Re: [ATM] Ronchi programs vs.Foucault testing

[Mark Holm <mdholm@telerama.com>]

jeff newsom wrote:
> Howdy folks-
>   Has anyone compiled info on the accuracy,say, of the Ron Win program, with its ruler function
> as an example,vs the accuracy of Foucault testing? Aren't both
essentially subject to
> interpretation? Halving  never foucault tested , I'm curious how the two would compare.  

Hi Jeff,

In principle, the Foucault test is quantitative.  Except for errors, it
returns a measured surface profile
of your mirror.  In that sense, it is not subject to interpretation.

There are two main reasons that it seems less quantitative than it, in
principle, is.

1.  The judging of zone null positions, central to quantitative Foucault
testing, is difficult.  Not
impossible, just difficult.  It takes a decent apparatus, decent test
conditions, decent eyesight,
patience and some experience to do it well.  None of these is a serious
hindrance to most
mirror makers.  It is actually quite important to have a test setup that
allows you to be physically
comfortable while testing.  Nothing kills testing accuracy and
repeatability like fatigue.

2.  Many mirror makers refuse to deal with the mathematics involved.
There can be many reasons
for this.  Some are:

a.  Some mirror makers use methods of testing and semiquantitative
measurement methods that
 work for them and are, for them at least, less involved than
quantitative Foucault analysis.  Pro's
and serious amateurs often set up null tests, of which autocollimation
is probably the best.  This
makes qualitative versions of the Foucault and Ronchi tests much more
suitable than they would
otherwise be for parabolic mirrors. (Setting up a null test is not
without difficulty/expense.)

b.  In the days before personal computers, the number crunching was
something of a barrier.
The availability of Foucault analysis programs (my favorite is Sixtests)
greatly reduces this problem.

c.  Some are math impaired and have difficulty working with quantitative
analysis, even when a
program does the donkey work.  It helps a lot if you followed math
pretty well at least through
the level of analytic geometry and the conic sections.  You don't really
need to dust off your
theorem and proof skills, but it is helpful at least to remember the
concepts.  A decent understanding
of first year calculus is a real plus.  You don't actually need calculus
to use Foucault analysis, but
the concepts of analytic geometry are fleshed out much better in terms
of calculus.  If you are familiar
with calculus concepts, Foucault analysis seems a lot less mysterious.

Some experiments with automated Foucault analysis have been very
successful.  This shows that,
when the human limitations in judging zonal nulls are removed, the test
really does fulfill its quantitative
potential.

Quite a few amateur and professional mirror makers have gotten good
enough at quantitative Foucault
testing to reliably produce good mirrors using it as their main
quantitative guide.  Automation isn't an
absolute requirement.

There are aspects of mirror quality that are not well judged with the
quantitative Foucault test, but most of
these are actually rather easy to judge with the qualitative Foucault
test and/or the qualitative Ronchi test.
Astigmatism is the big exception.  No simple version of the Foucault or
Ronchi tests is good at detecting it.

A decent introduction to wave optical theory is useful for understanding
both the quantitative and qualitative
versions of most all optical tests.  You can use simpler, ray theory to
understand the basics, but there are
phenomena that you will see in every optical test that simply can not be
explained in ray optics terms.  This
is nice background to have, but isn't required to use any of these tests.

One of the reasons I like Sixtests to analyze Foucault data is that Jim
Burrows, its author, is pretty solid on
both math and wave optics.  Sixtests packages some pretty sophisticated
math and optics in a quite usable
interface.

Mark Holm
mdholm@telerama.com

_______________________________________________
ATM mailing list http://www.atmlist.net/