Fan speed control for 3 pin fan (4 pin MB)

imperatormk

Joined Mar 1, 2011
96
So, since I cant find some good 4pin CPU fans, Id like to build some circuit that can use the PWM signal from the MB and vary the (3 pin) fan speed (supply voltage actually) depending on temps.

Last edited:

mcasale

Joined Jul 18, 2011
210
What do you mean by "MB"?
The third pin on many DC fans is a tachometer output (a pulse). You can read it with an MPU input, but make sure it is limited to the MPU supply voltage.

mcgyvr

Joined Oct 15, 2009
5,394
Forget it..or find a 4pin. 3 pin DC brushless fans have internal hall effect sensors and supplying them with a PWM signal just screws them up causing them to stall.
Just plug in the 3 pin and forget about the variable speed...

welkin87

Joined Nov 18, 2010
32
MB = motherboard

mcgyvr

Joined Oct 15, 2009
5,394
Nope that's PWM. The PWM from the MB is just turning that transistor off and on. It will screw up the internal hall sensors and the motor will just buzz and stall. If it were easy they wouldn't have 4 pin fans.

Of course by increasing or decreasing the dc voltage to the fan will vary its speed I don't think that's really the correct way to do it either. A schematic for something like an LM317 that you could vary its output voltage (5V min to 12V max or so) would work I guess.

Last edited:

imperatormk

Joined Mar 1, 2011
96
Who knows...Is the PWM signal from the motherboard DC or it oscillates?

BTW, what is the maximum PWM signal? 5V maybe?

kubeek

Joined Sep 20, 2005
5,748
Nope that's PWM. It will screw up the internal hall sensors and the motor will just buzz and stall. If it were easy they wouldn't have 4 pin fans.
Is it that hard to filter the pwm? Anyway I run my fans on a simple 555 pwm without any filtering and they do just fine.

mcgyvr

Joined Oct 15, 2009
5,394
Is it that hard to filter the pwm? Anyway I run my fans on a simple 555 pwm without any filtering and they do just fine.
Are they DC brushless fans? IDK I know I tried this a while ago and tried adding caps/resistor to gnd to the output to "smooth" the PWM and it just wasn't working.
My post about it is around here somewhere.I'll find it
Here.. you can try the SGT's suggestions.. but I gave up and actually just kept the fans at full speed.

kubeek

Joined Sep 20, 2005
5,748
In case anyone needs it, here is the pwm and the board. The board is designed for a stripboard - it is more of a component placement than a ready pcb as you can see.

Attachments

• 27.8 KB Views: 142
• 46.8 KB Views: 89

kubeek

Joined Sep 20, 2005
5,748
Are they DC brushless fans? IDK I know I tried this a while ago and tried adding caps/resistor to gnd to the output to "smooth" the PWM and it just wasn't working.
My post about it is around here somewhere.I'll find it
Here.. you can try the SGT's suggestions.. but I gave up and actually just kept the fans at full speed.
I used a standard pc 3wire 120mm fan, like this, but should work for any other I suppose.

SgtWookie

Joined Jul 17, 2007
22,221
I'd forgotten about it, but I did make a thermistor-controlled variable power supply for the BLDC computer fans that worked quite well for an application I had.

I used an LT1171 100kHz switching controller; I'd bought several of them for dirt cheap. They're usually in the ~$5-$7 range if you buy them from a major distributor; which I figured was more than someone would want to spend on a computer fan controller.

Our local Vietnam War Museum had an older Dell computer which kept overheating; there was just not enough air cooling flowing through the case. I modified the case to accept a much more capable fan; but the fan was quite noisy at full speed; it moved a LOT of air. So, I set this regulator up to have a minimum output of around ~6v; 5v was about the minimum it needs to start reliably. The thermistor is in the outgoing air stream.

I suppose something similar could be constructed using a 555 timer, N-ch MOSFET, inductor, and some other components.

Attachments

• 117.5 KB Views: 61

SgtWookie

Joined Jul 17, 2007
22,221
Kind of thinking out loud here ... the 3-pin connector fans usually have 12v power, ground, and a speed sensor line, to indicate that they're working - but this can vary considerably depending on the manufacturer. The 4-pin fans have 12v, ground, PWM in and speed sensor out.

Here's some helpful info on that: http://www.interfacebus.com/ATX_Motherboard_Fan_Connector_Pinout.html

Thing is, if the fan function is critical (like the CPU fan), unless the mobo gets feedback from the fan that it's spinning, it will shut down in about 30 seconds to prevent CPU meltdown. This is a good thing, but is really frustrating if you don't know what's going on.

The problem here is, that if you simply "fake" a signal to the mobo that the fan is spinning, and it is NOT spinning due to dust accumulation or other fault, then you will convert a perfectly good working mobo into a slag heap.

So, you really must not lie to the mobo on this; you need some method of providing a REAL indication of the fans' actual speed.

kubeek

Joined Sep 20, 2005
5,748
Are they DC brushless fans? IDK I know I tried this a while ago and tried adding caps/resistor to gnd to the output to "smooth" the PWM and it just wasn't working.
Now that I think of it, the problem could have been with the PWM frequency being too close to that of the rotating fan, creating beat frequency and disrupting the operation.

imperatormk: if your board has 4 pins, then using the 4th pin to drive a transistor switching the fan´s pin to ground should work just fine. The voltage of the signal doesn´t matter much, just use a base resistor appropriate for the 5V possibility.

But I have no idea what will this do to the RPM sense signal, so the board might think the fan is running at a completely different speed or not running at all. If it doesn´t work, try switching the 12V supply instead of the ground connection.