ATM Tri-space Addendum

[Roger Ceragioli <rogerc@dakotacom.net>]

I just played around with the tri-space design that I posted earlier this
evening, investigating the tolerances on the curves, thicknesses and
spacings.  Unfortunately, I don't have ZEMAX at home, so I did my work on a
much simpler, less versatile program.  But what I found was that the
tri-space is very forgiving of fabrication errors.  I purposely changed all
the curves by fairly large amounts (i.e. ~1/16" on the shortest radii,
~1/2" on the medium radii, and ~1" on the long radii), such as might occur
when using just a spherometer.  I investigated various deviations: random,
making the flints stronger than the crowns, and vice versa.

What I found was that even if you're pretty sloppy, you can recover the
performance of the system by varying r4 of the objective and aspherizing
r1.  The apo performance can be restored quite well.  The residual coma is
not visibly affected.

So it seems that this type of design deserves more study by ATMs dedicated
to making lenses.  If you can tolerate a long tube and can adjust the
centration of the corrector, you could have a large apo for a fraction of
the cost of a full sized ED element.  Even a 12" tri-space would only
require a modest 4.5" piece of KzFSN2.

Roger Ceragioli