[evlatests] a little progress

[Ken Sowinski]

This was a more useful day than Tuesday though the beginning was
spent recovering from things that did not work after the power
returned.

For the record:
Antenna 21 was out of sync; the L305 sync_enable did the trick.
The DTS module for 26D was in a funny state and had to be reset.
The CMP was not processing commands and had to be power cycled.

A few minor things were learned.  The -M switch for the executor
did not seem to remove VLA antennas.  The operator could not
find a way to modify the T5 bandwidth selection on the new
naseband display which works through the CMP.

Because I was so confused Tuesday the plan today was to switch
to the executor more gradually.  This made it easier to find
and fix the worst pointing antennas and the largest delay errors.
After a while things were stable enough the the executor and idcaf
were firmly in control and data looked reasonable.

Pointing is now good enough at L and C band; but we lose a handful
of antennas at X band and even more at K and Q.  Delays are now
within a few 10s of ns instead of a few 100s of ns for all antennas.
After the pointing run tonight I will import the refined pointing
model for all the moved antennas from the Modcomps.  There is a
file of delay measurements at L band that can be used to improve
the delays a little.

By the time this was done I was out of time and energy.


Rick ran his file which causes the VLA and EVLA to decohere at
L band and the problem was determined to be that the BD Flukes
are misset at L band if the BW is less than 50 MHz.  Empirically
this does not happen at other bands.  A casual look at VLALoIfSetup
has not yet revealed the problem.  The four Fluke settings at
12.5 MHz BW were: AC 118.75   115.625
                   BD 109.375  150.25!


Now that things are more stable attempt can be made next week to
create problems by using two subarrays.