[evlatests] L-Band Polarization Imaging

[Rick Perley]

    I thought it might be useful to image the last of my polarization 
test observations, which showed nearly perfect stability (other than 
antenna 11) in the antenna cross polarization.  The run was 3.5 hours 
long. 

    These observations were made of 3C147 -- an almost completely 
unpolarized (< 0.1%) source, at 10 frequencies from 1350 through 1440 
MHz, in polarization spectral line mode.  The data were bandpass and 
gain calibrated in the usual way.  I then combined the channels to form 
a 'channel 0'.  PCAL was run on each, and the R-L phases of each rotated 
to ensure the combined image would be aligned in Stokes Q and U.  (I 
arranged it so that all the polarized flux would be in Q). 

    The databases are DBCONed to form a single file, and a single image 
made.  This was used to enable one round of BLCAL, to remove small 
correlator-based offsets. 

    The result of this was a magnificent Stokes 'I' image -- rms noise = 
130 microJy, peak was 22164 mJy -- the dynamic range a sparkling 
170,000:1.  This certainly demonstrates that the EVLA and VLA can image 
together quite nicely. 

    The joint Stokes 'Q' and 'U' images were good -- but not as good as 
expected.  I then removed antenna 11, with immediate improvement -- the 
rms noises in each are then 180 microJy -- there is clearly some 
'scattered I' in the images from non-constant D terms.  But the level is 
low.  The VLA-alone image shows much higher residuals than the 
EVLA-alone, indicating there's (at least) one VLA antenna with 
significant variable 'D', or some bad data.  I'll look more closely 
tomorrow (Monday). 

    This is a vastly better result than what I found last October.  
Clearly, something has changed between then and now.  The removal of the 
4-band dipoles was done about October 20 -- a few days after the last of 
my earlier tests.