Debra Shepherd quoted Jim Pisano on the correlator setup time: "The correlator software will be able to support this switching between 2 configurations on 48 ms boundaries with one or more (the actual quantity is TBD) dumps discarded on either side of the configuration switch. Please note that correlator setups must be available 1.5 seconds before they are to be used. This allows for time to propagate the configuration to the correlator hardware and correlator computer software. This is a specification I got from Al Wooten which I've set as the baseline rule. A complication to this configuration switching scheme is in the use of subarrays. It is much more complicated to have 2 subarrays (or more) switching between 2 configurations. If this functionality is required, then we (correlator software & hardware groups) must be explicitly told that this is a requirement. We currently believe that having 1 subarray switching configurations while a second subarray uses a single configuration is doable without extra effort." Robert Lucas wrote: "1) This might be a real problem as the correlator will be a large fraction of the time split into typically 2 subarrays (one observing an astronomical project, the other calibrating some antennas e.g. after a move or maintenance). It looks like that a change of spectral configuration is an operation affecting the whole correlator, not a subarray ? As long as configuration switching is defined as available at the project level, it is a rather difficult scheduling issue to make sure that it is limited to one subarray at a time!! 2) One thing I wonder is whether the type of project we are thinking of (spectral suyrveys) really require re-configuring the correlator, or simply resetting the first and/or second LOs -- which is quite a different thing. For these projects one want to define a spectral resolution and bandwidths and use it for stepping through frequency space." Mark Holdaway wrote: "The typical case of using a second subarray for calibrating newly moved antennas may be fine, as we switch among correlator setups in the primary subarray doing astronomical observations and the second subarray will mainly be using a broadband continuum setup. However, one of the potential uses of subarrays would be to do two scientific projects at the same time, say a generic sort of observation on one and a monitoring project on the other. Generally speaking, both subarrays would need to go back and forth between some specific spectral line setup and a broad-band continuum setup at ~90 GHz for calibrating. If fast switching is used, the two correlator setups we want to use for a given subarray will have different frequencies, bandwidths, channel widths, and integration times." And Bryan Butler noted: "when calibrating a newly moved antenna, you won't always be in broadband continuum mode - you also need to find delays (so-called "cable lengths"), which is done in spectral line mode. to get the pointing model and antenna location (so-called "baselines"), you *will* be in continuum mode... it is my feeling that it should be a requirement that the two subarrays be able to switch between independent correlator setups." and Al said: "I agree that the restriction that two or more subarrays cannot switch between two independent configurations seriously restricts array use. Remember, as Robert noted, that from February 2007, when the Early Science array begins commissioning, there will be two or more subarrays on ALMA continuously until 31 December 2011. Thereafter, most of the time there may be at least two subarrays. There are a myriad of correlator configurations available, several of which are needed in normal calibration sequences. The subarrays need to be able to switch between independent correlator configurations. Several examples of this need have been given; it would be useful to turn them into more specific examples, or use cases."