IMHO,

A.O. Has never been closer to the reach of Amateur Astronomy.
As the industrial push for A.O. continues, the possibility of lower cost
components has already become evident with sites like Ebay where
I have seen tip/tilt and DMs for sale.

For Astronomy, Location, location, location, location.
You use AO with your best telescope at the best site.
Best results in a photon limited system comes from systems that need
the least static corrections (optics) and have the lowest optically
significant turbulence. AO can't make a bad site good.

We did some seeing measurements two months ago at the VATT and had
seeing from 0.5 to 1.5 arc seconds most of the time. 80 or 90 percent of
the seeing was near the telescope. (under 1 Km) We are not ready to say
how much was at the telescope/dome level but I suspect from a few
percent to all at times.

This means:
Speeds are not as high as the upper layers
Iso patch size is larger than in an upper layer dominated case

D.O.D. ie star wars did R and D using AO for various applications and
produced high order non photon limited systems with as high a bandwidth
as they could because the were tracking bats out of hell.

Mt Wilson has such a mirror, 256 elements ? and on the Hooker (100")
can go down to 12 th ? depending on the color of the star.

You need a big scope to go to faint.

The Old Mt Wilson AO system used a custom TEC CCD
and a lot of DSP boards not to mention 256 Hv amplifiers.

These systems used a wave front detector, usually a Shack/Hartmann
that breaks the light in to sub pupils used to control the tip/tilt of
each region buy the displacement transducer near the sub pupils. The
Reconstructor takes the wavefront signals and produces mirror control
signals. Faster is better as you are in a control loop.

All of the above is simplified by measuring and controlling the mirror
curvature. F. Roddier is the man to read the works of to understand all
about it. The re constructor for a curvature system is a diagonal matrix
making it fast. My first system was analog the computer just
watched the loop, taking data.

Curvature mirrors are being made for optical communications and as the
volume goes up the prices go down. These are low voltage pizo bimorph
mirrors and could be made cheap if you knew how. (i don't)

That leaves the detector and that's the rub. To work well it needs to
be noiseless with high Q.E. not to mention a fast readout time.

We used APD detectors and kHZ readout rates =$$$
and a fiber fed lens array (low fat) and that is another trick.

CMOS based sensors are in a fast lane to low noise performance.
It may be possible to obtain cmos sensors with machine vision processors
integrated in the near future.

I would start by using a bimorph to correct the static mirror errors
first. You would have a knob for tip/tilt, defocus, astigmatism X any Y
etc. That would allow people like me with aging eyes to be able to enjoy
not wearing glasses at the eyepiece.

When a detector comes along then go for it. While you are waiting you
could play with non photon limited bright systems. For vibration these
systems need to be fast with a large tip/tilt range. With a laser beacon
or possibly with techniques used in solar adaptive optics one could make
a daytime horizontal path system.

One last comment (Whew)

See Hardy and Wallner SPIE Vol 2201 Adaptive Optics in Astronomy
(1994) pp 77-87

"Wavefront compensation using active lenses"

Using two lenses that you tip/tilt and dispalce will correct
eight Zernike terms. They tell all.


Clear dark steady and closed loop.

Dan