[evlatests] A most curious end-of-scan problem ...

[Ken Sowinski]

On Sun, 31 Oct 2010, Rick Perley wrote:

>    For C and X bands *only*, the final 17 integrations (17 seconds) for
> each source were completely noise on all antennas and both frequencies
> within each of these bands.  No such effect was seen at L or S bands.
> The duration of the 'dead' data was exactly, and only, 17 seconds.

Somewhere along the way instructions to the correlator
to change net sideband are executed in anticipation
of the event.  We are planning to do this differently
in the near future and if that does not fix the problem
it will be debugged then.  Meanwhile, observations which
change net sideband should make the preceeding scan a
little longer than otherwise needed.