So you want PWM control of your new CPU fan?

[Lazzer408]

So you want 4pin PWM control of your new 3pin CPU fan?

I don't know if anyone can use this but I drew it up because I might need it in my ITX build. Feel free to try it.

What it is:
So you went out and bought a high quality aftermarket heatsink/fan but most of them only come with a 3-pin fan. Or maybe you want to use the cpu's fan controller to power additional fans. Or perhapse you want to take advantage of a high speed fan but you don't want to hear it or fork out the cash or deal with adjusting a manual fan controller. Most motherboards have cpu fan throttling adjustments in the bios but you can't take advantage of them with your new 3pin fan! This schematic shows you how to use only 3 components to allow the motherboard's 4-pin CPU fan connector to control 2 or 3-pin fans.

The motherboard's 4-pin fan header:
The 4-pin fan connector has power, ground, tach, and PWM connections. The PWM signal from the motherboard is open or grounded. When the signal is open the fan opperates at full speed. When the signal is grounded the fan is at it's lowest speed. The motherboard controls the fan by switching between these states.

How it works:
The transistor is held in the 'on' state by applying a current to it's base via. the 10k resistor from +12v to the base. The PWM signal from the motherboard sinks (grounds) the base to turn off the transistor. This resistor is needed to both turn the device on and limit the amount of current the motherboard has to sink to 1.2ma. The second resistor, also 10k, limits how low the pwm signal can sink it (and subsequently reduces the current current the motherboard's pwm circuit to .6ma). This keeps the fan opperating at a minimum RPM.

The NPN transistor was chosen because they are generally easier to find.

The 10k resistor in series with the PWM signal can be substituted with a 10k pot to give you 0-50% adjustability of the MINIMUM fan RPM.

 

[Cuzza]

Nifty idea.
I had to think about that for a while but I get it. Most of it - what does the 10k do? (sometimes I wish I'd taken up electronics instead of chemistry, grrrr)

 

[King Wookie]

Any specific transistors or mosfets you recommend?

 

[Lazzer408]

Nifty idea.
I had to think about that for a while but I get it. Most of it - what does the 10k do? (sometimes I wish I'd taken up electronics instead of chemistry, grrrr)

The 10k provides base drive to keep the transistor in the on state. It also limits the amount of current required to pull the base to 0v (during PWM low state) to 1.2ma.

Any specific transistors or mosfets you recommend?

As long as the transistor's power rating is higher then your fan it'll work. Nothing to specific about it.

EDIT - The circuit was changed to just a transistor to make it easier to maintain a minium fan rpm.

 

[<<Onafets>>]

Sticky?
Good job...I'll get my dad to help me try this!

 

[MKmods]

thanks for posting this, well done.

 

[ERazer]

sweet im soo gonna try this when i get home, ty bud

 

[Lazzer408]

EDIT - Circuit tested. Image updated. Videos posted.

The 3rd pic is how your circuit should look when assembled. I just used any old NPN transistor I had with the same pinouts. It works as planned and the RPM will shift from 100% to 50% when the PWM lead is grounded or anywhere in between based on the PWM signal from the motherboard. This works exactly the same way as the stock Intel cooler. The transistor is only noticably warm and will not require any heatsink unless your using a very powerfull fan or multiple fans. The TIP120 is rated at 5 amps and could probably control 6-10 fans with a proper heatsink.

The fan tested is a Magic MGT8012XR-A25. I couldn't find any info on it other then it's 80mm, 12v .39a and a medium-speed fan.

Here's some videos to show how the circuit functions vs. a generic 4-pin soc775 replacement fan (with a bad bearing)

PWM grounded = 50% speed
PWM open = 100% speed

Intel replacement 4-pin fan.

4-pin to 3-pin adaptor circuit.

 

[Lazzer408]

Here it is potted in hot glue and ready to "safely dangle" inside a case. Potting in hot glue is easy. Get yourself a piece of flat aluminum or copper. The base of an old heatsink would work great. Lightly coat the base with oil and use your finger to spread it around. Put some glue on the parts to be potted then press it against the heatsink and wait a few seconds. This will leave a nice flat surface and cool the glue. Spread the oil every time you flatten the glue. When you do the next side it will squeeze some glue over the side you just did. Just trim that off with a razor blade. Repeat for all 4 sides and wa-la.

 

[tw3akm@ster]

Thumbs up man. Nice mod.

 

[Lazzer408]

Thumbs up man. Nice mod.

Thanks man.

 

[Lazzer408]

Has anyone else made this yet? I wan't to hear some good news.

 

[Cuzza]

I totally would just for kicks, but I don't have a board with PWM. *Sigh*

 

[Lazzer408]

Just a quick update. I swapped out the 10k resistor (from the 4-pin's PWM signal) with a 10k pot. This gives you some adjustment for the fan.
This is how mine turned out.

 

[bobosoft]

Oh man, where do I start. Doing all of the math required to predict that circuit would make my head hurt. What I can tell you is that it is not what you wanted and the only reason it works at all for you is because you are using a TIP120 which is a darlington transistor which is an integrated circuit with 2 NPN transistors to create especially high gain. Anyone who used a normal transistor would just end up with a slow running fan. I can also tell you that won't work well when you plug it a motherboard. The fan will vary from slow (at 100% PWM) to slightly slower (at 0%). That is because a PWM signal is not grounded or open it is HI or LOW. In other words 0v or 5V, both of which would pull your circuit low. It also violated intels specified maxium 5V at the PWM by delivering 12V there although I doubt it would actually damage the MOBO.

The PWM signal varies the fan speed by changing its duty signal which is the percentage of time is at 5V vs 0V. For example, a 30% duty signal at 25Khz(intel specified frequency) would be a continuous pulse of 12 microseconds 5V followed by 28 microseconds 0V.

The circuit I would suggest would be to interrupt the GROUND wire (Not power) with an N-MOSFET. connect the source pin to the motherboard side and the drain pin to the fan side. Connect the gate pin to the PWM signal through a 100 Ohm resistor. It is that simple. The IRF510 available at radioshack would work great. Set the minimum fan speed in your bios.

 

[sneekypeet]

It's been stuck, nice work with the mod! I hope a lot of members can benefit from this short but sweet tutorial.

 

[Lazzer408]

Oh man, where do I start. Doing all of the math required to predict that circuit would make my head hurt. What I can tell you is that it is not what you wanted and the only reason it works at all for you is because you are using a TIP120 which is a darlington transistor which is an integrated circuit with 2 NPN transistors to create especially high gain. Anyone who used a normal transistor would just end up with a slow running fan. I can also tell you that won't work well when you plug it a motherboard. The fan will vary from slow (at 100% PWM) to slightly slower (at 0%). That is because a PWM signal is not grounded or open it is HI or LOW. In other words 0v or 5V, both of which would pull your circuit low. It also violated intels specified maxium 5V at the PWM by delivering 12V there although I doubt it would actually damage the MOBO.

The PWM signal varies the fan speed by changing its duty signal which is the percentage of time is at 5V vs 0V. For example, a 30% duty signal at 25Khz(intel specified frequency) would be a continuous pulse of 12 microseconds 5V followed by 28 microseconds 0V.

The circuit I would suggest would be to interrupt the GROUND wire (Not power) with an N-MOSFET. connect the source pin to the motherboard side and the drain pin to the fan side. Connect the gate pin to the PWM signal through a 100 Ohm resistor. It is that simple. The IRF510 available at radioshack would work great. Set the minimum fan speed in your bios.

I had posted a low-side switched mosfet version at one time but for simplicity sake I just left the transistor version. TIP120 was chose because RadioShack has it. Other then the voltage drop across the transistor it's more then sufficient for a single fan. As I stated, the resistors limit the current that the motherboard has to sink. It's cheap, it works. I've tested it with a varity of transistors. 2SD1273 (for example) is not a darlington (see pics) and functions the same as the TIP120 in circuit. My motherboard shows open/grounded on the PWM pin.

If the math for just 3 components makes your head hurt then perhapse electronics is not your strong point. Your more then welcome to post your own solution for a 4 to 3pin adaptor.

 

[bobosoft]

Id rather not argue about it, my concern was preventing frustration on the part people trying to implement your circuit who might think they had done something wrong.

 

[Lazzer408]

Doing all of the math required to predict that circuit would make my head hurt.QUOTE]

To help you understand the math I'll leave the transistor's gain, voltage drop, and fan current out of it. A resistor divider made of two equal values will give you 1/2 the -difference- in the voltage across the divider. If that voltage is 0 and 12v then the base receives 6v. The fan will run at aprox. 50%. Since the 'low side' resistor is adjustable, the base can recieve 0-6v so the fan can run between 0-50% when PWM is 0v. When the PWM is open, the base recieves 12v and the fan is ~100%.

IF the PWM signal is 0/5v then you just need to calculate 5 and 12 across the divider.
12-5=7v (difference)
1/2 of 7 is 3.5v
5+3.5v=8.5v
The base would then recieve 8.5v when PWM is high(5v). In that case the fan runs at about 70.83~% instead of 100%.

Is that what you were trying to get at?

EDIT - Clarification is not argument.

 

[bobosoft]

That's not quite how a bipolar transistor works. The current flowing from the base to the emitter determines the amount of current that can flow from the collector to the emitter. That proportion is called the gain so if you had 1ma through BE and a gain of 1000 (as is typical of darlington) you would allow 1AMP through CE. Calculating the current through BE after your voltage divider and taking into consideration the inductor it has to go through before ground is the math I was referring to.

 

[Lazzer408]

That's not quite how a bipolar transistor works. The current flowing from the base to the emitter determines the amount of current that can flow from the collector to the emitter. That proportion is called the gain so if you had 1ma through BE and a gain of 1000 (as is typical of darlington) you would allow 1AMP through CE. Calculating the current through BE after your voltage divider and taking into consideration the inductor it has to go through before ground is the math I was referring to.

I wasn't explaining the function of the transistor. I know it's not a voltage device. That's why I said I left the transistor's characteristics out of it. I was focusing on the point you made about the PWM's high state being 5v as opposed to an open state. I see how that could cause an issue and explained why. My board's PWM pin is going from 0 to infinity not 0-5v. If the circuit is connected, then PWM goes from 0-12v. Do you have a link handy to the Intel spec on the 4pin? I can't find it now. I'll look into later. I have to run out for awhile.

 

[bobosoft]

Here is the spec:

http://www.formfactors.org/developer%5Cspecs%5C4_Wire_PWM_Spec.pdf

You are right about it being open, not 5v, Intel says it has to be pulled up in the fan.

"This signal must be pulled up to a maximum of 5.25V within the fan."

the 2SD1273 is not darlington but it might as well be since it also offers veyr high gain (min 500, max 2500)

 

[Lazzer408]

It's being pulled anywhere from 0-6v with what I built. I'm sure it's lower then 6v if you consider the base current dropping the voltage across R1 with the fan load on the transistor. I doubt it's to 5.25v though. I tried to keep this as simple as I could. It worked good enough for what I needed so I thought I'd post it. Is it perfect? no, but it's dirt cheap and it works. Beats listening to a 120mm fan running at full speed. =)

I never looked up the 1273 datasheet. It was grabbed from my drawer of NPNs. The fan isn't a huge load so most to220 NPNs will work for this circuit. Just don't mount it with the tab.

 

[Wile E]

How in the world did I know this thread was started by Lazzer? lol

 

[Lazzer408]