"David A. Scott" wrote in message .1.4...
Louis Scheffer wrote in
:
[...] They just need to
make one or more of these mirrors not quite flat. This restores
diffraction limited optical performance without adding any new
elements.

But since not flat is harder to make than flat. Is it still as good
as it could have been if they did the job correctly the first time.

This is a reasonable question, but the answer is yes. The goal is to
make all light paths equal length. If the primary mirror was perfect,
then this mirror should be perfectly flat. With the mirror as is, it
should be slightly (a few microns) curved. In either case, what
determines the image quality is how far it deviates from the desired
figure. You are right that it's harder to make, but since it's small
and easy to measure in the lab, it's not too much harder, and the
surface accuracy should be very similar. Efficiency losses due to
surface accuracy are negligable after some point (lambda/14 is the
cutoff usually used for radio telecopes, where these are called Ruze
losses) and the mirror can be made much more accurately than that.

The same technique is used, on purpose, in other telescopes where it
removes much larger deviations. For example, Arecibo is a sphere, not
a parabola. But with a few extra mirrors (which are not even
remotely close to flat) they correct to diffraction limited
performance. I believe optical telescopes designed for wide fields of
view do similar tricks.

So the short answer is, yes it's a little harder to make the mirror,
but once it's done performance is identical to what it would be with a
correct primary and a flat mirror.

Lou Scheffer