[evlatests] C/X Phase test, Oct 3/4.

[Ken Sowinski]

On Fri, 6 Oct 2006, Rick Perley wrote:

> B)  VLA antennas.
>
>    A new and very troublesome phenomenon was immediately found -- Some
> antenna/IFs have non-repeating phases (after having changed band).
>
>    The effect was especially noted at X-band, but was present in C-band
> as well.   I emphasize that the EVLA antennas did not see any of the
> phenomena described below.
>
>    At C-band, antenna/IFs 12AC and 20AC showed multiple phase states --
> the phases are stable within each 2-minute observation, but the observed
> phases were as much as 80 degrees away from the 'base' state.   This
> effect was especially bad in these two antennas, but 'single' versions
> of this occurred in two others:  15AC, 25BD,   In addition, phase
> 'spins' occured in some scans in many other VLA antennas:  3, 7AC, 9AC,
> 10AC, and 11AC.
>
>    At X-band, the situation is worse.  3BD has a bistable phase, with
> phase spins between them.  Multiple (but mostly stable within a scan)
> phases are seen on 5AC, 6BD, 8BD, 9AC, 12AC, 20AC, and 27AC.
>
>    The consequence of these unsteady phases upon VLA imaging is likely
> to be significant.

Several test observations today were unable to reproduce this
problem.  I have written elsewhere why this is very hard to
explain by asking the correlator to do something wrong.  There
was some suspicion that a test version of the Modcomp program
G10 may have had something to do with this but observations today
with and without the new version displayed no problems at all
except for the unsurprising fact that it is hard to trace phases
when the weather is very bad.

It is not clear to me what can be done if we cannot get the
problem to recur.  I can only wonder if the test version of
G10 found itself in some state which caused it to misbehave.
However it is very much a 'stateless' program so that is not
a very satisfying explanation.