Electrical – Use motherboard PWM to control non-PWM fans with MOSFET

PWM frequency is 32 kHz so that's a period of about 32 us and half that time the transistor will be dumping energy into charging up those capacitor. It'll be doing it totally inefficiently too so maybe half that energy again is wasted in transistor heat.

Capacitor acquired energy is \$\dfrac{CV^2}{2} = \dfrac{2.2 \times 10^{-6}\times 144 }{2}\$ = 0.158mJ

BUT because charging the cap is done totally inefficiently, at a guess, the transistor will throw away just as much energy. It does this 32,000 times per second so that's a power of about 5 watts per transistor.

Now I may have over-egged this a bit by assuming that each 2.2uF would become totally discharged in one cycle of PWM but I'd wager that each transistor is doing over a watt just driving these capacitors back to full charge (32,000 times per second). Get rid of the caps and if you still have problems with noise then come back and tell us.

The lower the duty cycle the worse this will be because there will be a longer time for these caps to discharge and hence inefficiently re-acquire more of their 0.158mJ when the transistor switches back on again.

All DC motors are actually AC motors with some sort of commutation from DC with some DC coil resistance, DCR.

The BLDC draws excitation current from the coil resistance. (I=V+/DCR). As average voltage increases and it overcomes stiction, it starts spinning and the commutated coil impedance now is added to the DCR.

HOWEVER, an internal filter capacitor is needed by the fan to reduce incoming and outgoing voltage ripple, caused by current commutation in the poles.

• If you pulse that capacitor, with a non-PWM controlled fan, you may be pulsing that capacitor with more ripple current than the fan, and it may fail prematurely according to Irms and its rating, or cause weird aliasing noises.
• Nevertheless, you are imposing a new spectrum of current interference to the commutation for the motor and thermal feedback to regulate fans may have audible aliasing winding noise, from these ideal quiet fans.

Conclusion: Use the right fan or design a DC-DC regulator to control it, that guarantees no stalling from low start voltage by setting thresholds for voltage and temperature control range.

Anecdotal experience from early 1980s to now

• Fans generate acoustic eddy currents from the blade turbulence near a fixed grill. The solution to this is make the fan inline to a plenum. The design trick is to make this as short as possible to remove all the heat outside and not add much noise.

• For SMPS forced air cooling, it is this same eddy current turbulent velocity that reduces the thermal resistance of air cooling ferrite transformers and power transistors.

  • So maximize air velocity over the hot spots, with a plenum or folded cover rather than simple panel-mounted fan that pushes or pulls a certain volume of air.
  • The secret is that the air just blows over the top of the hot spots thus ineffective or weak thermal coupling, instead of turbulent air over the part at 1 to 3 m/s. In my case design it reduced full load hotspots in a restricted space from 70 °C to 10 °C temperature rise. I verified my design with cigarette smoke, thermocouples and mylar folded plenums. (This also helped me survive UL's safety "coke-spill and sledge-hammer tests" to my 19" 1U rack design.)
  • The motor senses rotation position by a precision Hall effect pole sensors which commutates the FET bridge.
  • All electrolytic components (batteries included) have a capacitance that degrades due to a chemical reaction accelerated by current and heat that affects the ageing rate.
  • This is defined by Arrhenius' law, which is approximately -50% MTBF for every 10 °C rise above ambient.

• The coil impedance rises above stall speed so the current now increases much slower due to wind loading.

• The stall speed might be around 25% starting and 15% stopping so that if stalled with no cooling the coil heats up the magnet and that could degrade the magnet over time if above its rating.

• Back in the 1980s Toshiba, Fujitsu using some higher power muffin fans would have a thermistor with self heating by applied power and self-cooling by the fan's air velocity to detect a fan fail and create a halt alert.

• The fan not only cools the host, but the internal parts as well, except when it is stalled. For small fans well designed, no problem, but if the coils heat up the rare earth magnets unusual failures can occur (dead-spots in starting fan, etc.). This usually only happens on one of four resting positions.

• For this reason for the last 35 years, any high volume production of systems I have been involved with, an automated start-stop incoming test was performed, testing for this fault. The design/process is a critical tradeoff between efficiency near magnetic neutral and offset so the circuit knows which direction to start commutating. This slim margin exists in every PM BLDC motor. Where the fan is not dead, it's just dithering back and forth, or motionless until you nudge it.
  • Even the famous Nidec Japanese brand fans occasionally had 1% of 100 fans with dead spots, so corrective feedback to OEM eliminated their problem by forcing them to fix it.
• I also gave our distributor my design for a simple test jig design that did this, Start <= 0.1 s stop <1 s repeat for 100 cycles, 100 fans in parallel.
  • Then we had perfect yields - no failures, three shipments in a row, so then I halted those incoming tests. Problem solved.
• Often these problems occur when the factory moves or some other minor process change.

p.p.s. reply

You probably blew the internal electrolytic cap from external PWM resulting in an abrupt change in source impedance with PWM now going open/close in parallel with fan's load including a capacitor with ESR, to give a low source impedance to the bridge. I suspected the capacitor ESR should be worn out with much higher ESR and lower C, so the fan has a load high regulation error due to a much higher source impedance. This explains your dramatic losses below 100% PWM from full speed to slow speed.

The capacitor ESR must be << 10% of the coil DCR for good voltage control RPM regulation or better, so use the proper fan or a linear design. A cheap fix may be to add a series R to transfer capacitor losses to series or RL to improve maximum speed with low Q for a field fix. Or do get it right the first time and listen to the wisdom of experience.