From: "Jerry" 
To: "'James Lerch'" , "'ATM
List'" 
Reply-To: "Jerry" 






        This has been a very interesting thread. The work you have done on
this is admirable. I use the old reliable Mark IV eyeball for testing. It
requires less software. But you are coming along so well with this I can
see that Robo-Foucault testing is going to work well and be easily
implemented by ATMs. Easy for others after you have done the hard part.
When you get the bugs out I might just try it.

Last night I gave some thought to this series of posts.

Vladimir Galogaza's comment,

>>>>>
"But robotic determination, I presume,  will be inherently more
precise if performed on caustic. I do not know how much more complicated
that will be than present method, e. g. RoboFoucault from James Lerch. It
would be interesting to have James opinion on this.

Vladimir."
<<<<<<


My comment

>>>>>>>
        When doing the Foucault test you do not focus on the image of the
slit. Not with the eye or, in the robo fashion, with the camera. The object
of the test is to determine the point on the axis that the beam from
opposite sides of the mirror zone cross the axis. That it is not in
"focus (corrected spelling)" at that crossover point does not
prevent the knife on axis from blocking half of the light from both sides
of the mirror zone that the knife is set for.
        The caustic may be a more determinate test. But the there are
probably interference methods that would be easier to automate and be more
accurate than either Foucault or caustic tests.

Jerry
<<<<<<<<<


And Jim Burrows' comment

>>>>
"Finally, it is possible to work Foucault on the image plane - I've
done it.  You replace the KE with a frosted glass slide and move the slide
longitudinally until the returns from the two mask holes merge.

         -- Jim Burrows  "


Which leads to thought of a host of tests at the center of curvature that
would not require a determination of "Gray" to determine the
crossover point on the axis. Most or all would require a mask but would
avoid the problem of determination of "gray". And some at least
could be done with a CCD with no lens, prime focus so to speak. As applied
to the caustic you would find the point that a zone opening focuses the
slit or pinhole onto the CCD. Or in the Burrows variant by locating the
crossover where the image from two openings converge to one. These are good
ideas for variations of Robo-Foucault if determining "gray" turns
out to be too problematic. We seem to all be reasonably sure that
"gray" is where the left and right zone beams cross the axis.
These other tests probably would be easier to program. If you can find the
crossover point on the axis without a mask most would find that preferable
if precision is not lost. That seems to only to leave finding the crossover
by determining "gray" or simultaneous "blinking" of
zone openings. Can someone think of other methods that do not require a
mask?


        Distortion in the camera optics can be plotted and corrected in
software. At least a first glance this doesn't seem likely to be a big
problem and probably wouldn't need to be corrected. But it is still worthy
of investigation

        Heat from the tester is one of those things that I consider. I have
a very bright HOT lamp on my tester. It does stabilize after being on a
while and I have taken some steps to get most of the heat away before the
slit and away from the light path. Heat matters not just because of
dimensional changes in the tester but also convection in the light path.



Jerry

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