Ok guys, so this might sound kindof dumb, but here's an idea. How about, instead of spinning CDs and DVD around, with all the moving parts, etc, why not just use a flat sheet of something like what's in digital cameras?
It wouldn't have to be nearly as good as a digital camera, because it wouldn't need to sense RGB, but just the one wavelength of the laser. And we wouldn't care about how bright the color is, just does it correspond to on or off. And it probably wouldn't even have to be as dense.
Well hold on, let me do some approximations on the density here. Let's assume a CCD in a digital camera is 1cm x 1 cm, and 7 mega pixels. So that's ~2645 pixels across 1 cm, or a resolution down to ~0.000378 cm, or 3.78 microns. From
http://www.usbyte.com/common/dvd.htm I found that CDs have a track pitch of 1.6 microns, and a minimum pit length of 0.83 microns. DVDs have 0.74 microns and 0.4 - 0.44 microns. Damn. Okay, so we need 0.4 microns resolution, which means 0.00004 cm, so 25000 pixels across, or 625 mega pixels. Although, instead of RGB triplet sensors, we'd only need an emmitter and a sensor pair.
Okay, so covering a surface the size of a CD or a DVD would probably cost a lot of $$$. But, getting 0.4 micron resolution shouldn't be impossible, given that the Pentium was built with 0.25 micron lithography, and 0.09 is mainstream. So maybe we could have a little sensor, and have it move up-down, left-right, instead of spinning the disc?
No no no, how about this: There's a little thin stick, and it's covered with this sensor thing, and that stick is as long as the radius of the CD/DVD (6 cm), and it spins in circles, reading the entire length of the disc at once. So, there'd be no seeking, and in one rotation it could read the entire disc (one layer). So, you'd only rotate the stick as fast as the sensor could sense, and as fast as it could actually transmit the data, so probably less than 1 rpm, which would take very little power, and practically no noise.