What Am I Doing in the Shop to Hurt Myself?

Many years ago, when I was in junior high, I was on a telescope mirror making bender.  I bought a number of Pyrex mirror blanks and the required abrasives and polishing compounds.  This is a task that requires more time and patience than I had at that age, and these rapidly turned into oddities on my shelves.

When we bought our first house in San Jose in 1986, I decided that I now had enough room and time to complete one of these projects. I had originally bought two 3" diameter, and two 1" diameter blanks.  (You buy these in pairs, because one will become the mirror, the other the grinding tool.  You can use a plate glass tool, but it was widely believed then that using two identical pieces produced a better mirror.)

My original "There's nothing that I can't do, I'm in junior high school" plan was to build a Gregorian reflector which uses a paraboloid primary mirror and an ellipsoidal secondary, which sits at the focal point of the primary mirror and reflects the image through a hole in the primary.  By adulthood, this was so obviously too hard for a first mirror that I decided to build a 3" f/8 Newtonian reflector.

If you have spent anytime looking at commercial telescopes, you have likely noticed that 3" reflectors are nearly always 3" f/10 spherical mirrors, not paraboloids, unlike larger aperture telescope mirrors.  The reason (which I should have researched) is that a 3" f/10 spherical mirror is within 1/4 wavelength of a paraboloid, and so close enough that it is hard to see a difference for telescope purposes.  But as usual I stumbled forward, making a 3" f/4.5 paraboloid (I went a little deep on grinding phase: hence f/4.5).  It turns out that the final figuring phase of a paraboloid mirror involves moving glass in 1/4 light wavelength depths from the surface.

How do you know you have achieved a paraboloid accurate to fractions of a wavelength of light?  A Foucault knife-edge tester.  This is a shockingly simple device, which allows you to verify paraboloid accuracy as long as you can measure movement of the knife-edge to and from the mirror's focal point in thousandths of an inch.  In a more primitive time, you did that with a nomograph similar to this:

The problem is that as the mirror diameter shrinks and the focal ratio gets shorter, the distances to measure get finer and finer.  This makes getting 1/4 wave accuracy (the usual minimum standard), harder and harder.  These days, with what I have learned, I would have mounted the knife-edge on a 1/4"-20 screw, put a 50 division knob on the end and no problem.  But I have learned much since those days, which nostalgia is making feel better by the day.  (It was not a good time.)

Amusing story: You do Foucault testing in darkness or near-darkness.  My wife came home, and the garage was completely dark except for a very faint light coming from the tester.  She was utterly confused by what she saw.  In later years, she would tell people of this and say being married to me was many things, but never boring!

So, I ended up with a 1/2 wave mirror when I decided to not put any more effort into it.  Not great, but about what the Hubble telescope had on its first failure.  It made a pretty effective, very light and compact telescope.  But because of its size, many of the individual components had to be cobbled together, with sometimes ugly results. 

A mirror cell holds the mirror in the tube and allows for adjustment of the face to the path of incoming light.  At the time in the 1980s, there were no web pages and finding parts was as second millennium as you might expect.  The best that I could find was a mirror cell for a 3 3/8" diameter mirror.  Using it as a source for parts, and with my limited tool experience of the time, I created this Frankenstein:

That is foam tape shrinking it to 3" diameter and the base that mounts in the tube is oak (which taps well and holds bolts better than you might expect).  I am now machining a mirror cell out of aluminum.  More details after my finger stops reminding me of my stupid.