[evlatests] Correlator offsets signals

[Ed Fomalont]

REPORT on CORRELATOR OFFSET EXPERIMENTS
Ed Fomalont
Sept 30, 2007

The brief report given on Sep 17 at the EVLA Monday morning was
incorrect.  Please ignore.


---OBSERVATIONS AND REDUCTIONS:

    Only 4IF mode (RR+LL, Two IF's, 7 channels each) is reported here.
The following observations, each about 4 hours, were


     Sept 16:  IF1  1365, 1465, 4885  intermixed
               IF2  1435, 1365, 4835  intermixed
     Sept 23:  IF1  1365                (AM889)
               IF2  1435
     Sept 24:  IF1              4885
               IF2              4835
     Sept 29:  IF1  1365                (AM889 clone)
               IF2  1435

No calibrations were made, but all interference was removed.  The raw
data counts are given below.  They should be multiplied by about 10 to
convert to Jy.  The slight contamination at 1.4 GHz from faint sources
in the field of view is only a minor factor at the shortest baselines
that have generally been ignored.  The short baselines were also 
affected most by possible lingering interference.


---CORRELATOR OFFSET SIGNALS CONSTANT OVER THE TWO WEEK PERIOD.

    The 'correlator offset' signals for each IF and polarization
remained remarkably stable over each day and over the ENTIRE two-week
period.  The offsets were independent of the observing frequency.

    The ONLY EXCEPTION WAS for the Sept 24, 4.8 GHz observations when
no significant offsets were observed.  What was difference?

     ONLY VLA ANTENNAS WERE AVAILABLE, YET THE MAJORITY OF CORRELATOR
     OFFSETS ARE ASSOCIATED WITH VLA ANTENNAS!


---OFFSET SIGNALS LIMITED TO A FEW ANTENNAS

   IF1-RR was the worst with about 40 correlators with significant offsets.
   IF1-LL, IF2-RR, IF2-LL had about 15 correlators with significant offsets.

   A significant offset is an offset signal more than about four times
that expected from the raw data average of 0.001 raw count average,

   The offsets were predominately associated with a few antennas.

     IF1-RR:  22 and 17 were bad  (3,9,15 fair)
     IF1-LL:  10 was bad, 17 fair
     IF2-RR:  13, 17 were bad
     IF2-LL:  8 was bad

     Antenna 22 was moved from W73 to W36 after the Sept 16 
observations.  This had no effect on the correlator offsets with antenna 22.


---MULTIPLICATIVE CLOSURE ERRORS?:

    In one observation, the offset signals were removed from all
sources and the number of multiplicative closure errors from an 0.5 Jy
calibrator during the antenna-based calibrations were significantly
reduced.  Hence, there is no indication of additional multiplicative
closure errors.


---CHANNEL TO CHANNEL VARIATION OF OFFSETS:

    The channel to channel offset terms were also constant over all of
the sessions, WITH THE OFFSET IN CHANNEL 1 dominant.  The ratios of
the channel offsets were typically the same for all baseline, IF's
polarizations:

          Chan         Relative offset
           1               1.0
           2               0.1  (uncertainty of 0.1)
           3               0.4         "
           4               0.2
           5               0.1
           6               0.1
           7               0.1

The phase between baselines and channels had no obvious relationship.


---FURTHER ANALYSIS, OBSERVATIONS?

    At the EVLA Monday morning I can show lots of plots, but the above
summarizes the properties of these correlator signals.