That all sounds good Cris, but PWM is OUT for quiet fan speed control. I have been working with a PWM-based thermal fan controller & it is not viable
Mike, I think the key to your statement is that the one you've been trying isn't viable
. PWM must be viable since that's how the whole PSU works. My guess is that the unit you have uses too low a frequency ("pulsing causes terrible variable noise"), and doesn't filter the output because inductor+capacitor filters are more than the designer felt they needed to invest in the problem.
Typical 'better grade' switcher/pwm designs use 100+khz, and L-C output filters to pass only clean DC to the loads. Like the PSU and the mobo's Vcore regulator. I know those two examples are high end compared to the target cost fit of a fan controller, but I expect that a suitable lower grade comprimise is possible. One potential snag off the top of my head though is the available source voltage. Most switchers use either high voltage, or invert their own. PSU's use rectified and/or doubled 120v AC, around 170-340v DC (those caps can pack a punch!). Starting at 12v and not inverting might seriously limit the maximum output, like maybe you would loose the top 25% of the speed range. Even a linear regulator will have that problem, but to a lesser extent (approx -.5v).
An intelligent control could also be done by using D to A output to control linear regulators. PIC's use PWM to make up their D to A outputs, but I think the frequency is quite high, and can be very economically filtered at such low power levels. To appeal to the flashy modder crowd, you could even use tubes for the output stages
laugh, there are $30000 audio CD players out there, and I don't think that price even includes the D to A converter
I'm glad the computer crowd doesn't seem to have gone that way yet, and I'm not likely to encourage it, perhaps unlike Aopen