ca-forum #18 minutes

date: Tuesday, Sep. 10, 1996. 11:00 - 12:30


Tied array (CACAL) - (M.W.)

The continuous phase calibration works OK. Further experience is required to set the update frequency.At present the machinery works in a box-car integration over N cycles. N was initially set to 8 cycles,but this is too long. MW now recommends 3 cycles - the problem is simply the rate at which the atmosphere varies, and the difficulty in keeping up with the variations.

Additional problem raised by J. Reynolds: should the phase calibration reference all the changes to the reference antenna, or would it be preferable to use the mean phase. The mean phase path makes some effort to keep the Narrabri signal independent of the atmospheric variations; using the reference antenna ensures that the atmosphere above the reference antenna is fossilised in the record, and will need to be removed during the processing. MW will look to using the mean phase as reference.

CASNAP issues (DJM)

This utility was revived at the specific request of the observers. The code is in bad shape, and is not reliable, as much of the supporting software has been modified (CAGET, in particular). There are some experiments where CASNAP is of great value, but the effort to make the product reliable is non-trivial. Derek McKay has summarised the problem in appendix A. User feedback is needed so that the Narrabri group can estimate the priority to assign to CASNAP.

>> potential users of CASNAP should get in touch with M.Wieringa.

X-band Beam-shapes (RS)

Bob Sault recently made an eight-hour observation to assess the X-band beam shape. The results are intriguing: there is a error of 0.23 arcmin in elevation, common to all antennas, with a small spread between antennas and hour angle. It is known that very low pointing rms can be found while tracking a single source, but a systematic 15 arcsec offset common to all antennas is not understood. The results presented were based on the stokes I data which will disguise the know XX and YY differences. Further processing is underway.

Software phase-switching (RS)

Bob ran a series of tests of the software phase switching. The procedure is a "poor man's" phase switch, as the cycle frequency is set by the basic integration cycle (10 seconds). The results show that the procedure introduces no undesirable artefacts. Unhappily, the non-phase switched data shows no problems, so we are unable at this stage to say whether the switching confers any benefits. The tests may have been at some disadvantage as the data were taken in the most compact configuration (122m); Bob will look to repeating the B. Gaensler experiment which did show problems (significant flux at the phase centre).

Reference Pointing - update (DJM)

The machinery is still in its shakedown phase, and users are encouraged to use it.

John Reynolds' recent trial revealed a problem: the program will not use a correction if it looks to be unreasonable. "Unreasonable", in its definition, is above some threshold when the previous reference pointing is within a specified time window. In John's experiment, covering a large part of the sky, this criterion was too severe, and genuine errors were discarded.

The decision algorithm needs to be reviewed - the signal-to-noise measure should perhaps be included; in other words, if the error is large, but the data are entirely consistent, and the signal/noise is satisfactory, then the error should be used, or some user feedback provided. Derek welcomes suggestions.

Focus Control - update (DJM)

The hardware is complete in antenna 3. Ron Beresford is steadily progressing through the remaining antennas. The software is now needing attention (ARRAY will need attention) - from Derek McKay who is otherwise tied up.

Pointing Discontinuities in the ACC - (GB)

This problem had been quiescent for a month or so, but has now returned at a greater degree of severity than before, and is now engaging DJM's efforts. A symptom not previously reported is that the problem is associated with a rapid increase in the reported number of encoder errors.

Planetary Data in the RPFITS files (RS)

Bob Sault's tests confirm that this problem ahs been fixed.

Frequency Switching and the SAC Machinery

The question was raised at a previous CA-forum as to the priority to assign to the revision of the attenuator machinery in mosaic mode. At the time the issue related to the number experiments such as the CEN A observations where serious differences in Tsys could be encountered from one field to the next. It was argued that the priority was perhaps low, as there were few such experiments (CEN A and SAG A). However, frequency switching experiments may also encounter the same problem (rapid and serious jumps in Tsys), which means that the SAC question should be revisited.

problems pending :
  • Faraday correction
  • DUTC ... (wew/mw)
  • Pulsar ephemeris data -> RPFITS (wew)

Appendix A - CASNAP problems.

CASNAP status

CASNAP is program which takes the visibilities from the ATCA and processes them on-line to present a real-time cleaned cut of the observered source, the point spread function, the dirty beam, amplitude and phase data as a function of baseline and fitted models of the source components.

It was written in 1992 by David McConnell for use with the PMN follow-up project. Since then it has had a chequered history of maintenance and upgrading, by both DMcC and DJM.

The original used to run off data provided by a program called SNAP. SNAP is now decomissioned and will not be recommissioned. CASNAP has been converted so that it works, albeit unstably, under SNAP's replacement - CAGET. CAGET is also used by other on-line programs such as CABSLN and VIS

The present version works, but poorly. Many of the features have been disabled.

It crashes regularly - usually to do with poor handling of the changed history mechanism between SNAP and CAGET, and to different handling of the on-line flagging. There is no model fitting, variable integration, or data logging (to name but a few) with the current version.

Based on the effort to convert CASNAP to run off the CAGET data, I would estimate 30 person days required to rewrite the program. If the program is not rewritten properly, it should be decommissioned. Perhaps expressions of interest from the user community would be useful in determining its value.

Next Meeting : Tuesday, October 8 at 11:00

Staff space
Public