as said before, when i played around with PWM control i ran into the problem of sizing this output capacitor (cap). as someone posted before, by using such a cap (and probably an inductor) you can convert the PWM signal to an analog voltage. if the cap is too small the cap's charge wont be able to hold the voltage for the supplied load and will have almost no effect. if the cap is too large, the cap will be loaded during the PWM pulse but wont be discharged completely by the load during the PWM pause. this results in a limited control on the lower PWM range (e.g. PWM 0%-20% results in 9V-12V, everything above 20% = 12V) and that's the issue i ran into. the cap needs to be sized correctly for the individual load to get full range rpm control!

the internal circuitry of some fans prevents them from working well on PWM.

i have some nzxt fans reacting well to PWM but i also have some sflex which have very limited control when used on PWM.

almost ever you will get erratic rpm readings at lower rpm as power to the rwm reading circuitry is switched off and on periodically at the PWM frequency which overlays to the rpm signal itself.

PWM frequency also plays a role here, this is the frequency at which the PWM signal is switched on an off.

as mentioned before, this ticking sound comes from the fan periodically switched on and off. if PWM frequency is in a range above the frequency we can hear, above 20kHz (which is also intels spec) the ticking sound problem goes away, but unfortunately lower PWM frequency works better in controlling the fan speed than higher frequency does at least with the fans i tested here.

to me using a more costly and inefficient analoge (or linear) voltage control circuitry is the way to go. i don't really care about this additional one or two watts per fan when the whole system burns hundreds of watts already. i like the idea of PWM and
  • yes it will be more efficient, thus produce less heat

but if it
  • does not allow me to control all of my various fans
  • prevents rpm reading at lower speeds
  • makes this ticking noise

its of no use to me.

this all holds true for 3-wire (non PWM) fans only as 4-wire (PWM) fans already have their own PWM controller with their own switching transistor built in and use the PWM signal supplied on the fourth pin just for controlling the built in switching transistor.

martinm, as posted by bing: your controller is just generating the control signal of 4-wire (PWM) fans for their built in circuitry with its own switching transistor. the cap we were talking before is the output cap behind the switching transistor where you have no access to.