The setup make a fairly good reproduction of the MSE spectrum, except: I think the MSE filters are a little bit wider than the real MSE peaks. The spectrum will only be exact on axis (of the image) as the MSE filters will vary wavelength with FOV, depending on how they are aligned in the end. Required: A optics bench with lots of space off one end to a wall. 3 bulbs and power supplies (preferably intensity controlled) 3 small lenses (Mine were 100mm) 3 narrow MSE filters (Borrow old ones from Matthias). All in manually adjusted veritcal and horizontal tilt stages. 2 semi-silvered mirrors (Polarisation preservation isn't important, but they should be ~50/50 transmission if possible) 2 polarisers, one in an accurate manual adjustable rotation stage (~0.1°). 1 Spectrometer - good enough to measure the peak wavelength of the MSE filters to ~0.2nm (My spectrometer was only 1nm resolution, but you can calculate the centre of weight more accurately than that). A λ/2 plate at the correct wavelength, on a stepper controlled rotation stage. (you could ignore this and do the setup with a polarliser if you don't want to rotate the polarisation later) 1 visible laser for alignement. - In place of the camera, setup the alignment laser horizontal and parallel with one row of holes in the bench, and mark where it hits the far wall. - Working backwards through the system, add each component except the filters and adjust them so that the laser hits the wall in the same place, and the small reflection from that component hits very close to the laser entrace hole. By the end, you'll have lots of dots on the laser. - Get all three bulb filaments in the laser beam. It's important to adjust the 100mm lenses first, to get the laser on the walls correct, and then do the bulbs afterwards. - Setup the spectrometer behind the pi/sigma semi-silvered mirror. It will probably put an extra dot on the wall, and not on the laser (because it's behind the mirror). - Replace the laser with the camera and turn on the first pi arm (pi-) bulb. - Manually rotate the pi arm polariser to be as close to vertical as you can. If your camera is very sensitive, you can do this by maximising the very slight variation of the intensity caused because the S reflection of the mirror is probably stronger than the P, but it doesn't need to be perfect. - To align the sigma polariser at exactly 90° to the pi: - Put a polariser in front of the camera, at any angle. - Rotate the λ/2 on the stepper through a few full cycles, fit the sine wave and return the motor to the position of max signal. - Turn off the pi- bulb and turn on the sigma bulb. - Manually rotate the sigma polariser until you get the best extinction you can (i.e. minimise the signal). The pi and sigma light should not be precisely at 90° to each other, and roughly aligned to the S/P of the last mirror. - Turn on all three bulbs and adjust the bulb positions so their three images overlap on the camera image. - Insert the three filters and get them roughly perpendicular to the the optic axes. The spectrometer behind the semi-silvered mirror won't report the same intensities because of the mirror, but it should give the correct central wavelengths if it and the camera are both on the axis. - Turn each bulb on individually and adjust the filter tilts to be where you want them to be. - Adjust each bulb individually to set their intensities correctly on the camera (pi+ = pi- = 1/2 sigma) Other Notes: To get the polarisation and spectrum good, the system has to be quite long, so the bulbs have to be quite bright because most of the light is lost to vignetting. The sigma bulb is in direct line of sight of the camera so watch out for the IR. The MSE filters and film polarisers are transparent to IR. This confused me for hours. LEDs can work, but you need to make sure they are very bright in the correct wavelength.