If you have XX and YY polarization data only, you can ignore (gratefully) this section.
A thorny part of calibrating ATCA data concerns the XY phase
differences (
). This is the phase difference between the X
and the Y feeds on each antenna; in principle they should be steady in
time. If we did not know 
, the phase relationship between
the XX, YY, XY and YX correlations would be arbitrary and we
could not combine them to make the Stokes parameters. Thus, 
is used to rotate the phase of the Y gain to that of the X gain, so
that the XX and YY visibilities are in phase (this gives the XY
and the YX visibilities the correct phase relationship as well).
is measured for each antenna every integration via a noise
source in each feed horn which is switching on and off continuously
throughout your observations. Before November 1992, this measurement
was quite poor because it detected the on plus the off signal. Any
ambient signal (and there is plenty, including your source and ground
pickup) would contribute to the measured value of 
. Since
this time, the difference between the on and off signals is now detected
and the quality of the 
values has improved significantly.
However, despite this improvement, they are still not of sufficiently good quality to use blindly. This is especially true in the 13 and 20 cm bands where deviations which do not appear to represent actual deviations occur.
If you are to be concerned with 
values, you will be
calibrating your data in MIRIAD. The calibration program solves for
for each antenna relative to the reference antenna.
Therefore, it is only important that you get a good measurement for one
antenna. Usually, the on-line reference antenna has the best
values, but this does not have to be the case. It also
turns out (see Sault, Killeen, and Kesteven 1992) that knowing the
absolute 
on the reference antenna is unimportant for many
observations so that these relative solutions are usually sufficient.
Currently, our preferred path with regards 
values is to
deal with them entirely in MIRIAD. It is still possible to apply
them with AIPS ATLOD as current users will be familiar with, but
since MIRIAD offers greater flexibility we would like to isolate
application of 
to that part of the process. Please refer
to the MIRIAD manual for details. You should be aware of a
convention difference between AIPS and MIRIAD for 
.
A formula to convert is given in the MIRIAD User's Guide.
Last update: 11/04/95
1