[mmaimcal] ACA/nutator discussion summary

[Al Wootten]

The entry of Japan as a partner in the ALMA Project enables some
enhancements over the originally proposed and costed instrument.
Two particularly appealing enhancements are the construction of receivers
to cover the remainder of the ALMA bands, and the construction of
a supplementary array of smaller antennas, the Atacama Compact Array (ACA).

Several memos and many discussions have investigated the form which the
ACA might take.  Scientific discussions have concentrated on definition
of the size and number of the atennas, including a summary of discussions
by Wootten, followed by documents by Guilloteau and Welch and imaging 
simulations by Morita and Yun.  Recently, Jaap Baars,
in his draft memo of 21 September 2000, summarized, noting that
the Atacama Compact Array (ACA) serves two primary purposes.

1) The ACA will measure missing spacings in the uv plane, since the 
specification on proximity of the antennas does not allow them to be closer 
than 15m, yet the 12m antenna size limits measurement of spacings, particularly
in the range 4-12m or so (see e.g. Guilloteau's Fig 1 in his 11 July memo).
The antennas of the ACA should be of a size to optimally measure spacings
in this range and of a number to provide optimal imaging when combined with
ALMA.  The ACA will be parrticularly important in measuring missing spacings
at the highest frequency range of ALMA, where its specifications limit 
imaging performance the most.  The ACA should be designed for excellent
performance in the upper frequency bands of ALMA.  From a purely scientific
point of view, a 6m antenna optimally measures the missing spacings, as
discussed by Welch and Guilloteau in their memos.  An 8m antenna was originally
planned for the MMA, so some studies antennas of this diameter are available;
closest spacings were 10m.

2) The ACA may prove useful as a dedicated array for observations in the
supra THz windows where the site provides transmission approaching 30%
 (Paine et al. 2000 PASP 112, 108) about 15% of the time.  The antennas
of the ACA should be of sufficient surface accuracy to be useful at 1.5 THz
and of sufficient sensitivity to provide suitable calibration at this frequency.

Jaap notes that the standard ALMA frontend unit will provide all bands on
the ACA and in addition will allow the installation of supra-THz inserts
in the slots of some bands, such as the lower frequency bands, for which
short spacings may not frequently be needed.  If this occurs, some of the time
the ACA will have different receivers from ALMA, and will be incapable of
working with it.

If the ACA fills one of both of these conditions, then:
At least some of the time, and perhaps a significant fraction of the time
the ACA will operate in an independent manner.

Jaap posed the question:
 Do any, or all, of the ACA antennas have to be equipped with a 
  nutating subreflector?
 Or can we do without them, and make life a lot easier and cheaper?

Holdaway and others of us in the imaging and calibration group have 
discussed this, and I summarize:
If 1) above is the exclusive use for the ACA, then one may be
able to use the ALMA/ACA combination relying on total power measurements from
the 12m antennas alone.  However, at the upper frequency ranges one would
expect errors to affect the 12m total power measurements to be at their
most severe.  The ACA can complement these measurements using total power
measurements made on elements of the ACA either through fast motion, or
through a nutating secondary.  Since the primary elements are smaller, 
the ACA should be more effective at this than ALMA's primary antennas are.
>From a pointing and surface error point of view, the ACA data on short
spacings might be superior to the ALMA 12m data; however this might be offset
by calibration errors, as calibration is more difficult with the smaller
antennas.  Simulations of an ACA working with ALMA have shown that pointing
performance of the 12m antennas can limit achievable image dynamic range
though fidelity index is improved by the addition of the ACA.  Would inclusion
of an independent set of data from ACA antennas equipped with nutators
change this conclusion?  It seems to me it might improve fidelity index,
but probably wouldn't alter the conclusion on dynamic range.

If the ACA performs 2) above, a scientifically productive option, then it 
would be useful to outfit some of its elements with a nutator.  As I 
mentioned at the ASAC meeting, a model in which Band 10 was first implemented
on the ACA while the technology matured allows us to reap science in that
band while postponing investment in 64 receivers until the technology improves.
In this case, the ACA functions as an independent array also.

Total power measurements are most likely to be affected by variations
in differential spillover.  In rapid antenna movement, such as in creating an
on the fly image, the varying ground will be compounded by antenna errors
in coupling to it at the higher frequencies.  Motion must be fast enough
to cancel atmospheric errors.  The nutator, particularly a good performance
nutator such as might be built for a smaller antenna, should provide good
atmospheric cancellation and moderate throws.  At submillimeter wavelengths,
total power images from bolometer arrays show that the sky has a complex
extended structure.  Imaging could be compromised by insufficient beam throw.
For the smaller sources most appropriate for interferometry, at the higher
frequencies, a nutator should provide better imaging performance.

Another question involves practicalities of construction and cost.
Simon Radford has done a preliminary investigation of this and reports:

'Since the ACA antennae will be smaller than the ALMA antennae (6-8 m vs.
12 m), presumably the subreflector will also be smaller (350 - 500 mm
vs. 750 mm). For equal specifications (beam throw, transition time,
etc.), the nutator force requirements (and size) scale with the mirror's
moment of inertia, which scales at r^4 (or even r^5 when the thickness
is included). As an illustration, the SMA and ALMA nutators have similar
specifications. The SMA nutator, for a 350 mm mirror, uses a pair of 4
lbf (17.5 N) motors. The ALMA nutator, for a 750 mm mirror, will use a
pair of 100 lbf (444 N) motors. This is a scaling of r^4.25.

Hence ACA nutators could be much less powerful (and smaller) than the
ALMA nutators (3-20%). Equivalently, ACA nutators could have higher
performance goals than the ALMA nutators.

The cost of prototype nutators, however, probably would not scale nearly
as steeply as the motor and physical size. With the quantities planned
for ALMA (4 or 5 nutators), we probably will not see much economy of
scale in production.'

Advantages of nutators on the ACA for obtaining total power are
1) minimized ground pickup differences between 'on' and 'off'
2) good cancellation of atmosphere, especially where it is strong and
    variable at submm wavelengths
3) somewhat higher performance nutators could be provided for smaller antennas
4) operation in the supra THz band would be enhanced
   better calibration (a corrolary of the above)

I think that from these considerations, the project should consider 
nutator development for at least some of the ACA antennas.  Since their number
size and configuration is only loosely defined, assume that an ACA
observation has minimal redundancy in its shortest spacing.  Then an
ACA single element observation should have an integration time similar to
that on any of the array baselines.  As with the large array, the
factor of two difference in the switched mode suggests that four antennas
with nutators should provide the necessary total power information.