Guide Intel 4 pin to std 3 pin fan converter

I needed to convert a 4 pin fan header to 3 pin. I couldn't find such a converter on the internet, apart for one which was for sale. So I designed my own. Part choice came purely from what was at the top of my parts bin, so no doubt someone could offer more optimal component choices.

I checked out the standards for 4 pin mobo connections. The important things to note are.

The mobo header works in open drain on pin 4 to provide the PWM signal to the fan. The mobo's are capable of draining 5mv of current minimum.. That is why I used a 1k pullup resistor. I reckon i could have used higher no probs.

The fan itself requires a 3-5v pullup, ie the mobo pwm controller does not supply voltages. Hence the reason for the LM78L05 voltage regulator.

The circuit above uses a BFY51 medium power transistor. The 12v fan I have tried it on is rated at 0.3A. With an Hfe of 120 on my multimeter, and 5 mv of current at the base, this circuit should be good for supplying two fans at a push. I have several BFY51's that have higher HFe and so could comfortably supply two 120mm fans.

Why that particular transistor? Well its NPN. Can handle 800mv of current no problem, and most importantly I had some in my parts bin that I had no other use for.

The rest of the circuit is easy. As a motor (fan) is an inductive load, it does not require a current limiting resistor. The only other part of importance is the 4n4001 diode between the +/- pins of the 3 pin fan. The motor is an inductive load, so when swithing off produces voltage spikes. The diode is there to protect against them.


Advantages:

The intel header on the cheapo motherboard I needed this for is capable of smoother pwm control of the fan. I can control a 12cm (rebranded Yate Loon) 1300rpm fan down to a mere 77rpm with this circuit.

Many aftermarket heatsinks come with only 3 pin plugs.

Its cheap and simple.

If you choose a transistor with high enough hfe and current capability, you should be able to run two fans off the same header with this..

The PWM from the 4 pin header seems smoother and quieter.


Disadvantages...

Speedfan cannot measure the RPM through the sense wire of the fan very well at all with this circuit. Spurious readings occur frequently when PWM is applied.

Its obviously not as easy as just plugging in to a 3 pin header. Though parts wise the costs are minimal.

Piccies:

 

Interesting but going by my old knowledge,
you need PWM to fan + (red)
gnd to gnd (black)
sense to sense (yellow)
But as I said I'm going by old memories.

 

Hi. What you describe is the header out for a 3 pin fan. What this circuit does is take output from a 4 pin fan, you know the ones that are popular on intel boards and convert it to work on 3 pin fans.

On the Intel board you have (and not in order)

12v
Gnd
PWM
Sense

unlike the 3 pin header, the intel 4 pin equivelant gives a constant 12v power supply to the fan. The PWM line is 3.3-5v, and gives the control signal to the fan. It does this not by supplying voltage but by sinking it to produce a signal. The 4 pin fan contains electronics in the housing that uses this control signal to vary the 12v voltage through the fan motor.

The sense line works more or less the same on both 3 pin and 4 pin headers.

In my circuit above, the transistor uses the pwm signal to rapidly switch on and off up the Gnd from the fan motor. It doesn't matter to the operation of the fan wether the 12v line or the Gnd is switched, the result is the same. I can't remember why I chose the Gnd for switching, probably because I guessed that the 12v line would be used for the sense line to work. On reflection, I dont think the sense line will work either way round that well with this circuit. You could place the transistor on either side of the motor.

The circuit works. I have been using it now for several months. If my schematic is wrong (and I dont think it is) its cause I built the circuit first and then made the schematic.

 

Hi i agree with AJB2K3

this is a better design

4pin on MB | 3pin fan
12v | Output of PWM and 12v CCT
GND | GND
Sense | Sense
PWM | -

but your design does this

4pin on MB | 3pin fan
12v | 12v
GND | Output of PWM and GND CCT
Sense | Sense
PWM | -

you curcit works but GND should stay GND its good practice

I made a simlar fan thig that used a TC64X PWM fan controler from Microchip, i got it to turn on the fan at a set temp and then ramp up the fan speed when things got hotter, this was for an old pc without these 4pin fans.

anway in the end when i got the thing working it modifyed the GND voltaged to control the output of the fan.

Credit to you for your work to

 

This shouldn't be, but is, beyond my A-Level Physics (Electronics) Knowledge..

 

Thanks for your interest.

Your right perhaps on maintaining GND as good practice, though in this example the consequences of a short would be similar either way.

The circuit is easy to modify so that it is the 12v line that gets chopped up rather than the GND line. As I mentioned two posts ago, I chose to do it on the GND line because I was guessing at the workings of the sense line within a 4 pin fan's electronics. I could find technical papers describing the standards for 4 pin headers but I couldn't find documentation on the sense lines. I would be interested if anyone could post a circuit. Big question for me really is does the sense line pull high or low. If I were building again I would probably use the transistor to chop up the 12v line. The speeds I run fans at, the sense line probably wouldn't work either way.

I had a look at the TC64X fan controller you mentioned. It looks interesting, but the datasheet gives what seems a fairly complex circuit to operate it. Was yours based on this, or can it be simplified? I can see where this would be usefull in controlling case ambient temperature... Nice that it can work autonomously. I build amplifiers and when I get round to making a big amp for a subwoofer, this might be worthwhile including.

I took a different approach for case temperature control a few years ago. I used a maxim 1668 5 channel sensor connected to the SMB bus. The Speedfan application read from the Maxim chip and could switch fans local to the hot area. In the end I got the whole PC to cool passively, and the fans were only a failsafe which fortunately was never required.

 

I think I found you fan tach answer here

Sorry, you'll need to open the spec for the diagram

This seems to be in line with the random info I found from some maxim and nidec datasheets.

The last page of this speedfan article is pretty interesting too.

AFAIK, the PWM signal has it's roots in the SSI CEB spec which states:

so I don't know how that would affect the tach signal.

 

here is the CCT i used it chops the GND rarther than the 12V



i modifyed it from a bittech project on another similar fan controler chip

the hardest part of the project was solderingt the SM TC646 chip to the vero bord

 

Thanks for the excellent post!
I've made this circuit and put it in between the fan and the header on my graphics card.
The card is a Sapphire 4850 Toxic, which has a large copper heatsink on it with a 3pin fan. That means the temps are really low, but the fan always runs, ignoring the ATI reference design PWM circuitry. Now it's quiet at idle and cool under load

 

Glad you found it useful clown. I had not thought of that one... I will be writing this one up for my website at some point soon. I didn't realise that there was another application for it....

My website got onto the Slashdot front page, So I need to negotiate with the hosting company who are not happy that I "affected the performance of their other customer's sites" They have shut me down for the moment.

 

5V

Hi,
Nice post, it's very usefull.
What do you think about using the 5v of the power supply instead of "creating" it with a regulator ? Moreover, there is a very low current throw the 5v-regulator isn't there ? So couldn't we put 7v zener (or zener combination to set the point to 5v) ?
I hope the schematics can be simplify, what do you think ?

Sylvain

PS: sorry for my (probably) bad english level

 

Yes you could easily use the 5v line. I used a regulator in my circuit so that I could run the whole circuit from the 4 pin header. I wanted my cabling to be as neat as possible cable management wise. A zener is also a possibilty. At the time I put this together I just used what was at the top of my parts box. As it happens I have zeners as well.

I am interested in developing the idea further, since finding out that it can be of use to graphic card modders as well. I think it woul dmake for an intersting article for my website

 

I for one am eagerly watching this thread and your website. While I know little about it, I can grasp it and solder it. I have been wondering for some time how the 4pin pwm signal works, and how I could 'replicate' it out to 3pin fans or for that matter, just 'replicate' it to more than 1 fan.

Turns out cloning it is easy, as you can use a 12v molex/gnd for the power for 'more' fans, then tag off the sense line to one of the 'more' fans, and then feed the control line to each of the 'more' fans. Theoretically, your max load could be your 12v rail. Seems to work anyway, as long as you are using pwm fans.

Your schematic is a simple one, much better than ways I was going to attempt. I would very much like to see how you develop a molex power source. I am also curious to see what you would do in the situation of 'replicating' this one signal out to 'more' fans.

And for that matter, on the topic of 'more' fans, I find that many boards now have only 3 or 4 pin headers, some have both. Yet always even a good board does not have enough for what could be done with fans in newer cases. So I look for a simplistic approach. Let Speedfan control 2 zones of fans (not including cpu fan at all), which it can do on many boards. Then, with amalgamation of what you are doing, let the 3pin and 4pin headers send signals to say 2 fans each. Now you control 4 fans via Speedfan. True not individually, but automatically anyway.

Your schematic opens up possibilites for conversion of only the 4pin header pwm to standard 3pin fans. This is good for many instances. Consider if nothing else there is not currently any LED pwm fans. I have looked for 2 years for good pwm fans. Of late they are becoming more and more.

Now let us assume that one machine could be used for media or gaming or coding compilation etc, heavy workloads. Imagine where you would have cpu fan, psu with 2 fans, case fans on (top,side,front,back). Now imagine a common situation where you have only 2 controllable zones (not including cpu zone) with either 3pin or 4pin or combo. You can control 2 case fans. This leaves with fanbus for options. Or if you know your mobo specs, possibly using fan splitter off mobo pin header. For to use large 120mm fan with more cfm speed control is bonus.

So simple idea, have each mobo header 'replicate' signal to drive fans with load driven from 12v molex. No more worry of fan header exceeding specs due to load. However, now for voltage driven speeds, voltage follower or similar circuit is still to be deteremined. You have show how to drive 3pin from 4pin.

But I still wonder, why do you not mention using opto isolator/coupler, to ensure segregation from actual mobo header? Only diode? I am again, only a learner so I understand some, but this is where some of my digging was leading me.

I am again enthused to see what sort of simplistic system can be borne of this thread.

MrWoo.

Edit: here is a link from times earlier when I was studying it more. Maybe some food for thought from the very good Cpemma
http://forums.bit-tech.net/showthread.php?t=136223

 

Bad student question

Hi,
I'm quite bad at electronics but I'm trying to understand how the circuit is working.
When PWM is high (5V) what is limiting the Ic current ? (The PWM current mustn't be greater than 5mA).
Does it needs another resistor between PWM and the base of the transistor ?
Thank you

 

My mistake

Hi,
In facts, we should have something like Ic= Gain x Ib so, with a 0.1A fan and a gain greater than 100, Ib is lesser than 1mA.
Do you have a solution to keep the same ground ?
Thank you

 

I myself have a Zalman CNPS7700 with a 3-pin socket, and wants to use the ASUS AI Suite to control fan speed rather than turning a knob by hand...

The RPM output from any computer fan is open collector, which means it is GND pulses. And the fan itself has no pullup. That's taken care of on the motherboard, which pulls to 12V.
The sense line may pick up things better if the GND isn't chopped up. Maybe.
Perhaps one can put a low pass filter in between to take out the 25kHz "ripple" on the RPM line coming from chopping the voltage to the fan.
This needs some thought.

In the meantime, reading the specifications for 4-pin PWM controlled fans may help.
http://www.formfactors.org/developer\specs\REV1_2_Public.pdf

[Edit]Sorry, I missed the link to that document in the other thread. I even found an older one.

 

If I understand correctly the circuit converts a 4 pin PWM signal to voltage to control a 3 pin fan? This is exactly what I need but I'm not cabable of building it myself. Any chance I could pay you to build one the same as the one in the picture for me?

 

I need to do the exact opposite: I have an old mobo with a 3-pin male fan connector, to which I would like to attach a new fan with a 4-pin female connector.

Any ideas?

thanks

 

Nice necro

If you leave the pins as is, so you're connecting 12v to 12v, GND to GND, and SENSE to SENSE, then if you connect the PWM input of the fan to 5v, then I think it should run at ~100% speed. I've only checked the specs, so I haven't tried this, but it should work... I think

If you want to make it go slower, then you can try experimenting with using 7v, or 5v or whatever instead of 12v for the input voltage, or you'd have to make a little PWM circuit.