The 4-pin PWM fan specification document

Control: management of fans, temp/rpm monitoring via soft/hardware

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

The 4-pin PWM fan specification document

Post by Felger Carbon » Thu Dec 14, 2006 9:24 am

114K .pdf here.

Highlights:
  • Pin 1 Gnd black
    Pin 2 12V yellow
    Pin 3 (tach) sense green
    Pin 4 (PWM) control blue
The PWM pin does not provide power to the fan. It is a 5ma maximum control signal. The specified fan lead colors are not compatible with 3-pin fan wire colors. There are three types of responses the fan can have (called A, B, C in the spec document) to the control input. The PWM frequency is 25KHz nominal, with an allowed range of 21-28KHz.

Lots of other info here. The main thing is, the fan itself has to contain some smarts. No 3-pin fan can be made into a 4-pin PWM fan by one of us ham-handed DIYers. Evidently, the 4-pin "smarts" include defaulting to full-on when no pin-4 control signal is present, which accounts for the observed behavior.

(I googled "4 pin fan wiring".) :wink:

cmthomson
Posts: 1266
Joined: Sun Oct 09, 2005 8:35 am
Location: Pleasanton, CA

Post by cmthomson » Thu Dec 14, 2006 11:57 am

Some newer motherboards modulate the 12V pin (pin 2) either under BIOS configuration control ("DC" setting) or by sensing the absense of a 4-pin connector.

I'm told that they universally put out a PWM modulated 12V signal on pin 2, rather than a reduced DC voltage.

SPCR member dinofx built a gizmo to convert 4-pin PWM to 3-pin constant DC, and documented it here:
viewtopic.php?p=307845#307845
i7 4790K CPU@4.6 GHz, ASUS Z97-PRO, 16GB G.Skill 2400C10, Intel 335 240GB SSD + WDC EFRX 1TB, Internal i7 graphics, Antec P180 case, Seasonic X-400 fanless PS, Megahalems CPU HS, Nexus 3-pin & AC PWM fans < 600 RPM, AcoustiPack foam, homemade ducts.

Felger Carbon
Posts: 2049
Joined: Thu Dec 15, 2005 11:06 am
Location: Klamath Falls, OR

Post by Felger Carbon » Thu Dec 14, 2006 12:31 pm

cmthomson wrote:Some newer motherboards modulate the 12V pin (pin 2) either under BIOS configuration control ("DC" setting) or by sensing the absense of a 4-pin connector.

I'm told that they universally put out a PWM modulated 12V signal on pin 2, rather than a reduced DC voltage.
Do I understand that 'Some newer motherboards universally put out a PWM modulated 12V signal on pin 2'? :?

cmthomson
Posts: 1266
Joined: Sun Oct 09, 2005 8:35 am
Location: Pleasanton, CA

Post by cmthomson » Mon Dec 18, 2006 8:03 pm

Felger Carbon wrote:
cmthomson wrote:Some newer motherboards modulate the 12V pin (pin 2) either under BIOS configuration control ("DC" setting) or by sensing the absense of a 4-pin connector.

I'm told that they universally put out a PWM modulated 12V signal on pin 2, rather than a reduced DC voltage.
Do I understand that 'Some newer motherboards universally put out a PWM modulated 12V signal on pin 2'? :?
Let me phrase it more precisely.

Some newer motherboards that have 4-pin CPU fan headers offer a "DC" option in the BIOS (vs a "PWM" option). Many users would assume that this would mean that selecting this option and using a 3-pin fan would result in a variable voltage being output on the header. According to several posts I've seen, some folks have used oscilloscopes on pin 2 of the header, and determined that it actually is a PWM output, not a reduced voltage DC output.

This is why dinofx's device is of interest: it converts a PWM signal into a constant reduced DC voltage, which eliminates the clicking that many otherwise quiet fans (such as the Nexus) produce when fed a PWM signal.
i7 4790K CPU@4.6 GHz, ASUS Z97-PRO, 16GB G.Skill 2400C10, Intel 335 240GB SSD + WDC EFRX 1TB, Internal i7 graphics, Antec P180 case, Seasonic X-400 fanless PS, Megahalems CPU HS, Nexus 3-pin & AC PWM fans < 600 RPM, AcoustiPack foam, homemade ducts.

Shaggy
Posts: 13
Joined: Wed Jan 03, 2007 5:44 am
Location: Spain

Post by Shaggy » Wed Jan 03, 2007 6:13 am

Hi all, i`m not an electronic expert, so according to 4-wire pwm pin features, could i use a npn transistor like this to manage the speed of a fan with this circuit?
Image
It works in 3-pin pwm headers, but i`m afraid of very low current allowed in pwm pin, on 4-wire pwm headers.

Thank for your replies.

pcy
-- Vendor --
Posts: 137
Joined: Sat Aug 13, 2005 2:46 am
Location: Central London, England

Post by pcy » Wed Jan 03, 2007 1:23 pm

Hi,


That works in principle, but in practice yoyu won't get enough amplification from a single transistor cct.


I've had one designed by an electronics engineer, and we ended up using 4 transistors to get it to work smoothly and provide some adjustement etc.



Peter

Shaggy
Posts: 13
Joined: Wed Jan 03, 2007 5:44 am
Location: Spain

Post by Shaggy » Thu Jan 04, 2007 7:10 am

Weel, ok.
pcy wrote:Hi,

I have a similar product. So far we have only built protptypes, but they work fine controlling varuous 3 pin fans using the PWM output fro an Intel DP965LT mobo.


I was thinbking of making them available for about £5 - $9.50 plus postage...



Peter
Have you tested this prototipes yet? is the output of the circuit pwm?
Are there yet on sale?

Thanks

pcy
-- Vendor --
Posts: 137
Joined: Sat Aug 13, 2005 2:46 am
Location: Central London, England

Post by pcy » Thu Jan 04, 2007 12:22 pm

Hi,


We have tested the prototypes, and we are using them on machines that I build, while we organize larger volume production.


It is a PWM to DC converter. The input to the circuit is PWM (4 pin) and the output is ordinary 3pin with the power voltage between 3V and 12V depending the PWM input. There is no point in taking a 4pin PWM in an producing a PWM output.



Peter

Shaggy
Posts: 13
Joined: Wed Jan 03, 2007 5:44 am
Location: Spain

Post by Shaggy » Thu Jan 04, 2007 2:26 pm

Ok, i`m anxious to see it.

pcy
-- Vendor --
Posts: 137
Joined: Sat Aug 13, 2005 2:46 am
Location: Central London, England

Post by pcy » Fri Jan 05, 2007 8:32 am

Hi,


You mean you want to see a pic, or you want me to send you one?



Peter

cpemma
Posts: 351
Joined: Sun May 11, 2003 2:31 pm
Location: S Yorks, OK
Contact:

Post by cpemma » Fri Jan 05, 2007 11:47 am

The way I understand it, the 4-wire fan provides a steady 12V to the fan electronics and a pulsed supply that controls power to the motor windings. That way the speed signal isn't affected by gaps in the supply, plus there are said to be other advantages in that a high PWM frequency can be used (above the audible range, giving lower PWM noise), which most standard fans (allegedly) don't like.

The Winbond chipsets with motherboard fan control produce high-frequency PWM but convert it back to a variable DC level with a simple RC or LC smoothing filter as shown on the datasheet, I assume other chipsets do a similar thing. So the 4-wire system is simpler for the mobo maker to implement and less sensitive to the power of the actual fan used.

In theory it's very easy to build a PWM controller for a 4-wire fan, I look forward to seeing some decent low-noise ones in the UK shops. ;)

Going the other way, Micrel's Application Note#34 shows an op-amp PWM-to-linear converter, but it's for a 30Hz PWM input, not the 30kHz more likely from a motherboard.

Shaggy
Posts: 13
Joined: Wed Jan 03, 2007 5:44 am
Location: Spain

Post by Shaggy » Fri Jan 05, 2007 3:18 pm

pcy wrote:Hi,


You mean you want to see a pic, or you want me to send you one?



Peter
Sorry about my bad english, i mean i`m waiting for its presentation,

pcy
-- Vendor --
Posts: 137
Joined: Sat Aug 13, 2005 2:46 am
Location: Central London, England

Post by pcy » Sun Jan 07, 2007 5:47 am

Hi Shaggy,


At the moment we only have the prototypes - built one VeroBoard rather than a custom PCB.

And of course no retail packaging - just the circuit and a cable to connect it to a 4pin PWM header on the motherboard. It's mounted on a PVC plate with a self adhesive backing.


But they work fine.

I'll probably make between 10 and 50 that way while we get production organized.



Peter

Stephen
Posts: 305
Joined: Mon Jan 02, 2006 6:34 am
Location: Michigan, USA

Post by Stephen » Sun Jan 07, 2007 8:32 am

cmthomson wrote:Some newer motherboards modulate the 12V pin (pin 2) either under BIOS configuration control ("DC" setting) or by sensing the absense of a 4-pin connector.

I'm told that they universally put out a PWM modulated 12V signal on pin 2, rather than a reduced DC voltage.

SPCR member dinofx built a gizmo to convert 4-pin PWM to 3-pin constant DC, and documented it here:
viewtopic.php?p=307845#307845
I have the Bad Axe 2 mobo, which has a 4-pin cpu fan header. The fan with my cpu heatsink is a 3-pin. So, if I hook up my 3-pin fan to the 4-pin header, will my mobo vary it's speed or run it at full speed?

blackknighti30
Posts: 6
Joined: Wed Mar 24, 2010 10:32 am
Location: Heflin, AL USA

Post by blackknighti30 » Wed Mar 24, 2010 10:40 am

Hi guys,

I found this discussion and thought you guys might be able to help. I have a Mac Pro enclosure and it uses fans made specifically for Apple, but I think I've figured out how they work. Now I just want to get them working with the temp control from the motherboard.

Here's the situation:

These fans are 4 wire fans, but the 4th wire doesn't seem to be PWM. They work in the same way, except that the control wire seems to require analog 0-5V rather than PWM. When I connect it to the motherboard header, the fan does not turn at all. I've even tried disconnecting the 4th wire in order to set it in full speed mode (PWM specs) and it still doesn't turn. I am currently controlling the speed by connecting the 5V power line from the PSU to the control wire of the fan and using a voltage divider circuit with pot to reduce the speed. The fan is at full speed with 5V, but at about 1.9-2.0V it stops.

I'm wondering if there is a way to convert either the PWM signal from the motherboard to a 0-5V analog signal or reducing the 12V pin of the motherboard header to a 0-5V signal while retaining the variable output.

Thanks in advance,

BK

lodestar
Posts: 1683
Joined: Fri Aug 05, 2005 3:29 am
Location: UK

Post by lodestar » Wed Mar 24, 2010 2:49 pm

Might be better to start that as a new topic rather than bump quite an old thread.

blackknighti30
Posts: 6
Joined: Wed Mar 24, 2010 10:32 am
Location: Heflin, AL USA

Post by blackknighti30 » Thu Mar 25, 2010 10:44 am

Yeah, I hadn't noticed how old the thread was until after I posted. I found this thread from Google.
Primary Hackintosh:
OS X 10.6.2 & Windows 7 Ultimate / Full-ATX Mac Pro case / Gigabyte GA-MA790FX-UD5P / Phenom x4 9950 @3.2GHz / 8GB Corsair Dominator PC2-8500 / ASUS EAH4870X2 2GB@ 800/950
http://aquamac.proboards.com/index.cgi?board=hack1&action=display&thread=707&page=1#4706

sthayashi
*Lifetime Patron*
Posts: 3214
Joined: Wed Nov 12, 2003 10:06 am
Location: Pittsburgh, PA

Re: The 4-pin PWM fan specification document

Post by sthayashi » Mon Jan 31, 2011 7:56 pm

Sorry to bump up an old thread, but I have a completely theoretical design for controlling a 3-pin fan with a 4-pin fan controller. It's theoretical because I've simulated it in SPICE and it seems to work there, but I've been reluctant to build it and was interested in getting some like-minded folks to weigh in on the design. I'm not entirely sure about it myself, since I'm only using what I've read on the spec sheet as a basis for the design.

I missed Dinofx's design, so I have no idea how that compares. PCY sells one, but hasn't shown off the inner workings.

Anyone interested?
[size=75][url=http://www.twolf1300.net/sthayashi/SPCR/systems.html]My Power Rig, Storage Rig, HTPC and Main Rig[/url][/size]

lodestar
Posts: 1683
Joined: Fri Aug 05, 2005 3:29 am
Location: UK

Re: The 4-pin PWM fan specification document

Post by lodestar » Tue Feb 01, 2011 2:48 am

This has been done commercially, most recently by Nanoxia with the PWMX. It was quite a neat device, and included a variable fan speed control. As it disappeared fairly quickly I assume it was not a sales success. The issue, as always, is that it may be cheaper to replace a 3 pin fan with a 4 pin PWM rather than pay for a separate controller and keep the 3 pin fan. Before the Nanoxia PWMX there was a device available from a British company called PaQ . They continue to advertise it at http://www.paqt.co.uk/store/index.php?a ... roductId=5. There is a data sheet here http://www.paqt.co.uk/docs/PaQ_PWM.pdf.

Judging from this, http://www.analog.com/library/analogdia ... speed.html it may also be better in noise terms to use 4 pin PWM fans. To pick out a sentence from this page, this is the reason why four pin PWM fans are significant in terms of quiet computing "....With PWM, the fan can be run at speeds as low as 10% of full speed, while the same fan may only run at a minimum of 50% of full speed using voltage control.....".

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: The 4-pin PWM fan specification document

Post by quest_for_silence » Tue Feb 01, 2011 3:09 am

lodestar wrote:To pick out a sentence from this page, this is the reason why four pin PWM fans are significant in terms of quiet computing "....With PWM, the fan can be run at speeds as low as 10% of full speed, while the same fan may only run at a minimum of 50% of full speed using voltage control.....".

I think it might be questionable, from a practical perspective: IME I often found out that *actual* CPU PWM mobo's ctrl were less able to dial down a fan than a "regular" voltage-controlled 3-pin one (and however I never found a PWM one which were really able to lower the speed down to 10% of rated one).
Regards,
Luca

Support SPCR, use these links when you buy: NCIX, Amazon and Newegg

lodestar
Posts: 1683
Joined: Fri Aug 05, 2005 3:29 am
Location: UK

Re: The 4-pin PWM fan specification document

Post by lodestar » Tue Feb 01, 2011 3:41 am

quest_for_silence wrote:I never found a PWM one which were really able to lower the speed down to 10% of rated one.
Yes, I agree. 10% of rated is I think a bit ambitious. Using the Scythe Slip Stream 120 PWM SY1225SL12LM-P, as fitted to the Mugen 2, as an example the rated speed is 0 to 1300 rpm, although Scythe qualify the 0 as being 200 rpm in practice. The one I have has an actual top speed of 1400 rpm (100% PWM duty cycle), and a low speed of 180 rpm (0% PWM duty cycle). So more like 14% of rated and 13% of actual, but I don't think you could get the speed down to that level with voltage control.

sthayashi
*Lifetime Patron*
Posts: 3214
Joined: Wed Nov 12, 2003 10:06 am
Location: Pittsburgh, PA

Re: The 4-pin PWM fan specification document

Post by sthayashi » Tue Feb 01, 2011 5:44 pm

lodestar wrote:This has been done commercially, most recently by Nanoxia with the PWMX. It was quite a neat device, and included a variable fan speed control. As it disappeared fairly quickly I assume it was not a sales success. The issue, as always, is that it may be cheaper to replace a 3 pin fan with a 4 pin PWM rather than pay for a separate controller and keep the 3 pin fan. Before the Nanoxia PWMX there was a device available from a British company called PaQ . They continue to advertise it at http://www.paqt.co.uk/store/index.php?a ... roductId=5. There is a data sheet here http://www.paqt.co.uk/docs/PaQ_PWM.pdf.
Yes, that's the one that PCY makes. For some reason, I never looked carefully at the picture of it. Rather I tried, but couldn't make anything meaningful from it. My design is similar, but it uses an op-amp instead of the NPN and PNP transistors, and it uses only one NPN transistor for the current amplification. Given the cost of transistors vs op-amps, the PaQ is probably cheaper to build, especially en masse.

I sorta fell into this as curiosity. I recently got a new video card, but replaced the heatsink with the Gelid Icy Vision that MikeC recommended. Unfortunately, I found the fan to be far too loud and even my regular fan controller couldn't make it quiet enough (it bottoms out to probably around 6v). I daisy chained a zalman fanmate to bring it to acceptable levels, but in all honesty, I want my video card to take care of its fan speed. The capability is already there, but the fan is not. It's not really easy to swap out the fan on this, and it generally makes me nervous to even think about.

I saw the PWMX, but also saw that it was out of stock pretty much everywhere. The PaQ seems good, but I don't know if I want to pay 10 quid plus shipping to the US for something that I can put together myself (with all due respect to PaQ). Besides, I'd need a different connector anyway for the video card, since it's not your typical header.
[size=75][url=http://www.twolf1300.net/sthayashi/SPCR/systems.html]My Power Rig, Storage Rig, HTPC and Main Rig[/url][/size]

lodestar
Posts: 1683
Joined: Fri Aug 05, 2005 3:29 am
Location: UK

Re: The 4-pin PWM fan specification document

Post by lodestar » Wed Feb 02, 2011 1:12 am

My favourite PWM control device was the Cooler Master Wind Rider, see http://www.coolermaster.com/product.php?product_id=2521 which again was not a commercial success. I think it was introduced too early and that today it would have sold somewhat better. With an adapter cable it certainly could have been used as a hub for two graphics card PWM fans. The lack of such fans on the Gelid Icy Vision is one reason why I opted for the Arctic Cooling Twin Turbo Pro instead.

SebRad
Patron of SPCR
Posts: 1121
Joined: Sun Nov 09, 2003 7:18 am
Location: UK

Re: The 4-pin PWM fan specification document

Post by SebRad » Wed Feb 02, 2011 4:35 am

Looking at the pictures doesn't look too hard to swap the fans on the Icy vision...
You can see my version of aftermarket PWM fans on aftermarket heatsink on GTX260 here.

Seb
i7 2600k under NH-C14 w/2xTY-147PWM fans, P8P67Deluxe, 8GB RAM, GTX560Ti OC w/TwinTurbo II and BIOS fanspeed mod. MX100 512GB & 2TB EARX in Scythe QuietDrive outside case. Antec Signature 650 fan swapped. CoolerMaster 201C with 2x92 + lower 92mm fan out rear, holes-in-floor and mesh top+front intakes.

lodestar
Posts: 1683
Joined: Fri Aug 05, 2005 3:29 am
Location: UK

Re: The 4-pin PWM fan specification document

Post by lodestar » Wed Feb 02, 2011 12:57 pm

The other solution for the Icy Vision would be to control the 3 pin fans from the onboard PWM socket. You would need one of those small graphics->normal size PWM socket adapters. And one of these http://www.sunbeamtech.com/PRODUCTS/Rheosmart/pci.html would avoid the need to replace the fans entirely.

Edit: It does not look too expensive either. This stockist http://www.performance-pcs.com/catalog/ ... s_id=30056 is quoting a price of $10 US plus shipping.

sthayashi
*Lifetime Patron*
Posts: 3214
Joined: Wed Nov 12, 2003 10:06 am
Location: Pittsburgh, PA

Re: The 4-pin PWM fan specification document

Post by sthayashi » Wed Feb 02, 2011 10:18 pm

For $10 + shipping, I'd be willing to check out how they do it.

Swapping fans on a VGA cooler is not a project I'm interested in getting into. I don't like using zip ties to mount fans and then I get the added bonus of dealing with 2 fans instead of one. Plus the cost of good fans and the wiring to deal with it all would be a lot more of an investment to an already expensive heatsink and video card (both of which I purchased at the same time).

With regards to PWM in general, I'm not fully convinced that a PWM fan is any more special than a regular fan with the right electronics as a separate subcircuit. After all, even the fan manufacturers have to implement the PWM spec somehow. I assume that it's electronics to modify signals and power to control a DC fan. Electronics that we can presumably recreate outside the fan.

I'd love to know how they do it though, especially on top of generating fan tach.
[size=75][url=http://www.twolf1300.net/sthayashi/SPCR/systems.html]My Power Rig, Storage Rig, HTPC and Main Rig[/url][/size]

trident
Posts: 56
Joined: Sat Jan 13, 2007 10:16 am
Location: Indiana, U.S.A.

Re: The 4-pin PWM fan specification document

Post by trident » Sun Mar 06, 2011 7:08 am

sthayashi

Would you be interested in collaborating?
Been working on a circuit with LTspice that has a P-channel FET for the output.
Would post, but ".asc" files are not currently allowed in this forum.

Regards

Post Reply