Controlling 3-Pin Fans (or Water Pump) using 4-Pin PWM Control from MoBo

[P!nkpanther]

This topic has already been discussed here in this thread:
https://www.techpowerup.com/forums/threads/so-you-want-pwm-control-of-your-new-cpu-fan.107135
and this one:
https://www.techpowerup.com/forums/threads/so-you-want-pwm-control-of-your-3-pin-fan.115752

Since I didn't want to bury my solution deep inside an old thread, I decided to open a new thread for it.

My recently acquired MoBo lacking enough fan outputs but my (also recently acquired) case having plenty of 3-Pin fans, I had to find a solution for this. I found the other threads above, but the solutions presented there were insufficient or flawed in some way or another (e.g. low-side switch instead of high-side: Bad for the tacho signal!).
As an experienced analog designer I had the knowledge and the tools to develop and test a proper solution, so I got my ass working...

Concept
The control signal comes from the MoBo 4-Pin connector but since you can't put a too heavy load current on these, I wanted the bulk of the power to be drawn directly from the PSU, which means from 12V pins of the Molex connectors.
Also I wanted my circuit to be able to supply at least 2 Amperes to the fans.
Since I had 3 (identical) fans to supply but only 1 fan connector on the MoBo for feedback of the tacho signal, I had to designate 1 fan as Main Fan, where the feedback tacho signal was to be taken from. The rest of the fans simply get the same voltage as the main fan.

Bottom right the 4-Pin connector to the MoBo
The 3-pin connector which goes to the main fan (the one from where I feed the tacho back to the MoBo) can be seen on the side of the PCB
The black-red cable is the Molex pair from the PSU
The black-yellow cable is the Molex to the fans without tacho feedback

Circuit
I have extensive simulator knowledge so I designed both circuits using an phased-out version of PSpice I have at home.
The main advantage is that you can not only see voltages but also currents and power inside the components, see simulation pictures below.

I came up with 2 variants of my circuit: Bipolar and MOSFET
The bipolar variant is designed for an output current of up to 2A while the MOSFET variant goes up to 4A, under the same thermal circumstances. It complies with the latest revision of the 4-Wire PWM Fan spec http://www.formfactors.org/developer/specs/4_wire_pwm_spec.pdf

For my tests I built up the bipolar variant, since I had the parts already lying around in my home, and did the tests successfully.
I haven't tested the MOSFET variant, but am quite confident that the results in practice will match the simulation as well as they did with my bipolar variant.

So this is the MOSFET PWM 3-Pin Fan Driver Schematic:

Red framed items are auxiliary parts/signals for simulation
Blue framed items are connectors
Q1/Q2/Q3 are general purpose transistors (300mW/100mA)
M1 is a P-Channel MOSFET IRF9540. Should not be too hard to get. Make sure to use this exact model or else try at least to match RDSon and gate capacitances. Also provide cooling to dissipate ~2W at 4A (see photo)
Capacitor C1 is optional but highly recommended. Watch for correct polarization.

Simulation:

Top waveform is the PWM controller signal from the MoBo
Middle waveform shows the 4A PWM current achievable with the MOSFET version
Bottom waveform is the power inside the MOSFET. It's around 1.8W during the pulses with a ~7W power spike at the edges.

Bipolar PWM 3-Pin Fan Driver Schematic

Again:
Red framed items are auxiliary parts/signals for simulation
Blue framed items are connectors
Q1/Q2 are general purpose transistors (300mW/100mA)
Q3 is an NPN power transistor BD241 which needs to be cooled sufficiently to dissipate ~2W at 2A (see photo)
Capacitor C1 is optional but highly recommended. Watch for correct polarization.

Simulation:

The bipolar version is quite slower than the MOSFET one.
Only ~2A PWM current is achievable under the assumption of the same power dissipation as the MOSFET version.
Bottom waveform is the power inside the MOSFET. Around 1.5W during the pulses but with quite a long ~6W power spike at the falling edge, so over all ~1.8W as well.

This solution has proved perfect for me and I hope it will be useful for someone else too.

 

[John Naylor]

Here's another solution .... I have about 6 of these, 3 in one box 1 for 6 rad fans on 420 rad, another for 4 fans on 280 rad and last feeding 6 case fans

https://www.newegg.com/Product/Product.aspx?Item=9SIA4RE5032993

 

[jboydgolfer]

I'm so glad this functionality is built into my motherboard, on my four pin headers ,in my bios, I have an option to switch between constant voltage, WPump, or case fan

 

[ZenZimZaliben]

I will give you props for the complete homebrew solution and the skill it took to build this out. Very Cool! But as John Naylor pointed out there are cheap and easy to purchase solutions.

 

[P!nkpanther]

I'm so glad this functionality is built into my motherboard, on my four pin headers ,in my bios, I have an option to switch between constant voltage, WPump, or case fan

My mobo has that same feature, but the point here is, that most mobos lack the power to fuel 3 or more fans.
My mobo only is capable of supplying 1 Amp. That's where my solution comes in, which draws the power necessary directly from the PSU, not via the mobo. This allows for 2 Amps or 4 Amps, depending on the version.

Here's another solution .... I have about 6 of these, 3 in one box 1 for 6 rad fans on 420 rad, another for 4 fans on 280 rad and last feeding 6 case fans

https://www.newegg.com/Product/Product.aspx?Item=9SIA4RE5032993

Yep, this seems to do exactly the same job. It takes the 12V line coming from the PSU, modulates it by PWM, controlled by the 4-pin PWM signal and distributes it to the 3-pin outputs.
Wasn't aware of that product, looks cool, thanks for pointing it out!