Dear Colleagues,

    I disagree with Al's statement on the need of wobbling subreflector for
ACA.  Let us consider several cases

1) ACA is only used at the ALMA frequencies.
    - then the total power is provided by the ALMA 12-m antennas, either in
total power (e.g. OTF spectral line)
      or using the few antennas equipped with a subreflector (4 is enough).
    - The ACA antenna would NOT need wobbling secondaries

I agree at the lower frequencies.  However, under Jaap's guidelines the
smaller antennas will have improved surfaces, and probably better pointing
owing to, for example, lower wind buffeting on overbuilt mounts, and are
likely to provide higher quality sampling of total power, especially when
equipped with subreflectors.  The ACA antennas would not need nutators,
but should have them for best performance of ALMA.

2) ACA covers all ALMA frequencies, but ALMA not (for cost reasons,
schedule, etc...)
    - the good way to solve that problem is to equip 1 or all of the 4 ALMA
12-m antennas which have the wobbler
      with the same receivers as ACA.

Again, this may work but the total power data may not be of the same calibre
as would be obtained by an ACA antenna alone.

3) ACA goes above one TeraHertz, where ALMA does no longer operate
    - given the small number of ACA antennas (< 16), a SINGLE  antenna
equipped with a wobbler is sufficient
      (4 makes it for 64...).
    - that can either be an ACA antenna, or the best 12-m antenna...
      A larger antenna offers some advantage, since it provides better UV
coverage overlap with the ACA array.
      The phase errors introduced by pointing errors at the overlap point in
the UV plane are actually 25 % smaller
      for a 12-m/8-m combination than for m 8-m-8m overlap when the pointing
error is proportional to the beamwidth,
      and 20 % larger if the pointing error is constant.
      On the other hand, 12-m antennas will have somewhat poorer surface
accuracy (to be quantified, using only
      night time conditions).
    - Arguments of high dynamic range and/or high fidelity above 1 THz seems
quite weak to me, and would need
      to be quantified. The fidelity estimate should take into account the
considerable difficulty of amplitude
      calibration at these wavelengths, which may be the dominant problem.

We will work on quantifying the performance of the array at high frequencies
as well as amplitude calibration, a comsiderable problem.  But I agree that
the minimum acceptable solution is to have one ACA antenna with a nutator.

4) In comparing wobblers for ACA and ALMA antennas, remember that the
wobbling frequency should be inversely proportional to the antenna diameter.

Yes, I agree with this.

As a conclusion, I believe that
    - it would unwise to let the close packing characteristics (which are
the prime concern for
    ACA antennas) be driven by the wobbler problem.
    - it may be wise to have one ACA antenna with a special subreflector
package, of different size, with a wobbler
    in it.
    - the real good solution would be to get ONE good 12-m antenna equipped
with a wobbler and the same
    receiver package as the ACA antennas.

Whether the close packing limit is actually affected by a wobbler or not is
an engineering problem.

I agree.  I think that the number of nutators should be driven by the close
packing characteristics, since it is basically the sampling of the short
spacings by the interferometer which must be matched by the single antenna.  

I agree that the wisest step would be to include a nutating subreflector
on at least one ACA antenna.  It would be even wiser to have more than one
but the exact number depends upon the final number and configuration of the
ACA antennas as well as the mode of their use.

I agree that in the best of all possible worlds, we would have a 12m
antenna with a ten micron surface to provide total power, with a nutator and
with the ACA receiver package.