A while back mental sent me a link to a DIY spectrometer. I finally assembled all the parts (with assistance from Lynne).
My modifications of the original design are: Glue the dvd vertically in front of the camera (because I couldn't work out how to attach it to the camera itself). The slit is made from two pieces of copper flashing I happened to have in front of me.
As you can see, I'm getting a fair amount of bleeding from the CCD when it is over exposed (the camera is as well focused as I can make it, and the slit is about 0.7mm.
top to bottom: CFL, CFL blacklight, red LED, amber LED, green LED, green LED again with blooming, blue LED, blue LED shining through piece of fluorescent orange material, white LED 1, white LED 2.
The white spot in the top spectrum is leakage light from a small hole in the case.
The reason the CFLs are tall is because they are surface emitters whereas the LEDs are point sources.
I believe the alignment is fairly accurate, so you can see that the white led is excited by the same blue as the blue led. Also, the CFL emits a fair amount of UVA, matching the blacklight emission spectra. I don't know how far into the UV I can see. I can definitely see IR from remote controls (but it's hard to take a picture because the led blinks on and off :).