[Raytrace] Re: A useful sequence

[Peter John Smith]

Peter Chen wrote

> I went through Peter Smith's suggested sequence.  It worked well and was
> very instructive.
> I had no problem getting up to the Newtonian, Cassegrain, and thru the
> Dall-Kirkham.  However, the optimization routines (Hammer, Golbal)  would
> not converge when going from Dall-Kirkham to RItchey-Chretien when the
> secondary radius, primary and secondary conics were set to
> 'variable'.   After several attempts, I found that if I started from a
> Cassegrain and set the merit function to 'ptv spot size' rather than 'rms
> spot size', it converged.
> So I guess the lesson is that, in raytracing, it helps to know where to
> start.  Am I correct?

The sequence I suggested I had checked with Zemax just to make sure that it
worked.

But I just used the plain optimiser which is far more direct and if you have
a good idea of what you want much faster.

For the other people, Hammer and Global optimisers are a special feature of
Zemax to explore possibilities far outside the parameters chosen and this
will be irrelevant to most people.

I have never been able to get much useful data from the Hammer and Global
optimisers and they caused a computer hangup on my old computer.  Must try
them on this new one.  But seriously, if you are more or less in control of
the situation they are not needed and even are a liability.

But Peter Chen has raised a point about optimisers.

There are so many settable goals possible that it becomes a bit of an art to
use them in some special cases and you may have to play with some of the
conditions.

Some optics situations are easy with optimisers - others seem laced with
local minima which trick the optimiser and lead it far from the wanted
direction.
Those in this sequence are I believe quite straightforward and the optimiser
should converge nicely even starting a long way from the solution.

A Wynne corrector has many local minima and it is best to start with a
prescription close to the final and only make small changes when playing
with the optimiser.

In general, it is best to start as close as possible to the final design -
especially for complex ones.  Then to chose anchoring constants so the
design stays close to this.  As an example, many eyepieces have a very deep
concave surface close to the image.  If this is not made a constant, the
optimiser very often leads one down the garden path.  Another trick is to do
the initial optimising with a reduced field, then increase it towards the
end.

The least satisfying aspect of using the optimiser is that it is more trial
and error than true analysis.  But, if the program has a good user
interface, the speed at which this can be done and the visual grasp of
changes in performance is such that the trial and error can be very usefully
directed.  James Bond would have killed for these facilities 50 years ago -
and it was becoming available back then in the right circles.  Abbe would
have just thought it was a fairy story.

If anyone has gone this far, I would suggest taking the Dall Kirkham
variation and adding two plane parallel pieces of BK7 about 200 mm in front
of the image.  They can be seperated by a few mm and each be about 10 mm
thick.

Have as the variables the conic of the primary and every surface R.  Leave
as constants all spacings and the conic const of the sec.  Use a reasonably
wide field since this configuration can correct for a wide field quite well.
Up to say 1 degree or more towards the end.  Dont forget to generate a new
merit function for the optimiser - especially after changing field.

Run the optimiser.

If you are lucky, the result will be a reasonable and quite interesting
design.

There are more than one possible promising design types here.  To tie the
optimiser down you will have to make constant some of the surface R values.

For instance, the pieces of plane parallel glass will each assume meniscus
shape.  They can be either back to back or front to front to work reasonably
and to explore each you may need to anchor one of these surfaces.

In Zemax, one of the settable merit terms is the effective focal length
(effl).  I would also set this and leave one of the spacings floating.  Dont
know how to do that in OSLO but it would be worth finding out because this
will be often used to stop the optimiser going crazy.

If you want to play around with this basic design, there is an example on my
web site called the DK11 meaning Dall Kirkham plus two single elements.

Peter Smith.