[evlatests] Something wrong with RQ gains ...
Rick Perley
rperley at nrao.edu
Fri Jan 16 16:06:26 EST 2015
Something is amiss with the RQ gains.
In the past, application of these gains (in AIPS, using 'TYAPL')
perfectly restored the observed visibility amplitudes to constant values.
Now, using exactly the same programs, the visibility amplitudes are
only partially restored. Eric and I have confirmed that TYAPL is doing
the same as it did in the past. So something seems to be wrong with the
recorded values.
Some evidence: I chose two antennas for which the gain variations
(due to RQ changes) were fairly large. Below are the visibility
amplitudes for ea09 and ea25, along with the RQ values. These are
supposed to scale exactly (and have in the past). Each row is a
separate, sequential observation of our point-source calibrator. Units
are arbitrary. The values should be proportional to amplitude (voltage).
ea09R ea25R
Amp RQ Amp RQ
-----------------------------------------------------
18.2 13 18.9 15.5
17.9 13 18.5 15.5
17.9 13 18.9 15.5
17.9 13 18.5 15.5
17.9 12.9 19.2 15.5
13.2 11.9 16.9 14.4
8.7 10.0 7.4 11.0
11.5 10.9 11.0 12.0
14.1 11.5 15.2 13.4
14.3 11.5 16.1 14.0
13.7 12.0 16.7 14.5
8.8 10.4 7.4 11.0
9.4 10.5 8.3 11.0
------------------------------------------------------
Were the system working properly, the RQ values should scale with
the visibility amplitudes. Hence, their ratio should be constant. But
is will be immediately apparent that the RQ values, although showing the
same trend as the amplitudes, change much less. Hence, when applying
them, the correction is far short of what is needed to restore the
amplitudes to a constant value.
All IFs (and polarizations) for a given antenna behave
identically. Each antenna is different, although some (like the two
above) are similar.
This effect is seen identically at both X and C bands (data taken
the same day). And for both, the system stability was very good for the
first 1/3 of the run (as seen above), after which these rather large
changes were seen.
Either the RQ values are incorrect, or the amplitudes are. I see
no evidence for the latter -- indeed, the quality of the data is very
good throughout.
Any idea of what's wrong here?
More information about the evlatests
mailing list