ATM Autocollimation Testing via Flats

[Roger Ceragioli <rogerc@as.arizona.edu>]

Hi, Colin.

I've used the Ritchey-Common test a number of times over the past few 
years and here's my little bit of knowledge/suspicion.

As others have said, it's a challenge to align the mirrors for the test, 
but it can be done o.k. starting with a flashlight.  The equations given 
in Tex and Malacara for deducing the residual curvature of your flat 
assume a geometry in which the flat is tilted 45 deg. to the spherical 
test mirror.  This produces a dim foreshortened (i.e. elliptical) image 
on the return.  Silvering your flat, as well as your spherical mirror 
(which is absolutely necessary), can help a lot.  In order to do the 
full test, you need a knife edge which cuts into you pinhole source at 
two right angles, as Tex describes.  Not too hard to set up, but a small 
challenge.  If you're just testing for zones so ROC of the flat, even a 
Ronchi grating can work.

The zones, if there are any (as there usually are) appear elliptical 
too.  This can make isolating them and measuring their position/width a 
challenge.  The test is extremely sensitive to slope errors, and so 
light even trivial zones can look quite bad.  Moreover, it can be 
extremely hard to make a flat without a moving spindle.  Even with a 
spindle, you often get zones when you wouldn't during normal spherical 
work.  Tom Waineo did an article in one of the later ATMJ volumes which 
is very useful to read and covers some issue not touched in the ATM 
volumes.  You might wish to read it.

The 45 degree geometry also means to you can test a larger flat using a 
smaller sphere, but the limit will be a sphere which is 0.707 times the 
size of your flat.  So your 8" sphere is too small for a 16" flat, 
unless you increase the tilt between flat and sphere/light source, which 
is a bad idea.  Also with a subdiameter test sphere, you vignette the 
top and bottom of your flat, further confusing the return image.  I.e. a 
12" diameter reference sphere will work much better.

What about a null lens for your mirror?  I large Ross null lens, say 8" 
dia. wouldn't be too expensive or difficult to make compared with a 16" 
flat, and could be used for may other mirrors too.  Takes some 
calculation and ray tracing to work out the test geometry but it's not 
too bad.  Personally, I strongly dislike autocollimation testing, though 
Mike Spooner and others on the list are quite good at it and can give 
you lots of pointers.  IMHO the knife edge with Couder mask with star 
testing for confirmation can do a good job on your mirror.  But that's 
just me.

Two problems with the autocollimation test: 1) unless the parabola and 
your flat are quite square with one another AND you can put your light 
source/eye on the same axis, then rays striking the parabola on the way 
out will not retrace the same path, which causes a slight averaging of 
errors on the surface; 2) if you use the Ronchi test to evaluate the 
asphericity, then remember that if you put your source/eye inside or 
outside of focus in order to see several Ronchi bands, you will not be 
testing at the focus and infinite conjugate.  In other words the Ronchi 
bands SHOULD NOT look quite straight for the correct parabola. 
 Likewise, beware of using the knife edge test or Ronchi if your 
source/eye are NOT at equal distances for the mirrors.  At focus, though 
the mirror should still null out.

Well, enough blabbering for now.  Good luck!
Roger Ceragioli