[evlatests] Results from yesterday's full-polarization test on 3C147 at L-band

[Rick Perley]

    I have taken the calibration of the full-polarization observation of 
3C147 made yesterday about as far as it is worth.    The key results 
(other than those reported yesterday):

    1) The quality of the data varied between the polarizations -- RCP 
was much better than LCP.   In summary, the only RCP data which needed 
flagging were those records affected by the periodic 10-second drops.  
In LCP, about half the data had to be removed to obtain stable gain 
solutions. 

    2) The RCP gains were quite stable throughout the 2-hour 
observation.  In LCP, all antennas had a large (up to 8 dB in power!) 
gain change after the first scan.  Two antennas (18 and 24) had another 
gain change  about 12 minutes later.   The amplitude ratios show that 
these changes are not likely due to T304 attenuator changes. 

    3)  Antenna 1 -- only -- showed a 10 degree sinusoidal phase 
oscillation, with period 45 seconds, for about 20 minutes.  This 
happened in both subbands, and in both polarizations.  No other antenna 
showed any similar effect.  There is no hint of this in the amplitudes! 

    4) Images of 3C147 revealed the following:

       RCP.  The 'lumpy-bumpy' phenomenon has returned.  The image is 
quite poor, but the (weak) background sources are visible.  A closure 
correction was successful in removing the 'lumps and bumps', showing the 
background sources quite well.   The amplitudes of these background 
sources do not match well my 'gold standard' VLA image, however ...
       LCP.  The image is just appallingly bad.  The 'lumpy-bumpy' 
structure dominates all, with the rms noise 12 times higher than the 
(non-closure-corrected) RCP image.  The closure errors are obvious in 
the visibility plot, with amplitudes of some correlator more than 13% 
too high -- these are non-closing errors.  There is, however, no 
corresponding error in the phase -- the non-closing effect is entirely 
in amplitude.  The visibility plot shows these amplitude offsets are 
constant in time.  Sure enough, a closure gain solution (BLCAL) nicely 
removed these errors, and the resulting image was fairly decent. 

    5) Michael observed 3C147 with an offset of an arcminute -- these 
data are also affected by the amplitude drops.  I did not attempt to 
analyze this source.

    6) Michael also observed at a place named '0604+442', which I 
thought was a blank field.  It is not -- it's a 750 mJy point source, 
located one cell (0.5") from the phase center.  (Was this intended, 
Michael, or were you just unlucky in your random guess?) 

    7) The cross-polarization data (RL and LR) look quite good (after 
flagging), with fractional polarization of about 5% -- this is the 
expected antenna polarization.  The values are constant in time, as 
expected. 

    I'll add here that flagging these data is *extremely* tedious.  
Getting rid of these periodic drops has got to be priority #1!