[evlatests] Full L-band Polarimetry

[Rick Perley]

    I've determined the gross characteristics of EVLA and VLA 
polarimetry, across the full L-band tuning width. 
The data are from the 6-hour observation of 3C147, taken two nights 
ago.  Mode = PB, BW = 12.5, 15 channels with 790 kHz resolution.  
Observations were taken at 1100, 1195, 1290, 1390, 1490, 1590, 1690, 
1790, 1890, and 1990 MHz.  Lots of data to sift through ...

    Overall Results:

    1) Except for antenna 11, all cross-polarization amplitudes and 
phases are stable -- to fractions of 1% and a couple of degrees of phase 
--- over the duration.  This is very good news. 

    2) VLA antennas (referenced to one of their own) have X-polarization 
typically 1 to 5% from 1390 through 1590 MHz.  At 1290 and 1690, these 
values are about doubled.  At 1190, 1790, and 1890, the fractions are 
typically 35 % or more.  Also, at these last three frequencies, there 
are very strong gradients in both amplitude and phase with frequency. 

    3) EVLA antennas (referenced to one of their own), have 
polarizations typically 2 to 8 percent from 1290 through 1790 MHz.  A 
couple of antennas (notably antenna 14) are usually higher -- 10 to 
15%.  At 1890 and 1990 MHz, the typical polarizations rise to 4 -- 10% 
-- still quite good.  At 1195, the typical X-polarization is up to 5 -- 
15%, while at 1100MHz, it is 5 -- 30%.  As all of these more extreme 
frequencies utilize the VLA's old OMT, perhaps most of the rise from the 
typical 2 -- 8% of the middle frequencies is due to the OMT. 

    In general, the EVLA-EVLA polarization are perhaps 1.5 times what 
the requirements have set.  However, far more important than the 
X-polarization level is the stability over the length of a typically 
long and deep integration.   The evidence so far is positive, in that no 
significant changes are seen over 6 hours.  More tests will be needed, 
however, before we can be comfortable with a relaxed level of 
X-polarization.