This is another test of the polarisation solutions with the LBA. We are observing 3 targets (3C273 3C279 & 1921-29) with well known VLBA polarisation results, and 1934-69 (which should have zero polarisation). We wish to confirm that the polarisation results from the LBA match those from the VLBA.
There are a number of important polarisation related issues.
Of course the most obvious one is the polarisation purity. Please make a special effort to calibrate the feeds. 10% _purity_ should be considered the minimum requirement. This is not the same as 10% isolation between the two polarisations, as it refers to error from the true polarisation. The real time fringe checking should look to see the level of the correlation between polarisation L&R at ATCA (which I will assume is perfect) and the polarisation R&L at the other sites.
Observing comments for each antenna:
VX014 observed 3C273, 3C279, 1921-293, & 1934-638 at 8.4, two pols, two IFs.
Parkes, Mopra, Ceduna, Hobart, ATCA observed.
ATCA has a pecular pattern in the rates. See figure RATES.VX014.PS Is has an amplitude of about 2mHz and a cycle of 1 hour. I wonder if it is to do with the array calibration? Is that on that time scale?
CEDUNA has awful phases and poor SNR. The delays are higher and more changeable. The changes in the phase and delay difference between the IFs and the polarisations are significant with times scales of about 1 hour. See DELAYS.VX014.PS. There was some data loss. There is no Tsys data.
MOPRA polarisation was reported as very unstable. However it seems to be stable, but it is very high at 40%! (this will break the linear solution model used by AIPS.) The default tsys data has an incorrect index. Furthermore the values in the log file (100 for R and 30 for L) are not reflected in the data (as R and L are the same amplitudes).
PARKES is has, at times, a relatively high phase between RR and LL. See RRLL.VX014.PS
HOBART ... no problems! E-W Solutions worked.
ATLOD forces the AN table to be mount type 0, Polarisation Types X and Y (for ATCA). This needs fixing. It can be corrected inside AIPS with TABED.
After parallel delay, phase and rate solutions, & the left to right delay and phase solutions, the bandpasses look reasonably sensible. The parallel hand phases and amplitudes are flat with zero phase, the cross hand phases are also flat and the phases are to be solved in the polarisation calibration. It would be reasonable, if Ceduna was not such a mess, and if Mopra was not so large. See POSSM.POL.PS
A new version of logs2keyin was made (1.2.14) which can handle all the formats around (i.e. I added the new version of Parkes). The INDEX keyword could be improved as it still assumes that this is for VSOP and labels the columns as L1 and L2. The columns should read something like (I believe):
TSYS PKS TIMEOFF=0.0 FT=1.0 INDEX='R1|R2','X','L1|L2','X' /
Total power imaging was straight-forward. Ceduna had a jump in gain (an improvement) which is why it would be good to have the Tsys values. Was this an improvement in the Tsys -- or a improvement in the sensitivity?
All images match what would be expected.
Calibration of the polarisation D-terms, based on the Stokes I images, with LPCAL. See Mopra_D-Term_resid.ps for the pre and post fit of the D-terms to be Mopra data (on the complex plane). Ratio comes from ~0.5 to 0.05. Still not very good, not surprisingly, but better than I expected.
LPCAL works fine, although the residuals will be large. D-Term_Solutions.eps is the IF1 and 2 (connected line) for R and L for each antenna (based on 3C273, but they are a few percent from each other based on each source, and polarised/unpolarised).
- All solutions agree (Pol/Unpol <0.5% & Source A/B ~1.5%) resonably.
- Average D-term magnitude: CAT 3% Cedna 4% Hobart 9% Mopra 35% Parkes 1%
- IF 1 to IF 2: ~ 1.5% difference Right and 3% difference L. Both cases dominated by Mopra.
- Ceduna L, IF 1 and 2 are quite different (4%), which relates to the awful Cross-Pol bandpass.
PCAL, which is not recommended for VLBI, does not work very well. Further work required.
Finally the results are presented in pcln.eps, which shows 3C273, 3C279 and 1921-293 & 1934-638, which has no detectable polarised flux. Shown are the LBA results (B&W) and the MOJAVE results at similar epochs, where they exist. The same regions are showing polarised flux.