Sign in with
Sign up | Sign in
Your question

PWM Fan- this doesn't make any sense

Last response: in Components
Share
January 12, 2010 5:50:32 PM

I just got an Akasa PWM splitter cable (google AK-CB002) so I can run two PWM case fans and my CPU fan off one PWM mobo header. I've started playing around with it, and have got some funny rpm readings from HW Monitor. Can anyone explain this:

When the CPU can is plugged into it's PWM mobo header, HW Monitor tells me it idles at about 2000rpm. If I plug the splitter into the same mobo header, and then plug the cpu fan into the splitter (so it's effectively just acting as an extension), with no other fans plugged in, HW Monitor tells me it's doing about 3500rpm at idle.

Any ideas why this might be?

Thanks

More about : pwm fan make sense

a b à CPUs
January 13, 2010 1:36:21 AM

the splitter is probably drawing more power and the HW monitor is basing the RPM over power draw

nothing bad just plug the fan into the mobo directly if you don't have another fan needed to be plugged in
a c 134 ) Power supply
a c 199 à CPUs
a b C Monitor
January 13, 2010 1:44:01 AM

The splitter is so you can can run two identical (i.e. CPU HS fans) fans off a single MoBo header which will control the speed of the fans according to the temperature of THAT device (in our example the CPU). You what I assume are two different fan types getting / receiving a common signal.
Related resources
January 13, 2010 7:29:21 AM

Thanks for the comments.

I had planned to use the splitter to run the CPU fan and two case fans. However when I plug the CPU fan into the board through the splitter, it runs at max speed. If I also plugged the case fans in, they would also run at max speed. The whole point of the PWM thing was to reduce noise and unecessary fan use... seeems like a bit of a waste of time at the moment
a b ) Power supply
a b à CPUs
January 13, 2010 7:41:55 AM

Most mobos will have other PWM fan connectors for case fans. Avoid putting case fans on the CPU fan header.
January 13, 2010 9:43:13 AM

I do have another PWM header on the mobo, but no matter what I do the fans run at 2000rpm from it. I have PWM and auto fan control etc. activated in the BIOS, but no use. Not too enchanted by PWM- I think I might go buy a fan controller...
January 13, 2010 10:27:32 AM

If You put more then one fan on the same cable PWM and rpm signals will be messed up. If PWM is not working with one fan and bios options enabled it could be that fan is not made for PWM control
a b ) Power supply
a b à CPUs
January 13, 2010 12:58:59 PM

From your description when you plug the adapter in and only the CPU fan, the mobo believes the fan is running much faster, and you say this is actually true. Just to check: is that true? When you say the fan runs at full speed, do you mean merely that the screen display says it is 3500 rpm, or do you mean that you can tell that it really is running MUCH faster? Assuming it is true, for whatever reason the mobo is trying to feed full power to the fan, and appears to be receiving back from it a correct speed signal. All this is hard to understand given how the system is designed to operate.

First thing I suggest is to make sure you have the CPU fan plugged into the correct set of leads from the adapter. It MUST connect to the one set that has all four wires into it - black, blue, yellow AND green. Then be sure you have the connectors turned the right way - there really should be only one way to push them together. Also be sure you do plug the Moles connector into an output form the PSU.

Now, here is how it is supposed to work. With the adapter in place and all 3 fans connected, they all share via a parallel connection the Ground and +DC supply lines. The + DC line has a rapidly-pulsing (25 kHz) semi-square-wave signal with a "duty cycle" or "% on" that is varied by the mobo controller circuits. The single connector for the CPU fan has one additional line which carries from fan to mobo the fan speed signal, basically 2 pulses per fan revolution. The mobo uses that fan speed for two functions. The first is simply to display the speed for you. The other is to monitor it for any failure (indicated by no pulses at all) so that the system can be shut down quickly before the temperature of the CPU rises rapidly. In most mobos, the fan speed is NOT directly involved in the regulation of the speed.

Fan speed actually is controlled by a simple feedback control loop. There is a temperature sensor built into the CPU, feeding a signal to the mobo via one of its pins. The mobo monitors that and compares it to the setpoint entered in the BIOS Setup screens. Based on that difference, it varies the "% on" character of the PWM voltage signal sent out to the CPU fan, thus changing the fan speed. (Unknown to it, in your case there really are three fans using this same signal.) Now, what would cause the fan speed to increase? Only a rise in measured CPU temperature. What would cause the fan speed displayed on the screen to be higher? Only a higher fan speed signal coming from the fan. So, I am very intrigued by your statement that, in the scenario of using the adapter plus ONLY the CPU fan connected to it, suddenly the fan speed indication is much higher AND (asked you to verify) you agree that the fan speed really IS much higher. Merely inserting the adapter into the circuit should NOT have changed the actual CPU temperature nor altered the fan speed signal being sent to the mobo. By the way, when you did this, did you ALSO ensure that the adapter's connector to a Molex output connector from the PSU was hooked up? That is important because it appears they use that (and not the mobo pin connector) for the Ground lead.

Now, things could get fouled up if you try to use the wrong components, or if the connectors are wrong and the signals on the lines get mixed up. That is why I said to check the connections. But also, it is vital that ALL the fans be designed for PWM use. You can NOT use an older 3-pin fan on one of these adapters!

As a separate issue, I disagree with the concept here. The measured temperature of the CPU internally is the best way to govern the CPU's cooling fan operation. But to control the fans that cool the entire case, to me it makes more sense to use a temperature measurement in the case rather than inside the CPU. That is why mobos use a temperature sensor built into the mobo itself to run the feedback loop that controls at least one of its CASE_FAN connectors. In your case with only one CPU fan and two case fans, here is how I would use the adapter. NOTE: this only will work, as I said before, if BOTH of your case fans are designed for PWM control. Connect the leads marked for "mobo CPU connector" to the mobo's CASE_FAN connector that IS controlled by actual case (mobo sensor) temperature - read your mobo manual for this info. Connect the Molex connector to one of its mates from the CPU. Connect ONE case fan to the leads marked to go to the CPU fan (this one will provide the Case Fan speed signal), and the other case fan to one of the other two fan outputs. At the CPU cooler, just connect its fan directly to the mobo CPU_FAN pin header. In the BIOS, set both fan speed control systems (CPU and Case) to run automatically and, for now at least, leave their settings (like temperature targets and limits) at the default settings. In this scenario, the CPU fan will be controlled by the CPU temperature, and both case fans will be controlled by the temperature measured at the mobpo's built-in sensor.
January 13, 2010 4:54:20 PM

Firstly, thanks Paperdoc for such a detailed reply, I appreciate it!

Secondly, having been driven nuts by this for the last couple of days I came to my computer today to check on a few of the things you mentioned and, would you believe it, it now all seems to be working fine! Plugged the cpu fan directly into the CPU header- 2000rpm, then plugged the splitter into the CPU header and the CPU fan into the splitter- 2000rpm. I'm certain it was acting as I described before, so I can only assume that some of the BIOS settings I was playing around with last night have sorted it out.

As for the concept, I agree that it's not ideal however my intention was to use these fans on my wife's computer, which only has one 4 pin header on the mobo. As I have two 4 pin headers, I think I might keep the fans in my case as they do seem to work better connected the the CASE_FAN connector, as you suggest.

a b ) Power supply
a b à CPUs
January 14, 2010 12:24:44 PM

OK, now I get it. This adapter is ideal for the situation in your wife's computer in which there is no separate Case Fan header with its own speed controller built into the mobo BIOS. Glad to hear you got it all working!
!