12V 4 wire fan datasheet?

Associate
Joined
3 May 2018
Posts
604
Has anyone seen or can anyone suggest where I would find the datasheet for a 4 wire PWM fan?

Reviews and how-tos are all very well, but I need to know detailed particulars. Such as:

* PWM frequency range
* PWM voltage levels for high and low
* Tact output voltage levels for high and low

I realise they should all be fairly standard which is why most of the PC hardware sites skip straight past these details, but I want to build a fan controller so I need the exact details.

If I can just find one, FULL, datasheet for a standard PWM w/ tact fan I should be grand.
 
You're better off hooking up to a scope and checking for yourself IMO. This isn't an actual standard, so any data sheets you find will be for a motherboard or similar, not the electronic specifications of the interface.

PWM frequency is not fixed AFAIK, obviously it's usually 20-25KHz but I think I've seen 17 before.

The PWM control signal is, again AFAIK, pull-down i.e. low means enable. I've heard it's 5V but have nothing to validate that. This arrangement means omitting the control signal will give a duty cycle of 100% so fans will still run. Low means 0V in this case.

I don't believe the tach signal has a low/high voltage, in fact I doubt the high has a particularly fixed voltage. It's a pulse, so possibly not regulated at all. I could be wrong, PWM fans seem pretty advanced internally these days.

If you do find any useful data please do share it here :)
 
So I'm told that this is the 'official' specification:
http://www.formfactors.org/developer\specs\rev1_2_public.pdf

So PWM is open drain with a 5V pull up in the fan. This is for "fail safe" meaning if the PWM input drops out the fan will run at 100% rpm as the 5V pull up will hold it at 100% duty cycle. If you pull it low (short it to ground) for 25% of the time you get 75% rpm and so on. It states a maximum of 5mA when pulled low, so I assume it has a 1K pull up resistor.

The Tach is also open drain with the motherboard / fan controller providing the pull up voltage this time and the fan pulling it down for each pulse. The hall effect sensor then shorts the tach pin to ground so a 10K resistor should pull it up fine without too much current. It specifies 12V-13.6V pull up, others have said this could be lower, say 5V preventing the need to level shift the controller sampling the tach (which is probably going to be a 5V or 3.3V micro-controller).

It's just an electronics project I'd like to attempt to make a nice standalone fan controller with a nice open REST API interface. This might require the actual "box" of the fan controller be outside the case to get Wifi signal, though if your access point is close it would be fine tucked away inside.

Ideally:
6 independent PWM outputs, 12 is possible.
Several DS18B20 1 wire temp sensors (they chain along the same 3 wires I think you can technically support dozens of them)
Several PWM inputs so some motherboard fan PWM can be passed through. This would just bypass the fan controller and connect the PWM and Tachs with mosfets to configured outputs.
SATA (or molex) powered.

A lot of fan controllers exist, but they either have garish interfaces, are fairly limited, use proprietary windows only software and/or are very expensive, I figured I could come up with my own and do better!

Also found this to take most of the grunt work out of the PWM:
https://datasheets.maximintegrated.com/en/ds/MAX31790.pdf

Pain the A.R.S package to solder though.
 
Yeah I find it annoying how many useful ICs are like QFN only, etc. :( I'm really going to have to invest in a hot air rework station some time.
 
Yeah I find it annoying how many useful ICs are like QFN only, etc. :( I'm really going to have to invest in a hot air rework station some time.

I got an 898D clone for £35 on ebay, works a treat. You'll need solder paste too, some needle point micro-tweasers and some form of magnification like a jewelers visor (or for around £125 see below).
https://www.youtube.com/watch?v=rgcupYOnvBo

I also got a Andonstar USB/HDMI microscope, it takes some beautiful shots... here are some of my USB sound card I made:
0COBwae.jpg
6NjBcCU.jpg
FeA41QY.jpg

I hope to get a few shots of my GPU with this when I remove the air cooler to fit a Kraken G12.
 
Generally I don't need magnification... for now... though I suspect that will change when I'm the other side of 40 :s I've hand soldered the PCM2902 a few times to break out boards.

It is stuff like the CP2114 that really makes me want to be able to solder QFN, etc.

EDIT: So not fun to hand solder heh:

yDjb5sQ.jpg
 
Last edited:
Generally I don't need magnification... for now... though I suspect that will change when I'm the other side of 40 :s I've hand soldered the PCM2902 a few times to break out boards.

It is stuff like the CP2114 that really makes me want to be able to solder QFN, etc.

EDIT: So not fun to hand solder heh:

Nice job. Tac and drag I assume? To be honest sometimes it's easier to do them with the iron.

The worry I have with the TQFNs is not getting it soldered it would be getting rid of any bridges under the chip once soldered. You can solder them with an iron by running it along the edge just, but it's better to set it down onto paste and hot air it, making sure to get the chip itself over the reflow temp so the solder underneath melts and flows onto the pads.

EDIT, for some scale reference (in additon to the 50p) the PCM 290x chips are about 8mm long and about 4mm wide, smaller than your wee finger nail. It has 14 pins along each side which are only about 0.2mm apart.
 
Yeah one corner then drag soldered. I killed the second one I did but the first and 3rd and 4th worked fine though they seem a bit fussy on prototype board with regard to the crystal placement and distance, etc.
 
Yea crystals need to be right up close and the USB is a differential pair which needs to be distance matched on both D+ and D- though probably less critical.
 
Back
Top Bottom