Quote:
Originally Posted by
amesdp 
Actually no, it doesn't. See
http://en.wikipedia.org/wiki/Dlp: "The "Rainbow Effect" is unique to single-chip DLP projectors."
I think you're getting confused about terminology. A 3-chip DLP still uses time-based modulation as a single-chip DLP - it just does it separately for each primary color at the same time, rather than projecting each color sequentially. I agree that it's not quite right to call it pulse-width modulation - but the principle is the same as the other mechanisms for modulating color by the length of time that each color is projected within a small time window.
wikipedia is hardly an authoritative reference, and if you read below in the same article it also states:
"Three-chip DLP projectors can resolve finer gradations of shade and color than one-chip projectors, because each color has a longer time available to be modulated within each video frame; furthermore
they have a reduced potential for flicker and rainbow effect"
Now, again, as I stated above 3-chip DLP also creates colors temporally. The positions of each DMD mirror element has only two positions, and there are three DMDs aligned together. That is 2^3 or 8 possible combinations. The combinations are: White, Black, G, R, B, C, M, Y. All other colors are created by dithering these 8 colors temporally.
Note how 3-chip is different than 1-chip in that white is not created by sequential R,G,B. Note also how each DMD element has the full time available to it, and is not restricted by colorwheel timing. These factors lead to greatly reduced or practically eliminated rainbow-type artifacts. The artifacts too, are different in nature from those in 1-chip DLPs, but they are not fundamentally gone or fundamentally something else. They are still the effects of temporal dithering to create colors other than the fundamental colors possible at any instant in the system.
I stand fully by what I said above because it is correct and true. Now, if you want to quibble about semantics about whether the artifacts possible in 3-chip is not the "rainbow effect" that's fine, but that's not an argument I'm interested in having.