I'm looking at building a 4-wire PC fan PWM circuit. The goal of such a circuit is to send a PWM control signal to a 4 wire PC fan, so that the fan can modulate it's speed (NOT to pulse the 12V supply).

One circut for this that is widely cited is the 556 based circuit located far down on this page. The circuit looks like this:

enter image description here

I can't for the life of me see how this circuit can work. My understanding of the goal such a circuit is (per the specs here) to produce a stream of pulses at 25kHz, and to vary the width of each of those pulses.

This circuit has the pot tied into Trig1, so as far as I can tell it'll vary both the width of the pulses as well as the frequency of the pulses? Or do I just not understand it (won't be a shock).

I would have assumed that one should build an astable 555 triggering at 25kHz, and then feed that into a monostable 555 that varies the pulse width?

Side note: If this circuit isn't the best approach for a fan PWM, can anyone recommend a better one?

  • \$\begingroup\$ You can get dedicated fan controller chips. Just search for "pwm fan controller chip". \$\endgroup\$ – Andrew Morton Mar 1 '17 at 19:24
  • \$\begingroup\$ Or use for instance an ATTINY85. It would even read the RPM for you at the same time, in order to detect a stalling fan and ramp up the pulse width if it happens. \$\endgroup\$ – Dampmaskin Mar 1 '17 at 20:23
  • \$\begingroup\$ Thanks! Unfortunately I don't have a programmer. I seem to want to do a microprocessor project once every 3 years, so I'm not sure I want to buy one. \$\endgroup\$ – RobotAndy Mar 3 '17 at 12:18
  • \$\begingroup\$ Andrew Morton: Thanks! I hadn't discovered those. Most of them seem to want SMBus, which I won't have. I did find the MAX31740, which is interesting - but it wants an external thermistor, and I want a pot - but I suppose I could figure that out. The biggest issue is it's surface mount and I think it's unlikely I can deal with that... \$\endgroup\$ – RobotAndy Mar 3 '17 at 12:36

One of the 555 timers in the 556 is used as a typical PWM generator. The second 555 is used to buffer the output according to that article. You may only need the single 555 circuit detailed in the first part of the article.

The effect of the pot is to vary duty cycle from about 5%-95%. The frequency will be based on the resistor, pot, and cap values, and held fairly constant (although there will be some slight frequency variation as the duty cycle changes). The diodes types (forward drop) will affect frequency also. This article may explain the operation more clearly.

You could set up two 555s (or a 556) as you described to maintain frequency if needed. I think a fan would work fine with the frequency changing over the range produced by the circuit as shown.

There are more elegant solutions to this (a microcontroller would probably be best), but 555 is most likely the simplest solution you would fine, and probably more than sufficient for the application.

| improve this answer | |
  • \$\begingroup\$ Hmm - that link was helpful. I didn't realize you could discharge through the output. So now I'm wondering if the frequency would vary afterall? You're always charging/discharging through the pot+R1, so the pot is just shifting part of the timecycle between charge/discharge. I think with R1=1K, P1=100k, C1=680nF you get a constant 21kHz? \$\endgroup\$ – RobotAndy Mar 3 '17 at 12:58
  • \$\begingroup\$ You're right it should be fairly constant, but I'm not sure about your values. I think your cap needs to be much smaller. Maybe you meant pF? \$\endgroup\$ – AngeloQ Mar 3 '17 at 14:07
  • 1
    \$\begingroup\$ I made this simulation for you tinyurl.com/htopxae \$\endgroup\$ – AngeloQ Mar 3 '17 at 14:46
  • \$\begingroup\$ Thanks - now I'm curious why it runs at 15kHz instead of the 21kHz I'd anticipate - but I'll ask that in another question. Your answer still says that the frequency will vary, which I think is incorrect, if you're willing to fix that I'd be happy to mark your answer as correct! \$\endgroup\$ – RobotAndy Mar 4 '17 at 15:59
  • \$\begingroup\$ The frequency is not exactly constant, but very close. I think the variation has to do with the diode voltage drops. I'll edit the answer to reflect that. By the way the difference in frequency from the calculated expected value is also due mainly to the diodes. You can change the diode drop in the simulation and see the effect on frequency. \$\endgroup\$ – AngeloQ Mar 4 '17 at 17:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.