I work as a coputer tech (fixing pcs) in a major university as part of
> a work study program. Im also a junior in computer engineering, so I
> know my way around a breadboard.
> One of the machines is a newish dell btx, that's having major cooling
> problems (random page faults, hd temp goes up to about 130 F, etc). I
> narrowed the problem down to the main case fan, which is a four pin
> PWM (pulse-width modulation). Apparently, the bios (happens without
> hd/os) forces the fan to spin at a ridiculously low speed,
> approximately 10% of capacity. After tinkering a bit, I found that
> cutting the blue (PWM) wire was adequate enough to disable pwm, and
> forced it to 100%. But now it sound's like one of those handheld
> vacuum cleaners, and is likely to have a very low MFT (mean failure
> time). I've tinkered a bit more, and found that by grounding the pwm
> with a suitable resistor, it slows it down to an acceptible level.
> I've run a few tests, and I've come up with the following data:
> Fan Power draw at 100%: ~375 mA
> Ideal Power draw (flow vs sound): ~210 mA
> Pwm voltage (fan to ground): ~3.266 V
> Pwm Current (directly grounded) ~0.52 mA
> Ideal Pwm>Resistor>Ground: ~3.2 kOhm
> I was wondering if anybody had any additional input about this before
> I screw something up royally. With these Ideal values, everything
> seems to work fine, and I don't notice any risky voltages or
> currents. I'm on a tight schedule, so I'm likely to begin soldering
> everything into place soon.
If this box belongs to a customer, don't install any hacks. If you're stumped, ask the supervisor to help or give it to one of the other techs. It sounds like you need either a new fan or a new MB. But I would not trust a fan that's got a resistor to the pwm lead, and I would not trust a repairman who hacks on my equipment trying to fix some problem he doesn't understand.
Good Luck! Rich