[evlatests] Essential Results from C-Band Polarimetry

[Rick Perley]

    I used 7 hours of dynamic time last Sunday afternoon/evening to do 
deep polarimetry at eight frequencies across C-band, using the seven 
available antennas equipped with wideband OMTs (2, 3, 9, 15, 21, 24, and 
28).

    The source observed was 0217+738 -- a perfect point source with low 
polarization and 4.1 Jy flux density.  An observation of 3C48 was added 
at the beginning to check the flux scale.  Observations were made in 50 
MHz continuum, with 1.67 seconds averaging.   The frequencies used were 
4885/4385, 5385/5885, 6385/6885, 7385/7885 MHz.  (AC/BD frequencies). 

    Essential Results:

    1) Data quality was excellent.  One or two antennas, as some 
frequencies, were slow to lock up (not a pointing issue) by a few 
seconds.  No clear patterns for this were seen -- it's a rare event. 

    2) No phase jumps were seen.  I loaded the data as correlation 
coefficients, and most behaved normally.  Two antenna-IFs were peculiar 
at all frequencies, changing from scan to scan by about 10% (i.e., never 
within a scan, only between scans):  3D and 28C.  This cannot be a 
pointing or efficiency effect, so the system temperature is apparently 
changing -- perhaps from mis-set power levels to the sampler? 

    3) Antenna sensitivity follows the trends reported by Emmanuel and 
me in the past:  about 25% poorer between 4 and 6 GHz than between 6 and 
8 GHz.  Tests by Bob Hayward and me indicate this is due both to a 
system temperature increase and an efficiency decrease -- possibly due 
to illumination effects.  (Further elucidation will require 
holography).  Overall, sensitivity between 6 and 8 GHz is extremely good 
on all antennas:  Typical Tsys/effic. = 45.  Antennas 15 and 24 are the 
best, with Tsys/effic = 39 and 41 K,  respectively.  Conversely, the 
typical Tsys/effic between 4 and 6 GHz is about 55K, again with 15 and 
24 being the best (and 2, 3, and 9 the worst, as they are at the high 
end of the band). 

    4) Polarization calibration using AIPS was done, using antenna 28 as 
a reference.  All antennas have higher cross-polarization ('D'-terms) at 
the band edges than in the middle -- as expected.  Antennas 9 and 21 are 
outstandingly good, with cross-polarization of 1 to 2 % across most of 
the band.  Antenna 3 is typically 3 to 4%.  Antennas 2, 15 and 24 are 
notably poorer -- especially the last one -- with polarization typically 
4 to 8%.  All cross-polarization values are very stable, as the 
resulting polarization images (after calibration) are nearly noise 
limited. 

    5) Closure errors between the EVLA antennas are impressively low -- 
typically less than 0.5% and 0.2 degrees. 

    6) Images were made in all Stokes' parameters at all frequencies.  
All are close, but definitely not at, the expected noise level.   In Q 
and U, the discrepancy is no more than 30%.  Various explanations can be 
offered, but are not worth pursuing (IMHO), given the imminent arrival 
of WIDAR.  

    The data utilized here will also be analyzed by George (within CASA) 
and by Bob Sault (within Miriad).  They plan to use a full polarization 
solution (as opposed to the approximations employed by the AIPS program 
PCAL).  Time-variable analyses of the cross-polarization will be done -- 
this is a crucial parameter. 

    With luck, all results will be shown at the next test meeting on 
April 23.