Extremely easy fan controller using PWM output of a MB

[yiranhu]

Okay so I've been obsessing about system cooling for the past few days. There are lots of plans online on how you can build a PWM fan controller. Of course the designs vary from the simplest (i.e. without adjustment) to the most complex (where the speed controller is a temperature based).

Now it would seem to me that the most important component of these designs are the PWM wave generator. The actual driving circuit is as simple as a switch (MOSFET) plus a diode, which prevents MOSFET from being damaged.

The interesting thing is that if you have a motherboard that has a 4-pin fan connector (my DS3 has 2 - one of them is unused because most of case fans do not come with 4 pins), then the temperature dependent PWM wave generator is build in. There is no need to replicate that on a separate board. All you need to do is to connect the 3 relevant pins of output of the 4 pin fan port to a MOSFET switch and then connect the switch output to the 3 pin or 2 pin fan. The extra pin in the 3 pin fan can be connected to the extra pin in the 4-pin connector to allow for RPM monitoring. Then the only thing you need to do is to enable temperature based fan control in the BIOS...

Is this easy or am I just crazy? (Don't get me wrong, either way i'm still crazy).

 

[pcy]

Hi,


I don't think it's quite that easy, but still pretty easy.


I just paid an electronics engineer $100 to do me a design for a PWM->DC converter and it works just fine.


I'm planning to sell them.



Peter

 

[yiranhu]

Why would you convert the PWM signal to DC? It can be used to drive the fan directly. Converting PWM to DC essentially is an exercise in building a low-pass filter. The fan itself is actually a low pass filter, because of its nonzero inertia. So there's no reason to do an extra conversion in between, which completely defeats the purpose of PWM based control. The only reason why we need a MOSFET in between is because the PWM signal generator generally will not be able to supply the necessary current to drive the motor in the fan. Using the MOSFET allows us to draw current directly from the power supply in the same time limiting the current drawn from the PWM signal to almost zero.

I have a DC based fan controller from COMPUSA that cost 6 bucks. But it lacks the automatic temperature based control. This is why I want to look into a PWM controller that uses automatic feedback from the temperature sensors. Since the motherboard provides the PWM signal (which controls the CPU fan for example), I can just tap it, so that I don't have to replicate that part of the circuit.

I'm going to try this later today to see what kind of result I can get, which will be posted once I get it working...

 

[pcy]

Hi,

I'm converting PWM to DC because I cannot obtain 120mm and 140m PWM fans.

The PWM output of the Mobo works jsut fine, and is already temperature controlled. If you have a PWM fan there is nothing more to do - just plug it in.


I think might be missing soemthing here: if the DC fan controller has PWM input then why do you need anything else - the PWM output from te jMobo is already temperature controlled. But if the DC controller has a just a 12v in, but no adjustment then what is it for?


Peter

 

[yiranhu]

What I'm saying is that to convert a regular voltage controlled fan (i.e. 2 pins or 3 pins), all you need to do is add a transistor at the fan input. A fan (wch is just a DC motor) is inherently amenable to PWM input. The only thing that one has to do is somehow make the voltage input to the fan a switching waveform. That's not trivial since you cannot use the waveform generated to directly drive a motor. So a driver circuit must be made. The non PWM fans don't have this circuit. So we just need to make this circuit. I'm pretty sure a single transistor is good enough for this purpose.

 

[pcy]

Hi,

The cct we ended up with had 4 transistors because the PWM output was very low voltage and would not drive the transistor to control the fan directly. Also, there was the question of smoothing the G-Dc voltage (otherwise the DC fan vibrated) and provifiding an adjustment.


Thge total cct had 14 components, but you are correct, conceptyually it's just a transistor in teh 12v cct controlled by the PWM.



Peter

 

[TBSN]

Are the fans on CPU coolers 4-pin (temperature controlled)?

 

[pcy]

Hi,


If its a 4pin header on teh mobo , like a 3pin but with one extra pin than yes, teh only point of the 4th (PWM) pin would be to provide fan speed control, and I don't see what else it could be except temperatue controlled.


Peter

 

[cheesehead]

You know, a good old-fashioned 555 PWM controller would do the exact thing just as well. (You also could have saved $100 on the design.)

http://www.cpemma.co.uk/555pwm.html

A $2 FET will allow you to drive roughly 30 fans - that's a rough guess, but who actually uses 30 fans? As an added bonus, 555 circuts are often easily tweaked for things like temperature control.

Or, a PicMicro could be used - for a few bucks, you can program the thing for any sort of behavior you want.

I'm part of a robotics club with nearly zero money. If anyone's interested, we could cook up a kit, or sell assembled modules as a fundraiser.

 

[Operandi]

Originally posted by: yiranhu
you cannot use the waveform generated to directly drive a motor. So a driver circuit must be made. The non PWM fans don't have this circuit. So we just need to make this circuit. I'm pretty sure a single transistor is good enough for this purpose.

Are you sure? Many, many pre socket 775 and AM2 boards use PWM to reduce the speed of traditional 3-pin fans.

 

[VinDSL]

Sorry for being rudimentary, but what's the point of a this PWM stuff? Fan noise?

Here's the way I see it...

Many of you have grown up with computers keeping you company in your bedroom, e.g. your domain. Mom, and especially Dad have been absent in your life. You didn't like the reality of the situation, and formed your own reality online. And, you've always kept your 'Teddy Bear' close at hand, 24/7/365.25

The thing is... now you're growing up, and you're having a hard time sleeping with your partner, 'cause their fans are snoring too loud, right?

Blah, blah, blah... you know where I'm going with this!

Here's my advice: If you sleep in the same room as your 'Virtual Angel', turn it off at night!

PWM is a stupid idea for stupid ppl. Don't fall for it! Turn the damn thing off at night...

 

[Aluvus]

Originally posted by: VinDSL
Sorry for being rudimentary, but what's the point of a this PWM stuff? Fan noise?

Here's the way I see it...

Many of you have grown up with computers keeping you company in your bedroom, e.g. your domain. Mom, and especially Dad have been absent in your life. You didn't like the reality of the situation, and formed your own reality online. And, you've always kept your 'Teddy Bear' close at hand, 24/7/365.25

The thing is... now you're growing up, and you're having a hard time sleeping with your partner, 'cause their fans are snoring too loud, right?

Blah, blah, blah... you know where I'm going with this!

Here's my advice: If you sleep in the same room as your 'Virtual Angel', turn it off at night!

PWM is a stupid idea for stupid ppl. Don't fall for it! Turn the damn thing off at night...

You do realize that a lot of people keep their computers in a room they spend many of their waking hours in, right? My computer is in the living room, and I know my roommates appreciate it not making an ungodly racket when on.

Just because noise isn't a priority for you, doesn't mean others don't have legitimate reasons to care. In other words, your pop-psychology theories are crap.

PWM is a very good way to control the speed of fans and other devices. It's not hard to implement, it's effective, and it allows pretty precise control.

 

[VinDSL]

Originally posted by: Aluvus

Originally posted by: VinDSL
My computer is in the living room, and I know my roommates appreciate it not making an ungodly racket when on...

Define 'ungodly racket', then we'll go from there...

 

[thecoolnessrune]

I have two 34db fans... Doesn't bother me at all.

 

[thecoolnessrune]

Double post, my apologies

 

[pcy]

Hi,


Yes... I did see several designs like that when I was doing my initail resaearc:

Originally posted by: Cheesehead
You know, a good old-fashioned 555 PWM controller would do the exact thing just as well. (You also could have saved $100 on the design.)

http://www.cpemma.co.uk/555pwm.html

The problem is it does not do what is required:

You have a PWM output from the Mobo, and a 3 pin fan, so you need a PWM to DC converter. Nobody makes one that I could find...


Which is why I'm doing my own.






Peter

 

[pcy]

Hi,

Originally posted by: VinDSL
Sorry for being rudimentary, but what's the point of a this PWM stuff? Fan noise?

It does not matter wjhat the point is, if any.

What matters is that teh mobos now come with PWM fan headers so we either have to run the CPU cooling at full blast the whole time or we have to find a way to get the PWM header to control an ordinary 3pin fan.





Peter

 

[TBSN]

Um, what does a 4-pin fan do that the 3-pin ones cannot? Is it possible to get 4-pin 120mm fans? What is PWM? Thanks for any info...

 

[VinDSL]

Originally posted by: pcy
What matters is that teh mobos now come with PWM fan headers so we either have to run the CPU cooling at full blast the whole time or we have to find a way to get the PWM header to control an ordinary 3pin fan...

How so? Who says 'we' have to follow Intel PWM FSC specifications?

That's what fan controllers are all about...

Integrated PWM is for stupid, lazy, sheeple, IMHO!

Please, change my mind on this...

 

[pcy]

Hi,

We don't have to follow the Intel PWM spec.


However:


If we have an Intel mobo with a temperature cotrolled PWM CPU fan header, then most of us will fnd it convenient to use a PWM fan with it.

The cct we are talking about converts a 3pin fan into a PWM fan. Particularly in the larger, quieter sizes PWM fans are hard to come by.


I have this cct, and it's driving a 140mm fan cooling a C2D E6600. It'd been running Prime95 (two instances, one per CPU) now for 24 hours. The CPU is stting rock solid at 60C with the fan running at 575 rpm.

Seriously quiet....



Peter

 

[TBSN]

quick question: you can change the speed of the CPU fan on the motherboard manually (only 1 speed) right? Or does it always go on full blast? I am about to build a computer and I was under the impression that with a 3pin cpu fan you can at least change the speed manually within the BIOS or something.... Plz help me out, Im confused.

 

[yiranhu]

Several things:

1. Why not use 555 circuits? Well main function of the 555 chip (or PIC) is to generate a PWM waveform that is used to drive a MOSFET. That waveform is already being generated if you have a spare 4 pin fan header on your motherboard. There's no reason to use the 555 chip remake something that's already there. Just put in a transistor and bam you're done. Of course if you want to, you can build the entire circuit with 555 chip, and that's just dandy.

2. What is PWM? It is pulse width modulation. Your fan is a DC motor. To run a DC motor, you give it a DC voltage and it goes. The speed of the motor is practically a linear function of the voltage applied. Therefore lower voltage = slower fan = less noise. So voltage is a good method for controlling the speed of a motor. Unfortunately, a DC motor has inertia and friction, which makes it a nonlinear system. So below some voltage threshold, the motor will not rotate at all because the input power cannot overcome the inertia and friction. A more robust method is PWM. Instead of giving the motor a lower voltage, it gives the motor an alternative pulses between 0 and 12 volts. I.E. we're turning the motor on then off very rapidly. Because the motor has inertia, it cannot react instantaneously to each changing pulse. Instead, what it sees is the average value of the pulse. For example, if I give 12 volts 50% of the time and 0 volts 50% of the time, the motor effectively sees 6 volts at its input. However because the actual on input is never below 12 volts, the motor will spin at very low duty cycle. Therefore your controller can reduce the speed even further than a voltage controller could. The problem is that when you give a system a on/off signal (which according to Fourier transform can be decomposed into sinusoids at the harmonics of this signal), your output will also have a small component resembling this input. In a fan, this could result in a weird sounds, if the basic frequency of your input is still in the audible range. A well designed PWM controller will take this into consideration and minimize the result of this.

3. what is ungodly racket? For me that's the noise when my case fans (3) runs at full speed. For VinDSL, apparently it's when a nuclear weapon exploded in front of his house. For Billy Bob Joe, it's when a goat squeals because a pig is flying... get the point?

4. Who in here said intel PWM spec before VinDSL quoted someone saying that? i don't see it... All I see is people having a discussion on how to build fan controller to better control noise of the fans. PWM motor control has existed for ages. Find an electrical engineering control's textbook and you will see it. BTW, if we're stupid, lazy, and sheeps, we would be ordering fan controller online instead of coming up with ingenious ways of using available resources to build one. IMHO!

 

[yiranhu]

Originally posted by: TBSN
quick question: you can change the speed of the CPU fan on the motherboard manually (only 1 speed) right? Or does it always go on full blast? I am about to build a computer and I was under the impression that with a 3pin cpu fan you can at least change the speed manually within the BIOS or something.... Plz help me out, Im confused.

Depends on your MOBO. If you have AMD (don't quote me on this), there's good chance that the fan connector on your MOBO is a 3-pin connector. In which case, if you plug your 3 pin fans into that and has fan control enabled in the BIOS, your computer will control fan speed automatically.

If you have a 4 pin connector, then what you can do is
1. connect the +12 V pin in 3 pin connector to the +12 V pin in the 4 pin connector
2. connect the speed sensor pin in your 3 pin connector to the sensor pin in the 4 pin connector
3. connect ground in 3 pin connector to ground in 4 pin connector

that will run your fan. Then go to bios and change fan control method to voltage based control (assuming this option is available). Then your fan will change speed based on temperature. If you don't have the voltage based control option, then your fan will just run full speed.

 

[Operandi]

Originally posted by: TBSN
quick question: you can change the speed of the CPU fan on the motherboard manually (only 1 speed) right? Or does it always go on full blast? I am about to build a computer and I was under the impression that with a 3pin cpu fan you can at least change the speed manually within the BIOS or something.... Plz help me out, Im confused.

AMD AM2 and Intel 775 both use 4 pin PWM fan headers for the CPU so if you go with either of those platforms and use a 4 pin PWM fan you sure to get automatic fan control for at least the CPU fan.

 

[VinDSL]

Originally posted by: yiranhu
3. what is ungodly racket? For me that's the noise when my case fans (3) runs at full speed.

Why don't you just undervolt them in a conventional manner?

Originally posted by: yiranhu
4. Who in here said intel PWM spec before VinDSL quoted someone saying that?

Um... See your title... reciprocal relationship, yes?

Actually, I think I see the original intent of this thread now...

I was working on an inverter at work yesterday. It supplies current to a conveyor motor (shaftless design). The inverter was working just fine, but it wasn't getting input from the encoder, so I threw a scope on it (spectrum analyzer actually). Nothing was coming out of the encoder except 60Hz background noise. I was looking for a square wave from the encoder... blah, blah, blah...

That's when it struck me!

I think you're saying... why not use the output of the spare mobo PWM header, as an input signal, to trigger a circuit that would control the speed of ALL fans in your computer, not just the HSF, e.g. using the mobo PWM output to automagically adjust the fan controller settings for your case fans, et cetera.

Is that 'the point' of this thread?

If so, I'll redact my statement about PWM being for lazy, stupid sheeple. That would be a great idea!