Henry Spencer wrote:
In article . com,
wrote:
As somebody else mentioned, a large baseline optical interferometer
is
likely to give more info for a much smaller price.

No, it depends on what sort of information you are after. In
particular,
optical interferometers are useless for spectroscopy and are usable
for
imaging only on fairly bright objects (because you need a fair number
of
photons per second to form detectable interference fringes). For
many
astronomical purposes, there is just no substitute for lots of mirror
area.

Erm, interferometers are not really "useless" for spectroscopy.
Interferometry makes spectroscopy a whole hell of a lot more
difficult, but it's still doable. If the only way to observe
the target is through deep interferometric nulling, for
example, then you accept the difficulties and move on.
Specifically, what you want is to be able to sample the
configuration space (by rotation the interferometry assembly
along an axis which points at the target, for example) to a
degree that allows you to create a processed, properly
calibrated spectrum that you can trust.

For this specific example, the collection of the spectra relies
on the interferometry. This takes advantage of the deep
nulling ability of interferometry to allow the observation of
very dim targets close to very bright targets (e.g. planets
near stars). However, in general, spectroscopy does not rely
on the features of interferometry, so in most cases spectroscopy
and interferometry live in different worlds. For example, while
interferometry can create higher spatial resolution views, doing
so necessarily lowers the surface brightness "per pixel" of the
image, which is not what you want for spectroscopy (which
almost always prefers more signal over finer spatial coverage).
And, of course, in general the whole idea of interferometry is
to use parlay a small multiple in mirror area (e.g. 2x, 5x)
into a large multiple in resolution through separation (or a
high degree of nulling). So, in general, as has been pointed
out, most spectroscopic targets benefit more from greater
light collection area than from anything else.

Interestingly, there have been a lot of very innovative
alternatives to interferometry in regard to deep nulling and
such like which have been proposed for various projects,
such as terrestrial planet finding, recently. For example,
the leading contender for NASA's TPF mission is, I believe,
a coronograph rather than a nulling interferometer.
Reflecting, I think, the many advances in overall "old
school"ish more or less classic-design type designs which
have taken place in the last decade or so as well as advances
in instrumentation (dynamic range, etc). Compare, for
example, the sophistication and capabilities of, say,
Chandra, WMAP, and Spitzer versus say Einstein, COBE, and
ISO. Where once the TPF specs. were beyond the state of
the art for conventional telescope technologies, those
technologies are starting to catch up to the point where
just a, mostly, plain jane telescope can meet the specs.
As above, one of the great advantages of a coronograph
style design is that it would greatly simplify spectra
aquisition and improve spectra collection throughput.

An even more interesting, and certainly more innovative,
concept would be the gigantic pinhole camera*, which
would have somewhat similar problems to an interferometer
but could offer much deeper nulling and finer resolution
at significantly less cost.


(*)
http://www1.nasa.gov/vision/universe/newworlds/new_worlds_imager.html