Intel fan humming when powered through resistance

Schematic:

Yikes!

How do I modify the circuit to stop the humming? All I know that pins 3 & 4 on my fan aren't used.

Based on the schematic, I believe this fan uses a brushless motor, with an internal controlling circuit which handles the commutation between the phases of the motor. It's very probably _not_ designed to be driven by lower-than-expected voltages as a form of speed control. I'd guess that the control circuit is switching the drive on and off in a way (or at a frequency) which causes magnetostriction noise or direct vibration.

The proper (expected) way to control the speed of fans like this is:

(1) Supply them with the expected DC supply voltage. Don't vary this.

(2) Drive the "PWM" pin with a pulse-width-modulated control voltage, with a duty cycle corresponding to the proportional rotation speed you want.

There are quite a few ways to create the PWM signal to reduce the speed. A 555 (or CMOS-equivalent) timer chip would be one. A simple

Based on the fact that the fan is willing to run with the PWM pin disconnected, I'd guess that this pin has an internal pull-up resistor. Driving the pin through an open-collector (pulldown-only) circuit would work. Driving it directly from a microcontroller... better check the voltage on PWM when it's floating, and see whether it's being pulled up to a higher voltage than the micro will tolerate.

You can either pick a PWM duty cycle (experimentally) which gives you the speed you want (and then run the system open-loop), or you can have the microcontroller monitor the pulse-rate coming out of the TACH pin, and adjust the PWM duty cycle "on the fly" to keep the fan at the speed you want (closed-loop control).

How do I control the fan (with PWM) with a SP3T (Low/Off/High) switch instead of a potentiometer?

BLDCs usually do speed control nicely with DC voltage. We do that in temperature tracking loops. But his circuit is a (poorly defined) current source. BLDCs draw current spikes.

Could the high pitched humming be cause by circuit oscillation?

try it with just a resistor, that transistor circuit looks like it can behave unpredictably.

if you want a more predictable transistor circuit connect a potentiometer across the B-E terminals of the transistor and connect the wiper to the base.

potentiometers with total resistance between 1K and 20K should work fine here.

I'd start by reading this:

I'd consult the following information and schematic to learn how the fan is intended to be used:

But if you want to control the speed then pin 4 has to be used as described.

Note also the requirement for 12V +/- 1.2V

You start by designing or purchasing a controller module like the one shown on this page:

Here is a junk-box-special cheap and dirty variable PWM maker:

piglet

I needed to kill some time while waiting for a very important package. Here is your PWM circuit in LTspice. Very interesting effects as you change the duty cycle.

The frequency changes (a lot), and the behaviour at the duty cycle limits are interesting.

I am giving up on Google Drive. File sharing is too painful. TinyUpload is still broken, so sabercat is about the only option left. It may make you wait 15 seconds before downloading, but maybe google drive is just as slow, depending on your browser.

I can't open it. Please send a JPG, PNG, or SVG.

Your circuit is not "a resistance".

A single real resistor should work, with a pretty hunky bypass cap. These BLDC fans average out to fairly ohmic loads.

Still, with a resistor it hums.

We control similar fans with DC drive, and they don't hum.

Connect the fan to a variable bench DC supply and play with it. I do that now and then, and the fans get quieter at lower voltage. Use two or three flashlight batteries if you don't have a power supply.

A pretty big electrolytic cap across the fan, after the resistor, may tame it.

I don't have a DC power bench, but I have a 12v power adapter.

Do you have a name?