Recorded to ATNF V006B
Near the start CACAL was run continously: 3 cycles, phase only, 90^o included (i.e. circular),
This holds it to better than 20^o of phase. Number 5 is the only one drifting
At 01:59 CACAL was restarted with 6 cycles averaging and pol leakages >included. The leakages were all zero
02:04 Source change gives very different phases. Takes a minute for them to be solved for (of course). And one or two cycles to apply.
add ante 5 at 2:16
4:36 Comment on the phases when tied. When we change sources the phases are very different. For ca 5 or 4 in particular. CACAL brings them together. After this they wander but in harmony.
4:48 When we return to 1934 there is no repeat of the leakage calculation, but does a very bad job of carrying the phases forward from the previous scan. Basically is averages the data from before with that of this. This is fine when we are phase referencing – good in fact – but will add a few extra bad data points at the start of scans with large slews. So there are improvements one could make to CACAL. Make it so phases are not carried over long slews.
5:40 interference on this source for the shortest baseline
In summary - I reckon we have shown that we can use CACAL to tie the array. No surprise there. It has shown that we can keep all 5 antennae in line to better than 20 degrees most of the time and better than 60 all of the time (in this run). We saw that the phases would be wildly different changing from point to point across the sky, so I have no doubt that every major slew will require a recal of the array. In VT11L we showed that there was no detectable effect of running CACAL on the antennae phases. We also saw that the valid solutions are carried over from scan to scan, which is good for phase referencing, but bad for wide slews across the sky. This could be changed.
Comments on the processing:
Running CACAL all the time can produce this error
Long slews can (must) introduce different phases for the different antennae. (Would you calibrate from 1934 to observe 0823?)
The phases are corrected with in the minute. Sometimes in two steps as residual solutions from the previous observations are used (even though it is a long way away on the sky). Comparing this data to that of V255D/E (which did not use continuous cacal) one sees that in this data set the sources are noiser, but all the antennae would have phased up fine. Phases differ by less than 20^o. This is presuambly because CACAL was run on a nearby source (1934 and 1757?). There is no ca06 in this data, so they are all reasonable close. The weather was good. This will not always be the case.