[evlatests] WIDAR phase transfer

[Rick Perley]

    Michael took 4 hours of data at X-band on a northern calibrator 
(0017+815), and a nearby weak point source, 0207+841, offset by 2 
arcseconds.  The idea was to use the former to calibrate the latter. 

    Short Answer:

    Phase transfer works!  Quite nicely, in fact.   We got a 
recognizable image of the target, in the right place. 

    Some Details:

    1) A large delay error had to be removed:  393, 343, and 224 nsec on 
antennas 2, 3, and 4, w.r.t. antenna 1.  These were the same on all 6 
sub-bands.  I determined these values from a short observation near the 
middle of the run.  I have not checked its constancy throughout the four 
hours. 

    2) The first 12 seconds, and the last 1 second of each calibrator 
scan had to be flagged.  No surprises here, as the system flagging is 
not being applied. 

    3) Not knowing the flux density of the calibrator, I took it to be 1 
Jy.  The six sub-bands (all at different frequencies) then gave the 
following average gains (a lower number means a more sensitivity sub-band)
  
                    Sub-band         Freq.              Gain
--------------------------------------------------------------
                         1                 8808 MHz        10.7
                         2                 8680                   9.2
                         3                 8552                   8.3
                         4                 8424                   9.4
                         5                 8296                 12.1
                         6                 8168                  11.9
----------------------------------------------------------------

    This is about as expected -- the central bands are the most 
sensitive.  The only surprise is that sub-band 6 is as good as it seems 
to be ...

    4) Bandpasses were determined, and look normal.  I made a single 
solution, averaging all the calibrator data.  No attempt to track 
sub-band bandpass changes (in time) was made. 
   
    5) Amp and Phase calibration, on a 1-second interval, was 
determined, using the calibrator.  The amplitude stability was superb -- 
except for a few weak points on sub-band 6 for antennas 3 and 4 (23 and 
24).  Phases clearly connect from scan to scan, although some small 
jumps cannot be excluded. 

    5) The data were then 'collapsed' to a pseudo-continuum, applying 
the delays and bandpass function.  I used channels 50 through 1000 for 
this. 

    6) The two sources were then split out (calibration applied) and 
imaged, initially using sub-band 3 ONLY! 

    a) Calibrator:  The map peak was 1.000 Jy (as it should be!), and 
the rms - .0002 Jy (DR ~ 5,000:1). 
    b) Target:       Peak - 71 mJy.  The source was located 2 arcseconds 
west of the phase center -- as it should be.  The map looked rather 
cruddy, indicating phase errors, probably dominated by atmosphere.  I 
then did a phase only self-cal, using the image.  The peak rose to 102 
mJy, and the rms was .00026 Jy, but the residuals were less noise-like 
than I expected.  I then ran an Amp&Phase self cal, 'moving' the source 
to the field center.  The peak of this map was 102 mJy, and the noise 
was now very uniform and 'noise-like', with amplitude .000069 Jy (DR ~ 
1480:1). 
    At this point, I made an image with sub-bands 3 through 6.  Ideally, 
if each is equally sensitive, the noise should drop by a factor of two.  
In fact, it dropped by a factor of 1.53 -- which I think is about 
accounted for by the lower sensitivities of sub-bands 5 and 6.