I've been looking at calibrating some observations from mid-2011, and the first thing that I did was to split out the 1934 observation and plot the XX and YY product amplitudes as a function of time. It was immediately obvious that there is significantly more scatter in the ant 3 baselines for the YY product than for all the others. This suggests to me that there was something wrong with the Y feed on this antenna during the observations.
I want to know whether this information alone is enough to suggest a definite course of action at this stage in terms of flagging. In other words, do the plots suggest that I should just flag antenna 3 entirely (giving up on it for the entire set of observations), flag the worst-affected baseline (1-3) and push forward with the bandpass cal, or not flag anything at this stage and see if it seems to affect the bandpass calibration too badly?
Or is there not enough info in these plots alone to suggest any definitive course of action yet?
Cheers,
Craig.
Screen Shot 2013-02-15 at 10.50.03 AM.png (108.13 KiB) Viewed 19987 times
Screen Shot 2013-02-15 at 10.47.39 AM.png (111.58 KiB) Viewed 19987 times
Screen Shot 2013-02-15 at 10.47.31 AM.png (108.41 KiB) Viewed 19987 times
I would bet that the following scenario is happening.
You could be using the CABB 1 MHz zoom profile. In this case, there is a known problem on antenna 3 where the channels in the continuum band corresponding to block 8 (or block 28 if you're looking at IF2) alternate between good and bad with the cycle period. This can be enough for the correlator machinery to do gain scaling for those products involving that antenna (if these channels are in the tvchannels).
Even if this is not the case, there are a couple of things you can do.
The first is simple: it looks like you're looking at the continuum at 16cm. In this scenario, it is likely that the other IF (the one you didn't load) isn't affected (block 8 and 28 are not often affected with this problem at the same time). Load the other IF and see if you have the same problem.
The second is simple too, but may lead to other problems. The gain scaling that the correlator does looks at the noise diode. It assumes that the noise diode has constant power, and it scales the output of the correlation to keep this assumption. There are any number of reasons why this may fail (including the scenario outlined above), but mostly it works well. But you can reverse this action during the atlod stage by using options=notsys. This may fix the problem.
OK - thanks Jamie. I'll try it and see how I get on.
Out of interest, how stable is the ATCA bandpass in general? Could I potentially use the 1934 observation from the previous day instead if it is unaffected?
Actually - while I'm on the topic I just looked at some observations of 1934 from a different epoch (1.5B config), and on the short baseline I'm getting this:
Screen Shot 2013-02-15 at 11.52.10 AM.png (48.71 KiB) Viewed 19983 times
This doesn't look like the same problem to me - it looks like actual fringes. Could this be the Sun or something sitting in a side lobe of the antennas? It seems more prevalent on short baselines, with a lower amplitude and higher frequency oscillation on the longer BLs. It does however seem to show up more prominently on the YY product than the XX again though, so not sure what to make of it.