In recent 12.5 MHz bandwidth spectral line data the noise level is too
high by a factor between 2 and 3. The only exception are observations
at mode PA, where the cross-hand pairs show the expected noise, but
parallel hand data show the increased noise. This occurs at all bands
and affects all baselines, whether formed by VLA antennas, EVLA
antennas, or either.
When did this start?
We are aware of data taken as recently as December 2006 which do not
show this increased noise, so the problem must have started after
that. This appears to exonerate the new correlator controller which
was deployed in September, 2006.
Initial tests appeared to indicate that the problem goes away when the
so-called correlator self-test is turned off. We have been unable,
however, to confirm the results of those tests: the increased noise
remains regardless of whether self-test is on or off. Other
bandwidths do not show this behavior; we therefore strongly suggest
observers avoid observing at 12.5 MHz bandwidth.
Since the problem is unrelated to the type of antenna (VLA or EVLA),
we believe this problem is related to the VLA correlator, and will go
away once the WIDAR correlator is deployed.