So, in an effort to highlight the dangers of having Zemax, a 3D printer, and too much free time, I made a 3D-printed guided spectrograph. It actually works, and pretty well, so I thought I'd share some details. Pictures below if you want to skip to the "it works!" part.
The design is a classical spectrograph, using a 600 lines/mm reflection diffraction grating from Edmund Optics. The spectrograph features a "slit" consisting of a 35-micron laser-drilled pinhole from Lenox Laser. The pinhole is drilled into a very thin sheet of stainless steel, the surface of which acts as a mirror to send light to a guide camera.
Light comes into the spectrograph and is focused onto the pinhole plate. Most of the target starlight goes through the pinhole and into the main spectrograph optics. These consist of a 20mm diameter x 80mm focal length achromat from Anchor Optics. The achromat collimates the diverging beam which then hits the 1/2" diffraction grating. The nominal design has the grating tilted 9.5 degrees to the incoming beam. This diffracts light at 5700A at 30.5 degrees, meaning the outgoing beam is tilted a total of 40 degrees to the incoming beam. This angle and wavelength were chosen so that the H-alpha line would fit on the CCD, but more on that later. The diffracted beam then goes through another 20mm x 80mm achromat and is focused onto the CCD, in this case an Atik 490EX monochrome camera.
Meanwhile, some of the starlight that doesn't make it through the pinhole is reflected by the pinhole substrate, which is tilted 15 degrees, off a 1/2" flat mirror, and through two 12.5mm x 45mm achromats, also from Anchor. These relay the image of the pinhole and field of view of the telescope to the guide camera, a Starlight Xpress Lodestar.
Initial setup involves illuminating the pinhole plate (by shining a flashlight into the scope) and focusing the guide camera on the pinhole. Focusing is done simply by sliding the cameras in and out of their 1.25" openings, but works surprisingly well. The XY pixel position of the pinhole is noted. Next a star is focused on the pinhole plate as seen by the guider. The guider is calibrated in the usual manner in MaxIm DL. Then when guiding is started, the XY guidestar coordinates are manually entered to match the coordinates of the pinhole. MaxIm then drives the guidestar right onto the pinhole! It guides surprisingly well! The first light test was done with a C11 at f/10. Despite the questionable surface quality of the pinhole substrate, the images look fine and a 9.1-mag companion to one of the stars was visible in a 0.1-second guide exposure! The stainless steel surface works quite well!
Focusing of the spectrograph camera is most easily done by pointing the telescope at a compact fluorescent lightbulb or other line source and focusing on the narrow lines. The source need not be focused in the telescope itself.
For wavelength calibration, I simply held a neon bulb in front of the telescope (and stuck the dust cover over it to block stray light) and took a long exposure. I wrote software that lets me identify the known lines and will produce a wavelength solution, as well as a program to extract the spectra and fit the wavelength solution. I use iSpec, a free Linux program that is available in a self-contained virtual system that can run under Windows. It can normalize the contiuum, find elements from lines, find the radial velocity of a star, etc.
I designed the 3D printed housing in two halves, top and bottom. This makes it feasible to print and easy to assemble by simply dropping the optics into their slots and then putting the top on! The two halves bolt together with screws and hex nuts. I couldn't print an accurate enough 1.25" nosepiece that wouldn't wiggle in the scope, so I revised the design to hold a metal T-thread-to-1.25" nosepiece. That helped considerably. I would change how I mounted the pinhole as I didn't have a precise way to center it. This changed the available wavelength range and eliminated H-alpha. It ended up spanning from 4065-6305A. Print time for each half was 6.5 hours at low resolution on a Makerbot Replicator 5th generation printer.
For first light I pointed it at a few bright stars of various spectral types, and a high-pressure sodium light. All look about as you would expect. Hydrogen Balmer lines in the B and A type stars, lots of metal lines in the K and M stars. The sodium doublet is resolved in the stars that show it. I calculate the resolution to be about R = 1000, or about 5A. That is pretty much what the theoretical calculations give, so I guess it works! Exposures are only 1-5 seconds, depending on the star. I plan to try some fainter stars next time out.
Total cost for the four doublets, fold mirror, diffraction grating, and pinhole was about $250. I didn't keep track of the amount of material used by the printer, but probably less than $10 worth. $25 for the 1.25" nosepiece plus a few bucks in screws. Under $300 for a fully functional guided spectrograph. Not too shabby!
The images show the optical diagram, a cutaway model, the spectrograph being assembled, some stellar spectra, and the assembled spectrograph with cameras an a Stellarvue SV60 because it's funny.