VLAPEAK sets the character to use in the VLA observing deck (...obs.y file) to control the use of pointing corrections determined during the observations. The actual request to do a pointing observation is made by setting VLAMODE = IR in a separate pointing scan on a calibrator. The integration time must be set to 10 seconds, which is the default that SCHED does not yet have the ability to modify. VLAPEAK is used to specify whether or not a previous pointing solution should be used as the starting point for the new measurement, whether to save a previous solution for use as a future starting point, and, for normal scans, whether to use the corrections from pointing measurements to adjust the tracking position.
A pointing pattern takes 140 seconds if the VLA correlator integration time equals the default 10 seconds, and some extra time should be added. Use 150 seconds (after arrival at source) for one pointing pattern and 4 minutes for two patterns. The on-line system will keep repeating pointing patterns until the scan is over. The results will be averaged. Two patterns are recommended. Reference pointing is important at 7mm (Q band) and is useful, but not required, at 1cm (K band). The pointing should be updated approximately every hour (some claim faster), and any time the antennas move to a significantly different part of the sky. At elevations above about 70 degrees, the azimuth changes rapidly in both time and calibrator/source position offset. Reference pointing may be marginal at such elevations, and, if attempted, should be done more often than at lower elevations. For pointing at X band, a calibrator (or target) source should be used that has a flux density in excess of about 200 mJy on all VLA baselines. See the VLA documentation for more details and advice.
When observing at Q band (7 mm), it may be desirable to first determine initial pointing corrections at a lower frequency, such as X band (4 cm), and then, using those corrections as the starting point, determine residual pointing corrections at Q band. The final pointing offset will then be the sum of the two. Normally just the lower frequency determination should be adequate -- the relative offsets between bands are rather well known. But the ultimate pointing accuracy requires the final determination at the high frequency, assuming, or course, that there is a source of adequate strength close to the target. Pointing should not be done only at 7mm. The a priori offsets can be bad enough that the pointing source will be on the edge of the beam and a valid pointing solution may not be obtained.
VLAPEAK provides the necessary options to manage the pointing corrections. To understand the VLAPEAK options, consider that there are ``primary pointing corrections'' and ``total pointing corrections''. The ``total pointing corrections'' are those that are applied when running a normal scan that uses the reference pointing results. The ``primary pointing corrections'' are those used as the starting point for a pointing determination. This description is not quite literally what the software does, but is correct in concept. The VLAPEAK options are:
Usually, when reference pointing is used, there will be a short scan (see times above) with VLAMODE=IR and VLAPEAK=' ' to determined the pointing followed by the observing scans with whatever VLAMODE is needed and using VLAPEAK=T. The pointing scan will usually be at X band (4cm). This is all that is needed in most circumstances.
When double reference pointing is desired, the pointing is first determined at a lower frequency, then tweaked at the higher observing frequency. That is when VLAPEAK=S or R are used. The sequence would be a scan with VLAMODE=IR and VLAPEAK=' ' at the lower frequency, followed by a scan with VLAMODE=IR and VLAPEAK=S at the higher frequency, followed by observing scans with VLAPEAK=T and the desired VLAMODE, followed by another tweak of the pointing using a scan with VLAMODE=IR and VLAPEAK=R, followed by more observing scans. The VLAPEAK=R - observing mode scans combination can be repeated several times. Eventually the pattern should be restarted with a new low frequency scan. Whether double reference pointing is actually better than just doing the pointing at X band and leaving it at that is somewhat controversial.
VLBI recording may be stopped or left running during pointing observations. You cannot combine pointing scans with phasing scans because they require different VLAMODEs.