From: "James Lerch" 
To: "ATM List" 
Reply-To: "James Lerch" 


----- Original Message -----
From: "Mark Suchting" 
>
> At 09:24 PM 8/10/2003 -0700, John Sherman wrote:
>
> >A few people asked me if I still have the original interferometer data from
> >my Galaxy mirror. Yes, I do. See www.johnspics.com/hi/hudek.jpg
>
> The shame about this mirror ( in its original form ) is that most of the
> error is just pure overcorrection ....1.6 waves  in fact, which refocuses
> at best fit to 0.4 waves P-V ( with focus subtracted,  as a shown on the
> interferogram). This equates to 15% overcorrection for a 22"
> F4.5.

Hi Mark,

Tell me more about this "refocus" trick...  Having translated a
Foucault data reduction program (I won't say I wrote it, as I didn't!) when
I read "refocuses at best fit" I think along the lines of finding
a best fit Radius of Curvature for the reference parabola.

Once the best fit RoC for the reference parabola has been found, and the
Strehl reported, "Refocusing" to a different reference parabola
has never made the Strehl and RMS values improve (else the code should have
used that one to start with..)

Once the best fit RoC has been found, if the optic shows 15% overcorrected,
the only 'trick' I've found to make the Strehl and RMS improve is to select
a different, non parabolic reference surface (ie changing the Conic
Constant, Deformation Constant, "little b", what ever we want to
call that term that describes on optical surface as spherical = 0,
ellipsoid  -1, parabolic
= -1, hyperbolic = <-1)

I ask all this as I was once told the following:

"... the mirror is indeed undercorrected. I have as much as 1/3 wave
total aperical aberration, which can refocus to about 1/16 wave by my worst
axis measurement (not the two combined)."

I studied that paragraph for some time to try and decipher it.  My testing
showed that the optic was undercorrected with a 1/3.8 lambda (550nm) P-V
wave front error.  Assuming the author didn't change the definitions of
"1/3 wave" and "1/16 wave" mid-stream in the same
sentence, I presumed the author was referring to P-V wave front error as
the original "1/3 wave" is very near my 1/3.8 P-V wave front
value.

After presuming to solve the "1/3 wave" mystery, I set about
trying to solve the
"refocus to about 1/16 wave" mystery.  I first found that
changing the best fit RoC wasn't the answer as this made things worse, not
better.  After a bit of head scratching I found that if I changed the Conic
Constant to "b = -0.92" I got a 1/19 lambda (550nm) P-V wave
front error, well that's pretty close to "1/16 wave".....

Based on my limited experience, and your writings Mike, I wonder if in
certain circles "Refocusing" really means changing the Conic
Constant?  In your case the
optic could be advertised as having a 0.859 Strehl, but it would be false
to claim that Strehl AND say it is a parabolic mirror....

> I note also that the section showing what `terms'  have been removed from
> the data has been chopped off the printout  ie focus , astigmatism. . I'm
> well familiar with what  Zygo `Zap' software printouts should look like in
> completeness.
>
> I would suspect from the three equispaced bumps on the phase map, that
> astigmatism has been subtracted too. This seems to be standard practice
> amongst companies offering interferograms, because of the astigmatism
> induced by self weight deformation, but I would question how much effort
> goes in to analyzing how much real astigmatism these mirrors might have,
> before it is `subtracted out'.

Another interesting point!  I've been trying my best to Star Test Charlie's
mirror (the latest Robo test subject) using an artificial star and a CCD
camera.
Unfortunately Charlie's mirror is odd in the fact that front optical
surface is a larger diameter than the back side diameter of his mirror. 
This gives the mirror a Conical shaped edge, which is near impossible to
support with the mirror vertical for the following reasons:

#1 The outside edge of the mirror is very glossy and smooth, AND its cone
shaped!

#2 Slings and posts that try to support the mirror without touching the beveled
face, cause the mirror to slide forward against the mirror clips as the
mirror nears vertical (needed to aim at an artificial star)

#3 Slings and post that support the mirror ON the bevel (only way to keep it
from sliding forward, off the cell and onto the mirror clips) introduce
"Stupid"
amounts of astigmatism that goes away when the Telescope is pointed at a
REAL star.  Unfortunately the Real Stars move, and the ones that don't move
that much
(like Polaris) are too dim to image without long exposure.

#4 Gluing the mirror to the cell might help solve the "aiming at an artificial
star" dilemma, however Charlie's existing cell is not setup to
properly support that much weight in tension directly from the 9 support
points (the cell works great in compression!)

Current efforts to image the star test results for Charlie's mirror are to
be done on an Equatorial Dob platform and a real star such as Vega.
Currently were just waiting for a clear night. (apologies to Central
Florida observers, the monsoon like rains of late may be our fault
)

> Probably the only way to produce an honest interferogram of a large thin
> mirror is to build a testing tower so that the optical axis can be vertical
> and the mirror lying flat on a flotation cell. Standard practice for large
> professional mirror labs.

Point well taken!

James Lerch

--- BBBS/NT v4.01 Flag-5