Re: ATM An $18.00 Null Test

[Aplanatic@aol.com]

Hello again.

It appears that the original Dall Null Test was performed at the RoC of the 
mirror (correct me if I am wrong) and that, as Peter John Smith noted:

"In S&T Sept-76 there is an article about extending the Dall test to other
conics. Mention of varying the distances from a centre of curve position is
made, but the computations were too hard to summarise simply so could not
be included in the article.

I am sure if the author, or Dall, had raytracing software available that
would find the position of, and evaluate the quality of the setup they
might have used it. Computing power is magic in optics !!!!"

So it seems that a non-RoC measurement has been previously proposed.

Indeed, ray tracing IS magic in optical design and test as it allows us to 
easily devise tests that would be impractical to analyze on paper.

Many have commented that the test I proposed requires a very accurate lens.  
I don't think this is so, but this statement is quite hard to prove.  There 
is one great benefit, in my opinion, that has been overlooked.  By moving the 
point source relatively far from the mirror, we are performing a test that is 
more nearly a null test. Obviously, if the point source is displaced far 
enough away, no correcting lens is needed at all to achieve good nulling 
accuracy.  Since much less correction is required of the lens in this test, 
the fractional accuracy of the lens surfaces are reduced.

One could argue that the deviation of the lens surfaces from spherical are 
important.  Yes, this is so.  However, since the lens can be placed close to 
the KE, only the central portion of a larger lens needs to have high 
accuracy, eliminating the usual problems of edge effects in optical surfaces.

One could argue that the quality of the diagonal is of great importance.  
Indeed, this is so.  But many companies are now supplying diagonals of known 
good quality as measured by interferometry.  Once again, it would be prudent 
to use only the central portion of the diagonal, ignoring the outer 5 mm 
where edge effects are hard to control.  If one can't buy a diagonal accurate 
enough for this test, then what about the optical quality of the finished 
Newtonian?

I have extensively analyzed several examples of this test on my ray tracer, 
including the 10" f/6 example given in my original post.  Without delving 
into the fine details, this analysis shows that the sensitivity to parameter 
variations including the lens radii, index of refraction, element locations 
and radius of curvature of the mirror are well within reasonable measurement 
bounds of the amateur.  Non-spherical lens surfaces do cause problems when 
the error is large, but a lens twice the diameter of the light cone with 1/4 
wave, fourth-order inaccuracies still provides better than a 1/20th wave null 
when used in this test.

Unfortunately, the only way to see the accuracy of this test relative to the 
Ross or Dall Null tests (at the RoC) in light of lens or placement 
inaccuracies is to ray trace it for yourself.

Dave Rowe
Torrance, CA
Medium Format Astrophotography:
http://members.aol.com/aplanatic/photos/astro.html