity at 1810 MHz (the highest frequency which worked 
for the test) show a reduction by about a factor of four in sensitivity 
(equivalently, Tsys is four times too high, or efficiency four times too 
low) compared to the middle L-band frequencies. 

    3) VLA sensitivity at 1210 MHz is similarly a factor of four worse 
than at the middle frequencies.  In this case, the cause is a waveguide 
resonant cutoff, clearly visible in the bandpass spectra. 

    4) The EVLA sensitivities are nearly completely flat between 1260 
and 1960 MHz.  There are some exceptions:  antenna 13 steadily worsens 
with increasing frequency (this is a long-known problem), antennas 17 
and 21 are less sensitive overall (likely due to high Tsys). 

    5) The calibration was done on 3C48, at an elevation of 25 degrees.  
At this elevation, the EVLA antennas have better sensitivity -- by 
roughly 15% -- than VLA antennas.   At higher elevations, the situation 
reverses.