[mmaimcal] question about frequency/band switching speeds and rates

[Bryan Butler]

all,

i'm writing some specs for software (for the SSR requirements on the 
control subsystem) on a bunch of hardware rates, and thought i'd get 
some input from you - what is the current thinking and hardware limit 
for these things (i'm out of the loop on alot of these discussions now!).

each of these is specified as a _rate_, which is how frequently such a 
reconfiguration must be able to be made, and a _speed_, which is how 
long it takes to complete such a reconfiguration.  my comments are in 
brackets below each spec.

. frequency switching within a band: rate=100 msec if within .03% in
   frequency, 5 seconds otherwise; speed=10 msec if within .03% in
   frequency, 1.5 seconds otherwise.
   [the speed comes from LO specs; rate is a guess]
. frequency switching to a warm band: rate=5 sec; speed=1.5 sec.
   [the speed comes from LO specs; rate is a guess]
. frequency switching to a cold band: rate=20 min; speed=10 min.
   [these are guesses - i can't remember what the spec is here]

. correlator reconfiguration: rate=1 sec; speed=48 msec if the
   configuration is preloaded, 2 sec if not.
   [this comes from the correlator specs, but i'm not sure on the
   2 sec - the last email i have from steve scott and jim pisano
   on this is that it was "several" seconds]  {as an aside, there
   also needs to be a spec on whether multiple subarrays need to
   have multiple configurations preloaded}

. OTF mosaic cell phase-center saltation: rate=100 usec; speed=10
   usec.
   [for the rate, i've assumed slewing at 1 deg/sec, and 950 GHz,
   with 2 samples per FWHM; speed is a guess - it will depend on
   how fast the correlator can reset delays]
. phase center switching in the primary beam: rate=5 sec; speed=10
   usec
   [these are guesses]

. nodding source switching: rate=5 sec; speed=1.5 sec
   [speed is antenna settling time; rate is a guess]

. subarray redefinition: rate=3 sec; speed=1.5 sec
   [complete guesses.  speed will depend on many things, but is
   probably limited by how fast the LO can be redefined.  if LO
   doesn't need that, can be quicker, but is still limited
   potentially by a bunch of other things like correlator setup,
   etc...]

also, can somebody point me to the most recent data rates document?

	-bryan