[evlatests] phase shifts caused by frequency differences gone

[Gustaaf van Moorsel]

During the Monday morning EVLA meeting Ken reported that the new executor
should address the phase shift problem caused by small observing frequency
differences, where 'small' means typical when Doppler tracking.  The exist-
ence of this problem made it impossible to apply phase gains determined on
a calibrator to a nearby source, which prompted us to urge observers not to
use Doppler tracking under any circumstances, not even when using only VLA
or only EVLA antennas.

Yesterday I ran a quick test to confirm that it now works.  I repeatedly
observed a series of 4 calibrators at L-band, separated by typically 15
degrees to simulate a typical source/calibrator pair.  The result is that
phases are now indeed stable: none of the previously observed phase shifts
were detected, so it appears the fix was successful.

Doppler tracking on these sources causes typical observing frequency diff-
erences of 16 kHz, or 3.4 km/s, which under the previous executor would
have been more than sufficient to detect phase shifts.

Note that this does not address the fluke problem on VLA - EVLA baselines,
but will allow Doppler tracking on pure VLA or pure EVLA subarrays.

Finally, antenna 1 is the odd one out - phases on baselines including this
antenna were not stable.  Ken informs me that this might be due to poor
synchronization, and that results of the stress test should confirm this.

Gustaaf