Barry Clark  5 Dec 2000:
> -=-=-=-=-=-=-=-=-=-=
> 
> (6) Fringe Tracking Centres -- Emerson
> 
> The LO, systems & software groups have decided that for OTF interferometry,
> the fringe tracking center position will be kept constant for the duration
> of one scan. This has implications for the realtime software (a great
> simplification), for the analysis software (a similar simplification), for
> potential maximum data rates (possibly higher), for the maximum OTF field
> size that can be covered in one scan, and for how OTF interferometry
> observing is carried out.
> 
I think this is wrong.  The data rates are not "possibly higher", but one
to two orders of magnitude higher.  I think we need to implement stepping
fringe stopping to save us from data inundation in the early years before
we can handle it.
------------------------------------------------------------------------
Robert Lucas  6 Dec
One month ago I had sent to this list my own reaction to memo 331 (see
below)

Best regards

Robert

------------------------------------------------------------------------
    Subject: 
            Re: [alma-sw-ssr] FW: [almanews] ALMA Memo 331 Released
       Date: 
            Mon, 06 Nov 2000 09:55:54 +0100
       From: 
            Robert Lucas <lucas@iraux2.iram.fr>
 Organization: 
            IRAM
         To: 
            alma-sw-ssr@nrao.edu, "Emerson, Darrel" <demerson@nrao.edu>,
ldaddario@inrao.edu
  References: 
            1 , 2

One remark on memo 331:

Larry and Darrel (following up a discussion that we had in Tucson last
  month), conclude rightfully that is is simpler to track a fixed
  point in the sky during an OTF mosaic scan rather than a fixed point
  during every integration.

However Larry and Darrel do not consider in detail option b, but they
  say "Continuous tracking of the beam center is possible"

I think it is not possible at all, due to lack of coherence.

Phase tracking the center of the main beam amounts to have the array
  aperture plane rotating at angular rate alpha^dot equal to the
  angular scanning speed.

At the end of a baseline of length D this plane moves by a length of
  alpha^dot*D*tau in one integration time, relative to the incoming
  plane wave from a celestial source; this length must be smaller than
  lambda/(2*pi)*0.36 not to lose more than 2% in sensitivity.

This limits the scanning rate to vanishingly small values (10^-7
  rd/sec at 300mum, 10km baseline). Basically it says than you must
  adequately sample the data during the time it takes to the
  _synthesized beam_ to cross any celestial source (and not the primary
  beam).

So continuously phase tracking the center of the moving primary beam is
not
 feasible at all. You should do this e.g. to track a moving object
 (such as a satellite) but not a celestial source.
------------------------------------------------------------------------
Steve Scott 6 Dec
 Robert Lucas wrote:

> However Larry and Darrel do not consider in detail option b, but they
>   say "Continuous tracking of the beam center is possible"
> 
>...
>
>   Basically it says than you must
>   adequately sample the data during the time it takes to the
>   _synthesized beam_ to cross any celestial source (and not the primary
>   beam).


The last statement is a good simple way to think of the necessary sampling 
rate with continuous phase tracking. However, I don't think this is the issue 
that Barry was referring to in his last email. Hopefully he can give us some 
more details.

Steve
------------------------------------------------------------------------
Barry Clark  6 Dec:
What Larry and Daryl are saying is to hold the phase center fixed at the
center of your mosaic, and to sample data fast enough that the residual
fringe rate due to the distant phase tracking center does not cause loss 
of coherence.  They suggest this to get around the difficulty of synchronizing
a jump in phase, due to a jump to a different phase tracking center, with
the integration times to a small fraction of the integration time, as is
needed.  To a rough approximation, for an N * N mosaic, this causes a data
output rate N times higher than for a jumping phase tracking center.
I think this penalty is high enough to be worth the work of synchronizing
the jumps in LO phase.

This essentially applies only to interferometric OTF mosaicing.  Jumping
the telescope positions requires flagging while they settle down, and the
LOs easily settle down faster than the telescopes do.

------------------------------------------------------------------------