[evlatests] X-Band Hi dynamic range imaging

[Rick Perley]

    As noted this morning, I used 4.5 hours of time last night to do 
some deep-ish imaging at X-band. 

    Mode used was 2BD, BW = 25 MHz, giving 16 channels in each 
polarization. 

    We started with 9 EVLA and 18 VLA antennas, but EVLA ant 24 dropped 
out early, for no known reason. 

    About 3 hours' duration was spent on the point calibrator, 3C286, 
and 1.5 hours on a randomly selected blank field, located 3 degrees away. 

    Bandpass and gain calibration were done in the usual way.   A very 
strange problem with the correlator was found, where by particular pairs 
of antennas showed oscillatory bandpasses for some whole scans (an 
erroneous lag value, I presume).  The affected antennas were 11 and 22 
in LCP for three scans, 28 and 13 in RCP for four scans, and 6 and 17 in 
RCP for one scan. 
This is clearly not an EVLA issue. 

    After essential flagging, the spectral line data were collapsed to a 
'pseudo-continuum' file.   A last round of amplitude and phase 
calibration was done.  The data were then split into two single-source 
files. 

    Imaging:

    3C286 was imaged with all antennas, giving a 35,000:1 ratio between 
the peak (5.2 Jy) and the nearby rms noise.  The noise structure clearly 
indicated 'closure' errors.  BLCAL was run, averaging over all times to 
give a single solution for each baselines.  Corrections are small 
(typically 1E-4), but certainly help the image.  The DR rose to 63,000 
for nearby blank field, and 110,000 for a distant blank field. 

    Maps were also made with:

    b) EVLA antennas only.  Effectively, only 8 antennas, and 28 
baselines.  The DR = 39 ,000 for distant noise, 31,000 for nearby.
    c) VLA only.  18 antennas, and 153 baselines.  DR = 95,000 for 
distant noise, 72,000 for nearby. 
    d) X-baselines (VLA x EVLA).  144 baselines.  DR = 55,000 for 
distant noise, 31,000 for nearby. 

    For all four images, a background source located 10 arcminutes away 
(beyond the -6dB point!) was seen, with a flux of 1.4 mJy. 

    Imaging was also done of the blank field.  I was betting I'd hit a 
random background source -- I was right, sort of. 

    a) With all antennas, the rms was .040 mJy, and the background 
source was 0.42 mJy.
    b) With EVLA only, the rms was .14 mJy, and the background source 
was not clearly obvious in the noise peaks.
    c) With VLA only, the rms was .05 mJy, and the background source was 
visible.
    d) With VLA x EVLA baselines, the rms was .057, and the background 
source was visible.

    While weak sources do indeed show up as they should with the 
EVLA-only, and the EVLA x VLA baselines, the impact of the EVLA is not 
as great as it should be.  The reason is likely the extremely poor 
sensitivity of two of the EVLA antennas.  Antenna 21, and antenna 14D 
have AIPS weights 1/4 the correct value -- their effective system 
temperatures are double (or the efficiency half) of what they should be.