[mmaimcal]Memo 428

[John Conway]

On Mon, 24 Jun 2002, Mark Holdaway wrote:

> 
> Configurants,
> 
> I think Mel is correct to ignore the partially shadowed baselines.
> The computational problem of mosaicing with such voltage patterns
> is basically insurmountable for all but the most pressing scientific
> problems (ie, cannot be used for every day observations) in the
> near term computing environment.  Such software is possible in
> principle, but we cannot count on this working for routine observations
> for the first 10 years of ALMA operations.
> 

Good. Thats what I've been assumed after Stephane put me 
right at ALMA week - its a lot simpler NOT to consider partially
shadowed antennas.


> Concerning 2 vs 3 configurations, I think there is a place for 
> config 3 in the +30 to +60 degree declination range where significant
> improvement in the beam shape occurs for config 3.  The main question
> is now a cost-benefit analysis: does the benefit of config 3 in this
> dec range justify the cost of the additional pads and the lost observing
> time in the move?  We have a methodolgy for answering that question.
> 

Mel's analysis was applied to a crude concept for the compact array 
which I had at the end of ALMA week in early May. I hadn't 
really thought much about the best way of making the N-S 
stretch arrays at that point. Esentially all I did was 
elongate the inner compact array by a factor or 2 and then 
found coincidences between these new target pad positions and those
in the compact array or spiral portion. Since the density of 
pads is low outside the inner compact array a signifant 
number of new pads were needed to be built. The design also took 
no real account of minimising shadowing.

Going along with Mels suggestion of moving the shadowed antennas
out of the compact part of the array first - I have tried to find 
optimised non-shadowed pad occupation schemes of the most compact 
array and then with the antennas I have to move out  I  fill  
an extended N-S ellipse of the spiral portion, then only a few
extra pads need be built. I believe its possible to get a
pretty good design with virtually no shadowing up 
to dec +35 and a round beam, by only building of order  6 extra 
pads. Hopefully I can show something tomorrow. 

The question is whether this array has good enough performance to cover 
the dec range  +35 to +60 when shadowing starts to kick in. I believe 
that it probably will be OK; and its not worth building yet 
more extara pads or doing many extra moves to accomodate a 
special array for these extreme declinations. Mel can run this 
new design through his MIRIAD scripts to calculate sensitivities
and beam sizes for the new array and we can decide ( I have most 
of these parameters in my own software but the MIRIAD route 
is probably more bug free).

My strategy is to decide the overall topology and pad-sharing 
philosphy and then do optimisation with Boone-uv and Kogan-beam 
optimisation - the software infrastructure for this is virtually 
all there now.

> The problem of the number of compact configurations is significantly
> simpler than it used to be: we no longer need to worry so hard about
> getting lots of short projected spacings at all declinations with the
> addition of the ACA (which, by the way, is costing a bit more than those
> 25 extra pads would have cost).
> 

This is a major issue though. Are we really assuming 100% certainty 
that ACA  will be built given its not in the bilateral baseline plan? I
think that we have to ensure that the array of 12m dishes has reasonable 
very short baseline coverage. Mark is making the argument I guess 
that with ACA we don't need to have many configurations for the 12m
antennas so that for ALL  declinations we can observe such that in
projection the antennas are just going into shadowing (so we are sensitive
to 0-3m baselines),  but surely in any case projection helps you mainly in
the filling in the  short baselines the v direction; in test uv
coverages I have done as expected the shortest centre-to-centre baseline
along the u-axis  stays around the antenna collision seperation distance 
of 15m. In conclusion using projection to fill in short spacings is 
only a partial solution at best and really to do a good job we 
need the ACA (..which is why this is top of our wish list)

  John

P.S The 15m minimum distance is the spec for antennas not to 
interact mechanically  if pointing in arbitary directions. 
If we really were not going to have the ACA then one might 
I guess consider relaxing this spec and have a few be closer
to each other and somehow ensure that are slaved to point in 
the same direction (but if one antenna fails all then cannot 
move). But with 4 million dollar antennas we don't want any 
risks I guess.