[mmaimcal] Re: [alma-sw-ssr] question about frequency/band switching speeds and rates

[Mark Holdaway]

Ed,

Are you referring to Fast Switching here?

A typical optimized fast switching cycle will look like this:


0.5 s observe calibrator

1.5 s move to target source & settle

25 s integration on source

1.5 s move back to calibrator



It is fairly efficient time wise, and should be designed to maximize
the overall sensitivity:

   sqrt( time_on_source / total_cycle_time ) * exp( - sigma_phi^2 /2 )

The first term is the efficiency from the duty cycle, the second term
is the coherence in the presense of decorrelating phase errors.
If the cycle is too long, the first term will approach 1.0, but
at the expense of larger residual phase errors and loss of coherence.
Alternatively, if we switch too quickly, sigma_phi goes down, but the
duty cycle gets worse.

And at last, the time on the calibrator is set by the residual atmospheric
phase errors -- it makes sense to have the thermal phase noise on the
calibrators 2-3 times less than the residual atmospheric phase noise, but
it is a waste of time to get 4-10 times smaller phase errors on the gain
solution.

    -Mark


> 4.  Nodding rate: The maximum nodding rate that may be used will, in
> fact, depend on the nodding speed (including settling time).  You
> probably don't want to spend more than 1/3 of the time not observing a
> cal or target.  If the WVR system works well, nodding may be minutes
> rather than seconds may be enough
>