[evlatests] Adventures in Imaging Cygnus A

[Rick Perley]

    I played around with my Cyg A data today, attempting to get the 
'ultimate image', orders of magnitude better than anything I did 20  
years ago.

    I failed.  But I don't think it has anything to do with the 
correlator. 

    In summary:  For Cygnus A, I followed the usual routines to edit and 
self-calibrate the data.  No 'heroic' measures were taken, just basic 
self-cal and editing of discrepant solution.  I utiilized only a single 
2 MHz-wide channel.  From the calibrator imaging (described below), I 
would expect the thermal noise to lie near 0.2 mJy/beam.  The final 
images at C and X bands (4.5, 6.5, 8.1 and 8.7 GHz, in order) all have 
about the same rms noise:  20 mJy -- two orders of magnitude too high!  
The dynamic range for these images is 2000 -- 4000:1.   My (strong) 
suspicion for the poor DR is in the imaging algorithms, which have not 
changed in essence from those I used 20 years ago. 

    But the story is quite different on the calibrator.  At X-band, the 
images (now made with 100 MHz BW) appear to be noise limited, near 15 
microJy/beam.  No 'closure corrections' were needed.  The dynamic range 
is about 250,000:1.  Not too shabby ...

    At 6cm, the initial best image had a noise of 53 microJy.  Clear 
signs of 'closure' errors are seen.  Doing a closure correction dropped 
the noise to 31 microJy.  It's clear at this frequency that Cygnus A 
(only 2.5 degrees away, and 350 Jy of hot power!) is creeping in through 
the sidelobes, despite the 100 MHz of bandwidth smearing.  The upper 
frequency, imaged alone (6.5 GHz) is much less affected (as expected) -- 
my best image here had 24 microJy noise -- close to what is expected.  
The 4.5 GHz image is quite a bit worse, but this is clearly due to 
Cygnus A.