[evlatests] Closure errors by band.

[vdhawan at aoc.nrao.edu]

		EVLA tests of 2005 Nov 23.

Rick Perley reported from data of Nov 18-22 that there appeared 
to be closure errors i.e. amplitude & phase variations on the 
baseline 14-16, which were less apparent or absent on baselines 
14-VLA and 16-VLA. C-band data had a time-variable closure error 
on ~4s timescales, X-band was steady.

The goal here was to see if these were real/repeatable. I stuck
to a single point source and did not investigate the elevation-
dependent phase offsets that Rick also reported before.

Continuum data were taken at 3s integration, at X, L, K, C bands.
 
Line data, 10s integration, were taken at 50, 25, 12.5MHz BW, IF A, 
for C, K, L bands. At X-band, IFs A,B,C,D were recorded at each of 
the 3 bandwidths.

I report mainly on the continuum data here.

1. Some data may have been lost due to interaction with another 
   subarray, I have not investigated the details yet; but all IF's 
   and all bands produced enough good data.

2. No rapid phase wraps or amplitude dropouts were seen. Antennas
   14 and 16 were as good as VLA antennas, to 0th order.
 
   Deviations are noted below. 

3. There ARE what appear to be closure errors, steady over ~10
   minutes at X and L bands, and variable on ~4s at K and C bands.

   Differences between the 4 IF's exist at a lower level, I have
   glossed over them here but kept plots and notes.

4. The combination of LOs and sidebands in use at the different RF 
   bands, along with the matrix below, may be useful in diagnosis. 
   I'll need some help from a hardware expert.

Band	amp 14-16	amp 14-VLA	amp 16-VLA	Comment.

X	0-20% HIGH	+_10%		+_10%		steady

L	0-10% LOW	0-10% LOW	0-10% LOW	steady

K	0-10% HIGH	= VLA		= VLA		jitter

C	10-30% LOW	5-10% LOW	10-30% LOW	jitter

The ranges are the mean values from different IFs - the jitter
is peak-to-peak deviation from the mean.

The clearest case is C band, where the jitter is invisible on
16-VLA, 14-VLA, and VLA-VLA, but obvious on 14-16. A phase
jitter of about the right amount, 5-10deg,  accompanies the 
amplitude jitter.

At K band the jitter is worst on 14-16, but still visible and 
maybe IF dependent on 14-VLA.

I need to do more on this, e.g. look at TSYS correction? Aliasing? 
Frequency dependence across the bandpass?  These can be investigated 
with the data in hand; I will send an update after I chew through the 
spectral line data - a dry turkey, without wild turkey to wash it down.

Cheers, Vivek.