I updated my old 2001 Ronchi test code to support arbitrary conic surfaces, and then uploaded the code to github.

My current 12.5″ project is an f/5 paraboloid:

If we were interested in making a spherical mirror, we’d have these as patterns:

Why bother writing this code? I’m actually interested in trying to acquire images using something like the Raspberry Pi camera, and then generating a matching pattern via an image optimization process. Having this code might be useful.

Addendum: Somewhere tucked away I have a mirror I did for a 6″ f/12 mirror. I think that I never really did a proper Foucault test on it, because the focal length is so long, I think I just figured it to as good of a sphere as I could. I thought I’d use my code to see what it would look like with the same offsets using the Ronchi test. Here’s the resulting patterns:

As you can see, there is a small hint of correction, most visible just outside of focus. I’ll try to dig out the mirror for testing this Friday, as well as a 6″ f/5 chinese reflector I have, which I think is just a sphere. With any luck, I’ll also get a new pitch lap made for the 12.5″ mirror.

It was a long time since I wrote the code that I used for Ronchi code, and while I had some confidence in it, I wasn’t 100% sure that I had verified it. So, today, I took a copy of Ronchi for Windows 2 (I downloaded it from here) and set it up to predict the patterns for a 100 LPI screen and the 126″ radius of curvature that my current 12.5 mirror would have. I had it generate the patterns at six different offsets. Here’s the result:

I then took the same measurements and offsets, and added them to my own program and generated the similar array. Here are my results:

They look pretty much dead on. I can feel pretty confident about using my own code.

Addendum: This used to be more important to me, because I often didn’t have a Windows machine to run programs like Ronchi For Windows. In fact, I don’t have a box that runs Windows now. To write this article, I did confirm that using the Wine compatibility layer to run it on my Linux box, without the hassle of using a virtual machine or some such. Nifty.