3-pin vs. 4-pin fan connectors, what's the difference?

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

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
levander
Posts: 21
Joined: Sat Jun 25, 2005 11:09 am
Contact:

3-pin vs. 4-pin fan connectors, what's the difference?

Post by levander » Wed Nov 15, 2006 12:39 pm

First question, what name do I use to differentiate between four pin connectors that plug into the power supply, versus four pin connectors that plug into fan connectors on the motherboard?

In this post, when I talk about 4 pin connectors, I'm talking about the type that plug into the motherboard.

Now, the first time I've seen a 4 pin fan connector was an article about the Core 2 Duo on tomshardware that said the 4th pin is there so that the motherboard (and assumedly software that talks to the motherboard) can control fan speed. But, I thought that it was already possible to do that even with the old 3 pin fan connector? I think I vaguely remember I've done that in the past with an applications called FanSpeed. Or, do I remember this wrong?

So, the second question is, if you can use the motherboard to control the fan speed with an old three pin connector, what's the point of the 4th pin?

tempoct
Posts: 136
Joined: Tue Aug 30, 2005 4:16 pm
Location: Los Angeles, CA

Re: 3-pin vs. 4-pin fan connectors, what's the difference?

Post by tempoct » Wed Nov 15, 2006 1:26 pm

4-pin on mobo is the new PWM fan connector. It takes different type of fan. I believe you can plug 3-pin in but it will always run at full speed. PWM is new kind of speed control
viewtopic.php?t=20969&view=previous&sid ... b1e4406fab

The 4-pin from PSU (that you plug into DVD drive, etc) is called "Molex" 4-pin connector.

I have the same problem myself and seeking solution too. PWM fan is hard to find and highly expensive. Only one that available in 120mm (AFAIK) is the JMC.
http://www.performance-pcs.com/catalog/ ... ts_id=3977
levander wrote:First question, what name do I use to differentiate between four pin connectors that plug into the power supply, versus four pin connectors that plug into fan connectors on the motherboard?

In this post, when I talk about 4 pin connectors, I'm talking about the type that plug into the motherboard.

Now, the first time I've seen a 4 pin fan connector was an article about the Core 2 Duo on tomshardware that said the 4th pin is there so that the motherboard (and assumedly software that talks to the motherboard) can control fan speed. But, I thought that it was already possible to do that even with the old 3 pin fan connector? I think I vaguely remember I've done that in the past with an applications called FanSpeed. Or, do I remember this wrong?

So, the second question is, if you can use the motherboard to control the fan speed with an old three pin connector, what's the point of the 4th pin?

levander
Posts: 21
Joined: Sat Jun 25, 2005 11:09 am
Contact:

Re: 3-pin vs. 4-pin fan connectors, what's the difference?

Post by levander » Wed Nov 15, 2006 2:57 pm

tempoct wrote:4-pin on mobo is the new PWM fan connector. It takes different type of fan. I believe you can plug 3-pin in but it will always run at full speed. PWM is new kind of speed control
But, couldn't you control speed of the 3-pin fans via the motherboard? Why do we need the 4th pin to do that?

pyogenes
Posts: 273
Joined: Wed Oct 11, 2006 10:38 am
Location: Chicago

Re: 3-pin vs. 4-pin fan connectors, what's the difference?

Post by pyogenes » Wed Nov 15, 2006 7:30 pm

tempoct wrote:4-pin on mobo is the new PWM fan connector. It takes different type of fan. I believe you can plug 3-pin in but it will always run at full speed.
That's not right. Whether or not the voltage can be regulated for the 3-pin fans depends purely on the motherboard - not the presence of the 4th pin. For example the Asus P5W DH Deluxe has a 4-pin PWM fan header but it will readily accept 3-pin fans and control speeds for those fans.

billrob
Posts: 11
Joined: Sun Nov 19, 2006 1:47 pm
Location: Rancho Palos Verdes, CA

Post by billrob » Sun Nov 19, 2006 2:04 pm

This isn't exactly a reply, but it is on the same subject. I have a Gigabyte motherboard model GA-55plus-S3G, with an AMD2 processor socket, in a Silverstone LC17 case. I have set the BIOS to enable System Smart Fan Control, which reduces the fan speed and noise a little. The CPU fan is still very noisy. The BIOS refers to a CPU Smart Fan Control, which may be enabled when using a CPU fan with a 4 pin connector. The Gigabyte fan model Classic AM2 (GH-ED821-FE) is a 3 pin connector. I cannot install the Zalman CPU cooler because I do not have enough clearance to the case (5.4â€

71monaro
Posts: 26
Joined: Mon Jun 26, 2006 6:03 pm
Location: South Australia

Post by 71monaro » Sun Nov 19, 2006 3:02 pm

The 4 pins used on the mobo are (not in order) earth, 12 volt power, pwm power and rpm sensing. 1) Earth - pretty self explanitary.
2) 12v power is just like the 3 pin mobo version and on many (but not all) motherboards this can have a voltage linearly variable with cpu temperature, hence varying speed from 7v to 12v.
3) the pwm is a pulse signal that tells the fan to turn on and off with full 12 volts pulses. The time that it is off is varied (longer off = slower fan, longer on = faster fan). Why do this? well it allows varying the fan speed technically all the way down to stop. whereas the 3 pin version has to have a certain voltage before it will start. Must note this is done inside the fan, hence special fans, the pwm signal from the mobo is insufficient to run normal fans this way.
4) The speed sense wire is a signal back from the fan to mobo. It is pulses with a frequency = speed. Now it must have a permanent power source.It can be varying 5-12v as long as continous not pwm turning off. Hence needing the 12v lead when operating a 4pin pwm fan.

levander
Posts: 21
Joined: Sat Jun 25, 2005 11:09 am
Contact:

Post by levander » Sun Nov 19, 2006 4:13 pm

71monaro wrote: 3) the pwm is a pulse signal that tells the fan to turn on and off with full 12 volts pulses. The time that it is off is varied (longer off = slower fan, longer on = faster fan). Why do this? well it allows varying the fan speed technically all the way down to stop. whereas the 3 pin version has to have a certain voltage before it will start.
Okay, so the 4 pin connectors allow more flexibility over the control of the fan speed? The motherboard has full control over the fan speed of a 4 pin fan. This is because the fan speed control and the power supplied to the fan are two separate signals with a 4 pin fan? With a 3 pin fan, the motherboard can only control the speed by regulating how much power is supplied to the fan?
71monaro wrote: Must note this is done inside the fan, hence special fans, the pwm signal from the mobo is insufficient to run normal fans this way.
I don't understand this note at all. Are "special fans" 4 pin fans that don't have PWM control implemented inside them?

Monaro, I'm just trying to verify my understanding of what you said. Thanks a lot for the reply. It was starting to seem very wierd to me that no one on one of the most in-depth technical communities on the internet knew what the whole point of this new fourth pin on the fans was, when 3 pin fans could do something that I now know was just similar, but not exactly the same functionality.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Sun Nov 19, 2006 4:54 pm

It was starting to seem very wierd to me that no one on one of the most in-depth technical communities on the internet knew what the whole point of this new fourth pin on the fans was
It's there, but you need strong search-fu to find it:

4-pin fans: a primer
what's the point of the 4th pin?


Overview:
4-wire fans are gaining in popularity and offer several significant advantages over the more common 3-wire variety. With typical 3-wire fans running with a PWM fan controller, the operating voltage is switched off and on to control speed. This presents a problem as power is not on long enough to complete an entire tach cycle. The PWM drive in effect chops the tach signal output and may produce incorrect readings. To obtain an accurate measurement it is necessary to occasionally provide a longer pulse. This technique is known as pulse stretching and can contribute to commutator noise.

In contrast, 4-wire fans provide constant power to the fan. Electronics continue to function at full power even as the drive coils are pulsed resulting in acoustics almost as good as linear (constant dc) power. The component count in circuits used to drive and monitor 4-wire fans is substantially reduced. Driving the coils at rates greater than 20 kHz moves the noise outside of the audible range, so typical PWM fan-drive signals use a rather high frequency (>20 kHz).
Also see the table in the link, it lays out the advantages of 4-pin fans very clearly.

71monaro
Posts: 26
Joined: Mon Jun 26, 2006 6:03 pm
Location: South Australia

Post by 71monaro » Mon Nov 20, 2006 1:01 am

levander wrote: Must note this is done inside the fan, hence special fans, the pwm signal from the mobo is insufficient to run normal fans this way. I don't understand this note at all. Are "special fans" 4 pin fans that don't have PWM control implemented inside them?
It is possible to construct a pwm signal from 12 volts to run a 3 pin fan but then you would not have usable speed sensing (can turn off and ignore). But the mobo does not do the complete job. It makes a small pulse. This pulse is then converted to a 12volt square wave inside the fan. It would have been a lot more universal if this was done by the mobo, but because it isnt then the only way you can use the 4th pin to control speed is using a 4 wire fan designed for this.

billrob
Posts: 11
Joined: Sun Nov 19, 2006 1:47 pm
Location: Rancho Palos Verdes, CA

Post by billrob » Mon Nov 27, 2006 8:48 am

I would like to add in with my understanding of the systems. Conventional (3 pin) fans control the speed by controlling the voltage. Current mobo (such as mine) sense the CPU temperature and adjust the constant voltage (say 8 volts) to reduce the fan speed. This has a disadvantage because there is a minimum voltage (maybe 5 volts) below which a DC motor cannot ovecome the internal friction and will not start, or maybe run. Maybe I am wrong, but when I start my PC (I have Gigabyte "system smart fan" enabled in the BIOS) there seems to be a voltage surge (maybe 12 volts), because the fans start fast for a second or two, then slow down. If true, very clever on Gigabyte, because a DC motor, once started, will run at a lower voltage than it will start at.
Now, PWM, as mentioned, applies a full 12 volts to the motor, but the power to the motor (hence the speed) varies as the length of the pulse in the cycle, and the cycle is about 20KHz. You can run a PWM DC motor at 1 RPM if you want to. Huge advantage in runnign a fan slowly. I assume (and this is the question) that if I plug a 4 pin PWM motor into the mobo, the mobo will sense that and assume a PWM speed control mode instead of a voltage control mode. Would appreciate thoughts/comments.

sjschwinn
Posts: 68
Joined: Wed Jul 05, 2006 1:38 pm

Post by sjschwinn » Mon Nov 27, 2006 9:24 am

jaganath wrote:
It was starting to seem very wierd to me that no one on one of the most in-depth technical communities on the internet knew what the whole point of this new fourth pin on the fans was
It's there, but you need strong search-fu to find it:

4-pin fans: a primer
what's the point of the 4th pin?


Overview:
4-wire fans are gaining in popularity and offer several significant advantages over the more common 3-wire variety. With typical 3-wire fans running with a PWM fan controller, the operating voltage is switched off and on to control speed. This presents a problem as power is not on long enough to complete an entire tach cycle. The PWM drive in effect chops the tach signal output and may produce incorrect readings. To obtain an accurate measurement it is necessary to occasionally provide a longer pulse. This technique is known as pulse stretching and can contribute to commutator noise.

In contrast, 4-wire fans provide constant power to the fan. Electronics continue to function at full power even as the drive coils are pulsed resulting in acoustics almost as good as linear (constant dc) power. The component count in circuits used to drive and monitor 4-wire fans is substantially reduced. Driving the coils at rates greater than 20 kHz moves the noise outside of the audible range, so typical PWM fan-drive signals use a rather high frequency (>20 kHz).
Also see the table in the link, it lays out the advantages of 4-pin fans very clearly.
This is very good information. I personally was not aware of this.

Now the question is, who makes 4-pin fans? I have never seem them...

billrob
Posts: 11
Joined: Sun Nov 19, 2006 1:47 pm
Location: Rancho Palos Verdes, CA

Post by billrob » Mon Nov 27, 2006 10:11 am

I am finding more as I research. Not sure if I know how to set up links. Try Evercool at sidewindercomputers.com, or ARX92mm lga 775 at adpmods.com. There seem to be some (probably more every day), but you need to dig. I'm backing away from a Zalman8000. It doesn't offer a PWM fan(??) It looks like a thermalright heatsink with a 92mm PWM fan might be a good combination.

truckman
Posts: 82
Joined: Sat Apr 08, 2006 10:31 pm
Location: Northern California

Post by truckman » Mon Nov 27, 2006 10:24 am

sjschwinn wrote:Now the question is, who makes 4-pin fans? I have never seem them...
It looks like Arctic Cooling has recently added 4-pin fans to their product line. They even have a feature that allows multiple fans to be slaved off one motherboard connector while still allowing the individual fan speeds to be monitored.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon Nov 27, 2006 11:06 am

You can run a PWM DC motor at 1 RPM if you want to. Huge advantage in running a fan slowly.
On the downside, many 3-pin fans including Yate Loon/Nexus etc react in a negative way to PWM, including commutator noise, other electrical noise, etc, as mentioned above. This is why PWM fan control is relatively uncommon on SPCR.

billrob
Posts: 11
Joined: Sun Nov 19, 2006 1:47 pm
Location: Rancho Palos Verdes, CA

Post by billrob » Mon Nov 27, 2006 11:47 am

Absolutely agree. I would never try to run a PWM fan in a mobo that did not have a 4 pin connector and support PWM operation

sjschwinn
Posts: 68
Joined: Wed Jul 05, 2006 1:38 pm

Post by sjschwinn » Mon Nov 27, 2006 12:02 pm

jaganath wrote:
You can run a PWM DC motor at 1 RPM if you want to. Huge advantage in running a fan slowly.
On the downside, many 3-pin fans including Yate Loon/Nexus etc react in a negative way to PWM, including commutator noise, other electrical noise, etc, as mentioned above. This is why PWM fan control is relatively uncommon on SPCR.
Yes, but I think the post above mentions that fact that some (much?) of the commutator noise from current PWM controllers (for 3-pin fans) comes from "pulse stretching" of the PWM signal to help get a better RPM tach signal. A fan designed for 4-pin PWM should be able to be almost as quiet as those designed for DC operation.

billrob
Posts: 11
Joined: Sun Nov 19, 2006 1:47 pm
Location: Rancho Palos Verdes, CA

Post by billrob » Mon Nov 27, 2006 1:53 pm

My point (and I may be wrong) is that a PWM system is inherently more capable of running a DC fan (designed for PWM) at low speeds. IF (big if) temperature balance in the CPU can be achieved at a lower speed than voltage control (via a 3 pin configuration) can run the fan, I would argue that a PWM system is always quieter.

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon Nov 27, 2006 2:23 pm

My point (and I may be wrong) is that a PWM system is inherently more capable of running a DC fan (designed for PWM) at low speeds.
Unfortunately, this is of little benefit as current non-PWM fan controllers can already take 3-pin fans down to inaudible levels; go any lower and the fan pushes so little air it may as well not be there. Also, it's not immediately obvious to me that PWM would be able to get a fan to spin at a voltage where straight DC cannot; if it won't spin at 5V DC, why would it spin when supplied with a simulated-5V PWM signal?
Last edited by jaganath on Mon Nov 27, 2006 4:16 pm, edited 1 time in total.

billrob
Posts: 11
Joined: Sun Nov 19, 2006 1:47 pm
Location: Rancho Palos Verdes, CA

Post by billrob » Mon Nov 27, 2006 2:50 pm

Good discussion. I can speak to the technical issue. Every DC motor has internal friction. Voltage control systems become more unstable (speedwise) as the voltage is reduced to a level that just overcomes the internal friction and back EMF. With a PWM motor, you essentially "zap" the motor at 12 volts with every pulse (remember it is 20KHz), and the internal friction becomes less of an issue. The total power to the motor (in both cases) may be the same. I can speak from personal experience (not computers) that you can achieve more stable speed control at low speeds with a PWM motor. I think that is what the market is saying as PWM becomes more available. You may be correct on the issue of "what is better than inaudible?".

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Mon Nov 27, 2006 4:34 pm

I can see what you're saying about the advantages of PWM, one very specific situation I thought of where 4-pin PWM really comes into its own is this: let's say you are a big system integrator (Dell or HP, etc) and you are sending systems all over the world, where ambient temps will differ massively and you have a temp-controlled PWM fan on the CPU, you will want that fan to have a very large speed range (say 500-2500rpm) to cope equally well in Alaska as it does in Australia, so in that scenario the excellent low-speed control offered by PWM allows you to hook up a powerful fan but run it slowly when the environment is cool. So for "beige boxes" it allows much more sophisticated fan control, but SPCR PC fans are normally tailored specifically to their environment so it's not such an advantage (ie SPCRers use such low-speed fans that reducing the voltage may not even be necessary, eg the Noctua 800rpm fan).

kater
Posts: 891
Joined: Thu Sep 07, 2006 11:20 pm
Location: Poland

Post by kater » Thu Nov 30, 2006 7:34 am

Seems I found a place where folks with indepth knowledge of PWM and related issues come to share info. Could you also please take a look at this thread and try to shed some light on the issue? Thanks in advance!

viewtopic.php?t=36188

Post Reply