This topic started in another thread, but it was a bit off-topic there so I'm starting this one.
In the other thread I said
"Perhaps Texas Instruments is simplifying the matter too much, but their explanation of how a single-chip DLP projector works ( http://www.dlp.com/tech/what.aspx ) says, in both text and graphic, that the DLP chip is struck by only red, green and blue light (the light that passes through the color filters on the color wheel), thus only red green and blue light hits our screens using such a PJ, and only one color at a time. It is the "persistence of vision" of our eyes that makes us see the full visible light spectrum. As I'm sure you know, it is the same phenomenon that lets us watch a CRT television without seeing the even and odd scan fields, but rather one image that doesn't flicker.
If what T.I. says is true, I don't understand how iridescence can come into play with such a PJ. "
Bill misunderstood me (unless I'm misunderstanding him

) and thought I was talking about a screen, such as the one by Sony (Chromavue?), that is comprised of transparent red, green and blue layers on a reflective substrate. I actually made a mini-version of such a screen using layers of red, green and blue transparent plastic and it simply doesn't work for DIY.
For those coming late to the party (

), it is a known fact that the mica particles commonly used in paints to create a metallic appearance act is prisms and break up white light into multiple colors. If a projector that shoots white light is used with a screen that was painted with a mica-based paint, color-shifting of the image will occur due to this prismatic effect.
What I'm getting at is that a simple one-chip DLP projector NEVER shines white light on the screen (see link above). Such a PJ shines red, green and blue light on the screen, and only one color at a time, never two or three. These also are the only colors reflected from the screen (except for any ambient light present). The reason we see the full visible spectrum of light is because of eyes "mix" the colors from the RGB light. I find that concept fascinating!
So, this means that even though the screen may be very prismatic under white light, when used with the PJ described above, the screen can, at best, break up the green light into a very narrow spectrum; and the same for the blue light. The red light from a PJ is almost monochromatic (one single color). There is a chart someone did of the individual spectra of their PJ's output, but I forget where I saw it.
It would be interesting to create a very prismatic screen on purpose and then see how the images looked when it was used with DLP and LCD PJ's!