[evlatests] Polarization Accuracy/Stability
Rick Perley
rperley at nrao.edu
Thu Nov 17 11:07:56 EST 2011
The PCAL solutions made yesterday reveal puzzling differences in the
antenna polarizations between the three paths. Curiously, the B0D0 path
(8-bit) antenna polarizations are much more similar to the A1C1 path
than the A1C1 is to A2C2! In comparing these solutions, we can see the
same basic structures in most cases. In general, the cross-polarization
phases are more similar than the amplitudes (something which I find very
odd).
Given that these solutions were made with a limited timespan of data
(about 2.5 hours), the parallactic rotation (needed to permit separation
of antenna a source polarizations) is limited -- we might suspect the
solutions are faulty. If so, the three parallel paths might give rather
different results for the source polarizations.
Not so! Overplotting the three sources polarizations for the three
independent paths shows astonishing similarity (actually, identicality)
of derived source polarizations -- both in amplitude and phase (position
angle). There are no differences to within the noise. We might
interpret this as confirmation that the PCAL program has found correct
solutions (to within the model it uses for these solutions).
There appears to be a small but significant linear polarization in
3C84. It averages (over the 14.0 -- 16.0 GHz frequency span) at about
0.4%, with many interesting structures where the polarization rises to
nearly 1% over 50 or so MHz spans. The PA is fairly constant over the
frequency span (I don't know the true angle since we didn't include a
polarization calibrator for this run).
J0303+4716 (the 'medium' strength calibrator) is about 1.5%
polarized, and also shows variations in its fractional polarization over
the frequency span.
J0349+4609 (the 'weak' calibrator) has very low polarization --
about 0.5%.
Another, much longer run was taken last night. This will allow us
to (we hope) deduce any time variability in the antenna polarizations.
We will also be able to compare the new solutions to those derived
yesterday.
More information about the evlatests
mailing list