The topic of controlling scan timing is discussed in detail in the Scan Times section.
DWELL is an alternate way to specify the duration of a scan. It is distinguished from DURation only in that the start time of the scan will be delayed until SCHED expects all antennas to be on source. Both the slew time, including acceleration, and the settling time from TSETTLE in the station catalog will be taken into account. The interval between scans will not be allowed to drop below MINSETUP from the antenna catalog to allow for finite scan setup times at some antennas. SCHED tries to arrange that useful data will be obtained for the full time specified by DWELL. With DURation, some of the specified time may not have good data because antennas are still slewing. Please see the description of DURation for a more complete discussion of the actions of these two parameters and their interactions with other parameters that influence scan times and recording activity.
As of Oct. 2010, DWELL has acquired second and third arguments. Argument 2 is the number of antennas to not wait for. Typically it will be a small integer like 1 or 2, but can be up to the number of antennas (not sure what that would do!). This should be useful if the there is a problem with long slews between pairs of sources near the zenith at one antenna. It would also be useful if you wish to let the less sensitive, but faster, antennas start observing once they are on source on the assumption that large, slow antennas might not need as much integration time. This is an issue with global observations or HSA. SUMITEM = EARLY can be used to inspect the results of the use of the second DWELL argument. For example, with that argument set to one, one antenna should have EARLY negative (got there after the start).
The third argument is a minimum time on source for the antennas that are not being waited for thanks to argument 2. The default zero for this argument allows the scan to end before the slow antenna(s) get to source. If a time is specified, then the scan will be extended until the last antenna gets that much time. In such a circumstances, the scan will still start when first group of antennas (other than those the second argument says not to wait for) get to source, so those antennas will get a longer scan than specified.
Please note that the model SCHED uses to calculate slew times is not especially sophisticated, especially in regard to the excess times beyond the slew that an antenna might take to start getting good data. It is adequate for most purposes, but, for example, it might not agree exactly with other programs such as the VLA scheduling program Observe. In such cases, the station specific program probably has the better answer. Also, for some systems, such as the VLA, SCHED is more interested in when VLBI data starts to be good which might be before a local correlator starts to get good data. As of Feb 2003, SCHED does take into account the time an antenna takes to accelerate to full slew speed, and should calculate slew times for short slews approximately correctly even if full speed is not reached.