[evlatests] Baseline labels are OK
Rick Perley
rperley at nrao.edu
Thu Jul 2 15:54:58 EDT 2009
Some more trials on the data at hand show:
1) I compared the deep 0217 C-band image taken with the full array
(via the old correlator) with that taken with the 10 WIDAR antennas.
Both show a single background source. I confirm that the WIDAR position
of this is *exactly* opposite (to within one cell on the map, and it is
312 arcseconds from the calibrator).
2) I attempted imaging this field with various subsets of the WIDAR
dataset: The 5 'flipper' antennas, the 5 'unflipper' antennas, the
baselines between these two groups, antenna 1 vs. the others, in an
attempt to see if any of these subsets makes a substantially different
image. All images give the same result. The amplitude of the
background source in the WIDAR dataset is about half of that for the VLA
-- but this is nearly certainly due to bandwidth smearing, as I used a
110 MHz-wide single channel formed from the spectral line data. (I'll
confirm this with the fundamental data later).
3) I returned to the short observation of 3C273 taken at X-band. I
have an ancient plot of the visibility function, from which I must
conclude that my earlier conclusion about the visibilities being
assigned to the wrong baselines cannot be supported -- the baseline
coverage and observation duration are not sufficient to deduce with
certainty the correct visibility envelope. These data do not have a
point-source calibrator, so there is no easy way to calibrate the data.
I attempted self-calibration, but it will not converge on a suitable
timescale. (Interpretation: It will be faster to take some more data,
with a suitable calibrator.)
Thus, I conclude:
1) There is good evidence we have a phase convention backwards.
2) It seems less likely that we have scrambled the visibilities
amongst the baselines.
I need a cleaner and more extensive dataset to confirm the above. I
suggest at least a couple hours on 3C273, with regular observation of a
nearby calibrator, at X-band.
Michael Rupen wrote:
> Ken and I observed a strong source with WIDAR this morning at C band,
> using our standard WIDAR-0 setup, but strongly attenuating the signal
> from each WIDAR antenna in turn, for one minute each (i.e., 1 minute with
> ant 1 dead; 1 min with ant 2 dead; etc.).
>
> I checked for this attenuation by (1) lagfan/lagcomp on the lag frames,
> and (2) VPLOTting the data (amp vs. time for each baseline) as passed
> through to AIPS, averaging over the central two channels of IF 1 = subband 1.
> Both the lag frames (blf) and the data passed through to AIPS accurately
> reflect the changing attenuation on all baselines. Thus the labelling of
> the antennas/baselines is correct.
>
> Rick previously found that the uvw's are also correct; and his C band images
> show that we can correctly image a double at C band, at least to within a
> 180-degree flip about the origin.
>
> So why are the L band maps from WIDAR data so clearly wrong?
> _______________________________________________
> evlatests mailing list
> evlatests at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evlatests
>
More information about the evlatests
mailing list