[mmaimcal] Brown message on small antenna array

[Steven T. Myers]

Some additional comments...

On Mon, 26 Jun 2000, Al Wootten wrote:

> From: Robert Brown <rbrown at NRAO.EDU>
> To: guillote <guillote at iram.fr>
> CC: awootten at NRAO.EDU
> Subject: Array of 8m antennas
> Date: Mon, 26 Jun 2000 10:20:49 -0400
> 
> Stephane:
> 
> In preparation for the Japanese visit Wednesday/Thursday I've been
> looking at the question of the array of 8m antennas.  I've been calling
> this the "ALMA Complement Array" (ACA) just to have a name.  Below is
> the logic I've been using; I would appreciate your views on any of this.
> 
> 1.  For operational simplicity the same receiver (front-end?) package
> that is used on the ALMA 12m antennas should be used on the ACA
> antennas.  Thus, the door to the receiver cabin on the ACA antennas is
> 1.1m wide.  I believe this consideration alone drives us to rather
> larger antenna diameter, the receiver cabin has to be big.

The main gain in operational efficiency will be to use the "standard" ALMA
reciever cartridges, which last I knew were going to be 100mm in diameter
for the 3 inner cartridges and 170mm for the 7 outer ones.  My guess is we
could make a smaller dewar based on these cartridges to accomdate
something like 4 small and 2 large (the extra small one is for a THz
receiver).  I dont think we need to accomodate the full ALMA suite all
at once and it might be easier to fit a smaller (600mm?) dewar on the
smaller antennas.  

If you want to keep the complete ALMA dewar+cartridge design, what about
going to a non-cassegrain design, with tertiary optics like OVRO?  A
larger receiver suite could be housed in a sidecab. 

> 2..  The primary purpose of the ACA is to measure accurately the
> uv-spacing corresponding to the range from 12m to about 1.5*12m.  (There
> are many secondary purposes).  Therefore, if the ACA uses 8m antennas
> with 1.5D configuration spacings we can accomplish the primary
> objective.  (if 1.5D could be made 1.4D this is even better).

I dont think 8m will be enough of a difference from the 12m main antennas
to be worth it - remember we justified the compromise of a larger antenna
by noting that 10.5m (and eventually 12m) were not so much larger than our
8m first choise that it was worth getting the increased collecting area.
My guess is that if we are going to the trouble of building a compact
array, we should choose something around a factor of two smaller, like
6m or 6.5m, with something like 12-20 elements.  BIMA has shown what
can be done with these, though it would mean that we would be duplicating
alot of what BIMA has but in the South. 

Note that for ALMA the shortest projected spacings will be 12m, and from
my mosaicing calculations for CBI we can reasonably expect to be able to
get good information on 8m spacings with 12m ALMA dishes from mosaicing
alone (though 6m will be hard).  Again, I dont think 8m will be worth
the expense - we should think seriously about 6m.

> 3.  Using ring arrays, the simpliest approach to maximize the number of
> 1.5D spacings is to surround the central element with an antenna ring
> with ring radius 1.5D.  Four antennas fit on that ring, so the total
> array is 5 antennas.  This is too sparse, not enough 1.5D
> interferometers, unusable if even one antenna fails.  We need to add
> another ring, this time at radius 3D; 6 antennas fit on that ring.  So
> now we have 11 antennas in the ACA.  This is pretty good--it gives many
> interferometers at 1.5D.  But it can be improved by "squashing" the
> array either E-W or N-S and adding another 2 antennas.  Then it is
> really good--you can get 18 interferometers measuring on 12-14 m
> spacings, and good coverge out to ~30m which will give plenty of overlap
> with ALMA so that ALMA and ACA can be easily put on the same calibration
> scale.

The best way to overlap is the cross-correlation ACA with ALMA (see
below).  I dont think we need to speculate on configurations at this
point, though it will need to be factored in to more detailed concept
designs and sensitivity calculations.  My guess, given the science niche
of the ACA (extended LSB thermal imaging in lines and continuum as
supplement to ALMA) a pretty much close-packed array will be necessary
(as we found with CBI).  I think it would be a mistake to design this
as a standalone array (arent we then just duplicating BIMA and SMA?).
Of course, I have to wonder what we would be gaining over a 12-15m single
dish optimized for submm and THz in any event if we go standalone also...
and one would then have to visit the question of a possible 24m single
dish.  We should stick to supplememnting the ALMA with a number of smaller
antennas to be cross-correlated in my opinion.

> 4.  I conclude that the ACA should be 11-13 antennas.  An antenna
> diameter of 8m seems best both from an imaging and sensitivity
> perspective.

I disagree (see above).  I would say something like 12-20 x 6-6.5m will
be a better complement to ALMA.  You advance no sensitivity calcualtions,
remember it will be surface brightness (temperature) that will matter,
not point-source sensitivity.  More needs to be done on this to decide
the question - I only note that we already know that 11-ish x 6m will
have good sensitivity especially at a superb site - ask the BIMA gang!

> 5.  ALMA and ACA do not have to be co-located because we can carry the
> calibration of one to the other.  There is no need to cross-correlate
> ALMA and ACA.  Therefore we have the possibility to locate the ACA at a
> higher elevation and take advantage of the 200micron window if that is
> seen as desirable.

I think it would be a terrible mistake to push the ACA to a higher site
without alot of careful consideration - it will be alot harder to maintain
on one of the peaks (especially if not one of the adjacent peaks), it
will preclude correlation with ALMA, and I dont think it will gain as
much as has been claimed.  For example, I doubt that there are many
really good nights where the inversion layer has not dropped below
the ALMA main site but will have above a slightly higher site.  I
saw only about one case where a slightly higher site would have been
clear but CBI was socked in.  I think that the info from the rather spotty
radiosonde campaigns should be viewed rather more critically than has
been in this case.

Also, I think running the ACA with ALMA will help cut the cost of having
a completely separate correlator system, and being able to
cross-correlate will help with sensitivity problems (eg. it will be much
easier to calibrate on a phase-reference source if there are baselines
to nearby larger ALMA antennas).  Again some studies need to be made
before pushing this hard.

Just my 10 Chilean pesos worth...

     -Steve