I have wanted to build the HDD pov clock for a long time, and finally got the L293D chips in the mail the other day.
after hours and hours of experimenting [I cannot believe I'm still up at 2AM working on it :o ] the most I was able to get was about 1200RPM.. and even then, if I bump or slow the motor it can stop and just chatter.
Here are a few key things I want to mention: I glued a small neodymium magnet onto the platter. I'm using a hall effect sensor, and I'm having some kind of magnetic interference. I have the 6 steps that it cycles through for each quarter-turn and when I write the 3 outputs, I first turn off an ENABLE pin, write all 3, then turn ON the ENABLE pin. However, if I connect the enable pin [on the breadboard], it causes the hall effect sensor to trigger false, and instead of once per revolution, it triggers multiple times a rev [I'm assuming back-emf?].
so for now I just leave the enable pin alone, or tied to VCC, and the sensor works.
Oh and the other thing, I have 2 L293D's in parallel so to speak. I hooked up 1, and it was getting too hot at ~500mA, so I put another one above it and put a jumper on each pin that was being used. shared the load, not quite even but it ran cooler.
that's all I can think of.. oh here's the full code: http://pastebin.com/8bQZWcFA I switched from millisecond to microsecond delays so the comments of the code might not all be 100% correct.
the main things I'm wondering: is not shutting off the power before I switch polarities [due to interference I get when enabled] limiting my max speed? if so, if I get an infrared sensor what speed can I expect to gain?
and for the project I'm thinking of [http://www.ian.org/HD-Clock/ -- I haven't cut the slot in the plate yet] would ~1200RPM be enough for POV? I'm thinking no, because it's 20 R/second, i.e. 20 FPS, and 24 FPS is what smooth animation looks like. [it would look choppy I guess?]