[evlatests] Results from Imaging in Ka-Band

[Rick Perley]

    Still being ignorant about JIRA, I continue with this old-fashioned 
method ...

    After removing the non-fringing and insensitive Ka-band antennas, I 
calibrated the data using standard 2-point interpolation (1 min in 
amplitude, 10 sec in phase). 
    The bandpass was solved for each 1-minute on-source calibration, but 
I used the overall average when imaging. 

    Imaging results:

    a) The sky is the right way up.  The off-axis calibrator 
(J0157+7442) popped up exactly where it really is. 

    b) The noise in the blank field integrates down exactly with number 
of IFs ('subbands') and number of channels, as predicted by the simple 
noise equations. 

    c) The noise on the strong calibrator (when calibrated as above) 
does not decline with number of channels, as naively expected.  The 
noise is about a factor of two too high when I used 57 channels, 
compared to a single channel.  The amount of 'degradation' increases 
with increasing number of channels. 
       However, when I SPLIT out this calibrator source, integrating the 
57 channels together to make a 'continuum' database, then run CALIB on 
this, the resulting image *does* exactly meet the expected noise.  The 
'final' dynamic range (114 MHz BW, both parallel polarizations, and two 
IFs) is about 20,000:1, and appears (to the eye) to be close to noise 
limited.  Not bad for only about 8 minutes on-source integration, and 
about 17 antennas.