[evlatests] Sys Cal saga.

[Vivek Dhawan]

The state of play: Data were flowing till the power outage. Flow has resumed,
but it is bit unclear if the data are now correct. They appeared more correct
before the power outage. Various oddities need looking into.  In summary, its
getting close, but not yet ready for prime time.

Details:

SY (syspower = Psum, Pdif, and PGN= requantizer gain setting) and CD (caldevice)
tables come through in a painless way, via BDF2AIPS. They behave in good and bad
ways, some listed here. No doubt more quirks lurk.

Viewing and plotting are available, with new features being added daily by Eric
and Bill Cotton. The UV flags can be applied to the SY table data with SNEDT,
this works well and shows that the times are correct etc.


CD table data: view with PRTAB.

For OSRO-1 (1 subband per baseband, 2bb, 2 pol) the tables are populated
as follows: IF1 and IF2 appear different, pol1 and pol2 are identical.

For RSRO data (8subbands per baseband, 2bb, 2 pol) all the subb are identical
and 2 pol are also identical. This feature is present in the ASDM table, not
caused by BDF2AIPS. (Thanks to Josh Marvil, Steve Myers for checking this
in the ASDM)


SY table: Viewing and plotting are available, with new features being added
daily by Eric and Bill Cotton. I used PRTAB and SNPLT quite a lot.

The UV flags can be applied to the SY table data with SNEDT, works well and
shows that the times are correct etc.

Most antennas (OSRO-1) data look reasonable, but there are small and big 
exceptions. I looked mostly at C band data, so ea13 has zero Psum; but so
does ea10, which works fine at C band.

Psum for some channels wanders more than others, even on the same antenna.
This is obvious when there is RFI present, but is seen even on the default
settings at C band when RFI is not.

I expected the (rms/mean) on Psum to be 1/sqrt(2*128MHz * 1sec) or (1/16000)
without systematics, perhaps 1/(few thousand) with typical total power 
stability; and the noise on Pdif to be about 30-50 times larger, since the
Tcal is a few % of Tsys. The measured noise is about 1/2000 on Psum and 1/200
on Pdif.

Pdif for a given channel should be independent of elevation, Tsys changes,
etc. Most but not all channels are stable in this way.

Psum vs. elevation should show the increase of total power from opacity and
spillover, and does. (Psum/Pdif) prop to  (Tsys/Tcal) is more uniform across
the different antennas and subbands, as it should be, so the application of
the calibration should work quite well. It remains to try it and compare with
3C286 or 3C147.

As mentioned in the last meeting, the (Psum/Pdif) is unperturbed by 1db step
changes in the gain path, as expected.

Enough for now - Vivek.