[evlatests] More details on phase winding from Dec 09, 2005 data

[Mark Claussen]

   Here is some more detail about the phase winding that we saw
   on Friday Dec 9, 2005.    Plots can be found at

   http://www.aoc.nrao.edu/evla-internal/techdocs/testresults/20051209.mc

To summarize, the goal was to use different combinations of LO settings
for the SAME sky frequency, so as to change/eliminate the amount of
aliasing in the 4GHz bandpass filter before the digitizer.

Aliasing was reduced for some settings, as reported on Monday by Ken 
Sowinski, but 2 other problems were provoked.

1. The changeover to a new scan happens a few seconds before its
expected time.

2. The L302s at one or both antennas seem to stop updating the fringe 
rotation, or get the frequency wrong. Sometimes they recover, sometimes 
not. At these times, the fringes on EVLA-to-VLA antennas disappear, but 
fringes at fairly high rate ("phase winds" in Rick Perley's terminology) 
are seen between 14-16.

More details follow.

   C-band continuum:

   We stepped through L301 settings 13184, 13440, 13696, 13952, 14208,
   14464, 14720, and 14976 MHz.  Each setting was observed for about
   2 minutes.  In the plot vplot1.ps, one can see that there were fringes
   on the EVLA-VLA baselines, for the first 4 settings.  For the next
   three settings, fringes on the VLA-EVLA baselines went away, but 
   recovered for the last setting.

   Phase steps are seen from scan to scan on antennas 14 and 16,
   during the time they were nominally 'working' - the first 4
   scans. The changes occur not at the scan boundary, but a few
   seconds before the scan boundary.

   In the next three scans, the phase on the 14-16 baseline winds quite
   dramatically  - and fringes to the VLA are gone.  The fringe rates 
   for these three scans correspond to sky frequency errors of 256, 
   512, and 768 MHz for the three scans - as if the L302 is doing the 
   fringe rotation for the wrong frequency.  Our guess is the L302 is 
   doing the fringe rotation based on one of the internal LO's (which 
   have been offset) not based on the sky frequency (which was unchanged).

   At the final scan here at the C-band continuum we see that the fringe 
   rotation is again correct and fringes are once again seen on VLA-EVLA 
   baselines (although there appears to be a slow phase slope on the 
   VLA-EVLA baselines).


   K-band continuum:

   We stepped through L301 settings from 13184 to 16256 MHz, comprising
   13 scans.  See vplot2.ps.  There are no fringes on the VLA-EVLA baselines 
   for any scan. There are fringes 14-16 baseline with a phase wind whose 
   slope increases monotonically; the fringe-rate starts near zero and 
   decrease to about -20 mHz.  This behavior, while possibly related to what 
   we saw in the C-band continuum, is somewhat different.  It's as if the 
   fringe-rotation is nearly correct at the start, but the L302 has stopped 
   updating, and errors are building up.


   C-band line:

   We repeated the C-band continuum experiment except using line mode 4, 
   with 6.25 MHz continuum.  vplot3.ps shows the channel 0 phase plot, as 
   above.  Again, there are no fringes on VLA-EVLA baselines, and the phases 
   are winding on 14-16 baseline.  Note that the phase wind is similar to 
   what was seen at the K-band continuum although the phase slope is now 
   decreasing in magnitude (from about -20 mHz to -10 mHz).

   K-band line:

   Finally we repeated the K-band continuum experiment except using line 
   mode, as above.  vplot4.ps shows the channel 0 phase plot.  Again, there 
   are no fringes on VLA-EVLA baselines; on the 14-16 baseline we see the 
   fringe-rate start at about -10 mHz, go through zero and on to about +10 mHz.

   The effects of the last three groups of observing (K-band continuum, 
   C-band and K-band line) suggest that the L302s (or perhaps just one antenna) 
   lost what it was doing and never recovered.  So perhaps, for some reason, 
   commands weren't getting to the L302.  The first problem, though manifested 
   differently, could be the same problem --- fringe rotation commands not getting 
   to the L302.

   Mark and Vivek