I want to use a 3-wire fan (e.g. this one), which will be operated closed-loop, powered by a variable DC voltage 0-12 V with tachometer feedback. When requesting a rather low RPM of the fan, the DC supply voltage (\$V_+-V_-\$) will be probably pretty low (well below 5 V).


simulate this circuit – Schematic created using CircuitLab

As the tachometer signal relies on the Hall sensor logic to generate a valid output state, is there a minimum instantaneous DC supply voltage below which the tachometer signal can't be generated despite a spinning fan? I had no luck finding this information in a datasheet but maybe some of you have experience in this regard?

As we know when a 3-wire fan is powered via PWM, the tach signal is invalid during phases when the input power is off (i.e. 0 VDC). The tach signal is valid during phases when the input power is 12 V.

Logically, this suggests that there is a minimum voltage below which the tach signal generation stops despite a running fan.

Does the tach signal generation have a minimum operation voltage?

I did a test and results are a little inconclusive:

The fan for the test was a Lian Li LI121225SL-B4-C.

When the supply drops below about 3.8 V, the fan stops and the tach-out is pulled active-low (the output is of open-collector type). If the fan is spun by hand, the tach remains active-low. The tach-low output is maintained down to a supply voltage of about 0.8 V. Below that threshold, the output goes into an open state.

However, this fan has a stall detection so it is likely that it realizes that it stalled and defaults to output a tach-low until it manages to re-start on its own.

It appears that the correct output logic state is thus maintained down to about 0.8 V supply voltage. Is this a realistic assumption for fans with tachometer output logic ?

  • \$\begingroup\$ Which fan specifically? Many have onboard drivers, so won't like being PWM'd. And what type of tach output, open collector? Totem-pole? If the datasheet doesn't say, you'll have to get one and try it. There are literally tens of thousands of fan variants, all of them slightly different. Datasheets tend to be skimpy on details also. \$\endgroup\$
    – rdtsc
    Sep 30, 2021 at 12:01
  • \$\begingroup\$ @rdtsc ok looks like I need to do more reading.. I thought that all 3-wire fans use 2 pins to power the fan directly and 1 wire to output a tach signal (open-collector). Shopping for such fans at least didn't make it obvious to me that this was false. \$\endgroup\$
    – tobalt
    Sep 30, 2021 at 12:15
  • \$\begingroup\$ A 3-wire fan does use two pins for power and one for tach. But whether that power is PWM-capable, and whether the tach is even an output or input, depends on the exact fan. Also consider that, for a fan which refuses to run on PWM power, a suitable R-L-C lowpass filter should be able to smooth out the PWM enough for it to run. \$\endgroup\$
    – rdtsc
    Sep 30, 2021 at 12:30
  • \$\begingroup\$ @rdtsc I don't want to PWM the fan, as stated in the first sentence of my question. It will be run via variable DC (for low EMI), I added a schematic. But if you are right and some fans use some blackbox internal drive circuitry, even a variable DC voltage is not going to work for all of them. :( \$\endgroup\$
    – tobalt
    Sep 30, 2021 at 13:00
  • 1
    \$\begingroup\$ I would suggest that most if not all fans of the type described have brushless DC motors controlled by proprietary controlled schemes. \$\endgroup\$
    – user80875
    Sep 30, 2021 at 13:31

1 Answer 1


Your biggest problem here is the tach output is not referencing ground. The fan and tach signal are referencing 12 volts. You are varying the negative fan terminal, floating it above ground to control the fan speed. What could you possibly do with a varying tach signal that floats above ground?

If you reference ground and control the + input voltage, you can pull up the tach signal to whatever logic level you are using, 3 volts or 5 volts for example. You do not have to pull up to the fan Supply voltage. This is a super common assumption I see all over the place where people try to compensate for not doing this correctly in the first place.

  • \$\begingroup\$ It's an old question, and I ended up using a PMOSFET and op-amp to control the V+ indeed. However, even as drawn, the V+-referred signal would be no issue. A single PNP on the output inverts it back to a GND referred signal if desired. Accepting anyway, to close the question. \$\endgroup\$
    – tobalt
    Oct 11, 2023 at 18:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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