a thought
Spline, your comments seem to be directed at the "noisy" performance of PWM controllers in general. As far as lamps or DC motors go, the load essentially incorporates the low-pass filtering for free. For LEDs and such, choosing a PWM frequency above what the human eye detects as "flicker" is necessary. I used to think 30Hz was adequate, but recently ran across a medical article that suggested 5 - 100Hz is a dangerous zone for photosensitive epilepsy (~ 10% of epileptics, according to the study), with around 50Hz being the worst trigger point. It may not be a bad idea to keep those multiplex rates above 100Hz.
Another means of lowering filter constraints at the extremes of modulation (near 0 or 100%) is to use a numerically controlled oscillator instead of pulse-width modulation. Basically, you replace the counters in the classical digital PWM with modulo (wrap-around) adders. Also called "distributed" PWM, this technique has been described several times in the 90's design literature (EDN), both for use with micros and with discrete logic.
OK, here's the thought:
It's amazing to me how simple tasks like a light dimmer, driving a 7-segment display or a hay-baler (another recent thread) can generate such interest and different ways of solving a given problem.
I found myself getting a little annoyed at the lamp-dimmer thread, since it seemed to be a purely digital problem to me (I have a large digital hammer and everything looks like a nail). Two pushbuttons in (digital) and PWM out (also digital). From a design standpoint, it seemed asthetically wrong to convert to the analog domain and then back to digital in order to solve the problem.
In fact, I came up with a simple, purely digital solution using a couple of inexpensive HC logic IC's. Rather clever too, I thought.
Then, I ran across the links I posted recently. EPOTs are ubiquitous and cheap these days, unlike the situation several years ago. Coupled with a simple schmitt-triggered astable multivibrator, it was a nail in the coffin.
I guess my point is to follow the Occam's razor principle, and try to go with the simplest solution that will do the job reliably. Not exactly what Sir William said, but close enough.
Try not to get married to your ideas, just because they're yours. I think there's a tie-in with the not-invented-here (NIH) syndrome.
- CAL