How do 4-pin PWM headers work? and how are they implemented?

[Blood]

This nerdy question is more out of curiousitythan necessity as I still rely on old school voltage control for fan control

Question:

1) How do PWM fan headers work? There are 4 pins. Ground, 12v, sense, and pwm control. Does the pwm control pin supply the actual current to power the fan? Or does the fan still take its current from the 12v pin, but have internal electronics tha vary fan speed?

2) PWM type control can be implemented on 3-pin headers, by "pulsing" the 12v pin current. How is this implementation different?

3) If I plug in a 3-pin fan into a modern motherboard's 4-pin header. Will the motherboard still be able to control the fan?

[cmthomson]

1) How do PWM fan headers work? There are 4 pins. Ground, 12v, sense, and pwm control. Does the pwm control pin supply the actual current to power the fan? Or does the fan still take its current from the 12v pin, but have internal electronics tha vary fan speed?

2) PWM type control can be implemented on 3-pin headers, by "pulsing" the 12v pin current. How is this implementation different?

3) If I plug in a 3-pin fan into a modern motherboard's 4-pin header. Will the motherboard still be able to control the fan?

1: The latter. The PWM signal tells the fan what percentage of current to draw.

2: A 4-pin PWM header supplies constant 12V on pin 2 and a low-current control signal on pin 4.

3: On a modern board (newer than about 5 years), a 3-pin fan plugged into a 4-pin header will run at full speed always; in this situation you need an external controller, such as a FanMate. Newer boards no longer use the chip that modulated the 12V pin.

[MikeC]

3: On a modern board (newer than about 5 years), a 3-pin fan plugged into a 4-pin header will run at full speed always; in this situation you need an external controller, such as a FanMate. Newer boards no longer use the chip that modulated the 12V pin.

This is not always true. Some boards have both VR and PWM control even for headers that are 4-pin; the Asus Z77 series boards, for example. But it is probably safe to assume you probably will not have mobo control over a 3-pin fan in a 4-pin header.

[Blood]

1: The latter. The PWM signal tells the fan what percentage of current to draw.

And how does the fan regulate the amount of current to draw? Thanks for your patience with me

[Vicotnik]

PWM means pulse width modulation. The fan gets pulses of +12v, a bit like this:
+12v, nothing, nothing, +12v, nothing, nothing, etc. would be almost like giving a "normal" fan constant +4v. Many, many pulsed per second. A PWM fan has electronics inside that controls this and the motherboard talks to this electronics with the fourth wire.

edit: Found this article on the subject.
- http://www.analog.com/library/analogDia ... speed.html

[cmthomson]

Some boards have both VR and PWM control even for headers that are 4-pin; the Asus Z77 series boards, for example.

Wow. That might be the only one... And a reason to upgrade from my Asrock Z68.

[Blood]

edit: Found this article on the subject.
- http://www.analog.com/library/analogDia ... speed.html

Great read. Thanks. Now I am wondering why high frequency PWM can't be implemented on 3-pin. Research for another day.

[Blood]

Some boards have both VR and PWM control even for headers that are 4-pin; the Asus Z77 series boards, for example.

Wow. That might be the only one... And a reason to upgrade from my Asrock Z68.

If that's the only reason... May I recommend the Aquaero (what I use) or T-balancer?

[lodestar]

May I recommend the Aquaero (what I use) or T-balancer?

The T-balancer has its uses, particularly because it offers either analogue (voltage) or PWM control of 3 pin fans. Analogue control of course is subject to the usual restraints such as starting voltage. And to control 3 pin fans by PWM means using the low frequency variant of this method with the potential issues of fan clicking at low speeds and buzzing. I am not sure that the T-balancer is good value for the money, for the same price you could buy two or three PWM fans that in most motherboards would offer both more control and freedom from the noise issues. With Asus Q-Fan control for example I can get the speed of the Noctua NF-F12 PWM down to around 300 rpm and at that speed it is absolutely silent.

The Aquaero is another analogue or low frequency PWM control device, so although it offers a high degree of control for a water cooled system it is subject to the same limitations on these two methods of 3 pin fan control as the T-balancer.

[lapon]

Great read. Thanks. Now I am wondering why high frequency PWM can't be implemented on 3-pin. Research for another day.

Pulsing power like that is hard on a system. Its why we have power factor correction on power supplies, so they don't draw power in pulses but continuously. Basically if the PWM was on the 12V system pin, the whole system would suffer due to the inductance of the fan. Every time the 12V pulsed, the current surge into the fan would probably make other mobo components, and CPU unhappy. Since fans are lots of coils of wire and magnetic fields and stuff, they don't want the current to change. This is why turning on a fan is harder than keeping it running. The PWM fans probably include a capacitor as well to remove the fluctuations to the power so that fan runs smoothly, otherwise the system would see a highly variable, high frequency load, which is always a nightmare for the power supply to try to keep up with.

[cmthomson]

PWM can and has been implemented on 3-pin headers. A controller chip would shape the pulses, and a capacitor on the supply side of the chip would smooth things out for the MB power plane. But this implementation cost a couple of bucks, so it was generally available on only high-end MBs ($200+), and faded away as 4-pin PWM fans became ubiquitous.

A completely different approach was the NoiseMagic NMT-2 and NMT-3. These controllers bolt onto the corner of the fan, and convert a +12V input to a lower analog voltage based on the temperature of the air flowing through the fan. They avoid startup problems by supplying 12V for a few seconds before reducing the voltage. I have a couple of these in my system, and a few spares on the shelf.

[lodestar]

Pulsing power like that is hard on a system.

All PWM fans draw a constant 12V, there is no pulsing of the power from the PSU. An integrated circuit (IC) fan driver chip within the fan itself generates the pulses and supplies them direct to the fan motor.

[lapon]

Pulsing power like that is hard on a system.

All PWM fans draw a constant 12V, there is no pulsing of the power from the PSU. An integrated circuit (IC) fan driver chip within the fan itself generates the pulses and supplies them direct to the fan motor.

Right, but if the PWM was implemented on the 12V line (3-pin PWM), the power system of the computer has to deal with the pulsing of the current draw. As cmthomson pointed out, the good mobos that do implement 3-pin PWM it have a supply side capacitor and pulse shaping to decouple the fan from the MB power. Clever, but expensive. Not doing this would probably hurt either the motherboard or the power supply voltage and current regulators. As you say, the 4-pin PWM has no such pulsing, neatly sidestepping the entire problem, at least from the MBs point of view.

[Blood]

Thanks for the breadth of knowledge and all the replies. Just read the Fan Expert review. I wish more motherboards implement this feature (Makes me feel like I got the wrong brand).

The Aquaero is another analogue or low frequency PWM control device, so although it offers a high degree of control for a water cooled system it is subject to the same limitations on these two methods of 3 pin fan control as the T-balancer.

Just FYI for anyone looking, the Aquaero can provide a higher starting voltage, and the latest versions allows me to control fans based on CPU temp / GPU temp, hard drive temp, CPU temp - case delta. It's a rather impressive but poorly documented piece of hardware. I initially bought it due to the need to control 140mm, utter lack of PWM 140mm fans with 140mm fan mounts, and desire for a specific PCI-express layout.