[evla-sw-discuss] net sideband and VCI documents

[Ken Sowinski]

Currently net sideband is conveyed from the executor in three
ways, 1) in the delay packets sent to the station boards,
2) in the multicast 'observe' xml document and 3) in the VCI
document which is sent in full to configuration mapper and
in part to mcaf.

Net sideband is needed in the station boards for the phase
calculations and to set the sideband flippers, and in mcaf
to assign 'sky' frequency to spectral windows.  Net sideband
is not used by configuration mapper, it is passed to the station
bpards where it is ignored in favor of the information in the
delay packet.  The CBE need not know about net sideband because
we guarantee, by use of the sideband flippers, that all correlated
data is net upper sideband.

model2script now generates scripts which use different correlator
setups for two scans which differ only in net sideband.  We require
this because mcaf examines a truncated VCI document to learn net
sideband, but a side effect is the long time it takes to needelssly
confiugure the correlator.  Recently this has restricted the choice
of programs that are in the observable pool.

I suggest that mcaf be changed to use net sideband as transmitted
in the 'observe' document, which it already must examined to learn
the signed sum of the LOs.  If this is done, then successive scans
which use the correlator identically, but change net sideband will
not result in reconfiguring the correlator.

model2script will have to change to take advantage of this.  It
must recognize that net sideband change does not require a new VCI
document; and must be sure to not needlessly change the use of
Widar subbands when changing band.