To: atm{at}shore.net
From: mdholm{at}telerama.com
Reply-To: mdholm{at}telerama.com


I'll comment on the replies so far.

1.  The whole cross machine/os compatibility problem is a thorn in
everybody's side.  I'd be just tickled pink if all programs would run
equally well on all systems, but there are a host of reasons, some of them
rather fundamental, why that isn't so.  I won't go into it further.

2. I am going to download Figure XP and give it a test drive.

3. Inches and millimeters work equally well for mirror making.  Getting
hung up over one or the other is silly.

4. I am firmly in the quantitative testing camp.  This doesn't mean I think
star testing is useless by any means, but when figuring a mirror I think
quantitative Foucault, wire, caustic, Hartmann or interferometric testing
done in a reasonably controlled environment with a minimum of optical
elements involved has some real advantages.

Advantages:

   Secondary mirror quality, alignment errors, tube currents, atmospheric
seeing, spider diffraction, are all eliminated or more readily controlled.

   You can write the test results down, subject them to statistical and
analytic mathematical analysis, record them for future reference and easily
communicate them to others for discussion.

   One of the key advantages comes from the availability of modern computer
programs: Data can reliably, quickly and easily be reduced using trusted
methods, with a minimum of possibility for human error, to graphs showing
surface figure in quantitative detail.  "Figure of Merit"
analyses of various types are almost trivially implemented once the
computer programs have done the basic curve fitting.  RMS and Strehl ratio,
EER and others that have real physical optical meaning and are directly
related to image formation are readily available.  This is a Huge
improvement over the situation when I first made a mirror, and I am not at
all interested in going back to the old situation.

Disadvantages:

    You have to build a Foucault tester that can give reliable measurements, or
something more sophisticated.  As a host of ATM's have proven, this is not
a great obstacle.

    You have to learn to make the Foucault measurements.  It isn't as easy to
do as one at first thinks.  (Star, Ronchi and other qualitative tests
really have the same problem.)

    In order to enjoy the advantages of computer data reduction, you have to
have easy access to a computer that can run the programs.

    Many people consider themselves to be "math challenged".  To
these folks,
any delving into the world of numbers is scary or frustrating.  This is a
very real aspect of the human situation.  I do not belittle it, or consign
the afflicted to the dust heap.  When it comes to math ability, I am
definitely in the middle of the pack, better off than some, considerably
worse off than others.  I know how it feels from both ends of the spectrum.
 The best medicine for this problem is ready availability of willing help
from someone who has already gotten over the worst of this hurdle.  This
e-mail list, atm clubs and mirror making classes are all excellent
resources.

So why do I like Sixtests?

It has a reasonably easy user interface.  It implements many of the key
mathematical tools available, needing only a bare minimum of math
understanding from the user.  The result plots are very clear.  It is easy
to compare the results to any conic section and to compare to parabola's of
different ROC. The result plots show deviations from the reference surfaces
directly in nanometers of glass.  I don't have to think about, is this
radius longer than that radius.  I can concern myself with where is the
best place to remove glass and how in heaven's name am I going to get my
lap to remove it from there.

Mark Holm
mdholm{at}telerama.com

--- BBBS/NT v4.01 Flag-4