Confused over PWM.

GreatScot · Post by **GreatScot** » Tue Oct 07, 2008 4:59 am

Hi folks.

I'm very new at the silencing business... got into it when I was shopping for new computer components and wanted to get a video card that didn't sound like a dust-buster at full speed. That led me here (SPCR) and SPCR led me to Antec 182, ThermalRight Ultra 120 Extreme, 8600GT Silent and a SeaSonic 380W PS.

The only thing that's getting to me now is the relative usefulness of a PWM fan on the TRUE. At the moment, I've got a Scythe S-Flex 21E 1200RPM fan, which I would control with a Zalman FanMate. I know this will keep it quiet, but will it also keep it cool enough when I'm doing some light gaming or rendering 800+ RAW files into 16-bit TIFFs?

The HSF combo is going to be sitting directly above that passively cooled PCIe graphics card, so I'm sure this will add to heat stress. OTOH, I do plan on using both top and back 120mm fans for exhaust, so the volume of air flow through the case will likely be ample, even with TriCools set on low or medium. And yeah, the TriCools will probably get changed out for Nexus or similar fans.

So, would a PWM fan on the TRUE be helpful for me? I'd like to be able to "fire and forget" the CPU cooling... slow and quiet when it's under low stress, faster and slightly less quiet when there's a little more heat stress.

Perhaps I'm obsessing over nothing, and should just wait until I get it built with the Scythe in it... even at moderately fast speeds I might not notice significant amounts of fan noise. I've just read some... conflicting... reports of PWM fans here. Tales of high-pitched whine, problems controlling the speed (?), stuff like that.

If a PWM expert could enlighten me, it would be much appreciated.

-Scot

Strid · Post by **Strid** » Tue Oct 07, 2008 6:12 am

Hi!

I'm not sure which CPU you're using, but if it's not a medieval 125 Watt monster, it will keep fine temps with a S-flex at like 800 rpm no matter load. Your GPU isn't a power hog so it'll do fine.

I have a passive GPU sitting few centimeters below a passive heat sink, and I my temps never go above 50, actually I don't recall more than 50 ever. So I'm pretty sure you'd be fine.

I'd rather have a fixed RPM than having the fan ramp up all of a sudden, just because the motherboard thinks the temp is too hot when it isn't.

And ditch those TriCools!!

GreatScot · Post by **GreatScot** » Tue Oct 07, 2008 5:37 pm

Thanks, Strid.

The CPU is an AMD dual core 5400+, and I have no intentions of playing overclocking games with it... no matter how mediocre it is by current standards, it will still be leaps and bounds faster than the Celeron that's in my current Windows box.

I'll try the Scythe at moderate speeds then.

This still doesn't help me understand the usefulness of PWM though... although I'm guessing it's a lot more helpful for the tiny, high-speed HSF combinations that come OEM with the CPUs.

Cheers,

-Scot

rdibley · Post by **rdibley** » Tue Oct 07, 2008 8:41 pm

Here's some of my assumptions on PWM fans:

-PWM fans shouldn't stall out. The control circuitry onboard should restart the fan if it stalls.
-PWM fans should be able to maintain a more steady speed. Theoretically, two fans of the same size and manufacture should run at exactly the same speed despite slight manufacturing differences.
-The power consumption is less than a variable voltage controlled header off a motherboard

Of course, the first two are based on my assumption that the fan itself has a closed-loop control circuit to drive the fan to the commanded speed.

Anybody have any thoughts/corrections?

pony-tail · Post by **pony-tail** » Wed Oct 08, 2008 10:03 am

I have PWM fans on a couple of systems that I use regularly .
I found that some PWM ( four pin header variety ) can have a very obnoxious / gritty buzz , a Zalman 8700 NT comes to mind . The fans I am currently using are from Arctic cooling and although some are very good a percentage are quite bad for noise . At a guess I would say it is a quality control issue . PWM works very well aside from fan compatibility issues and means that you can keep the fan speed down to a minimum for a given work load . The digital nature of PWM is what makes some fans sound gritty that would otherwise sound smooth when on straight voltage control . This is because PWM operates at 12v and varies the fan speed by switching the power on and off at a relatively high speed ( square wave ) , this causes bursts of torque which can translat to vibrations - and noise in fans not designed for that purpose .
My suggestion would be to give one of the Arctic Cooling fans a go - they are pretty cheap ! If you do try one , get the one with a normal type fan frame if you intend to mount it on a Thermalright cooler as the "open frame" one does not work well with the wire clips.

jhhoffma · Post by **jhhoffma** » Wed Oct 08, 2008 4:17 pm

pony-tail wrote:I have PWM fans on a couple of systems that I use regularly .
I found that some PWM ( four pin header variety ) can have a very obnoxious / gritty buzz , a Zalman 8700 NT comes to mind .

That's typical of Zalman fans. They all pretty much stink. All Zalman fans suffer from a noticeable buzziness when compared to SPCR standard fans. Luckily, on most of their heatsinks, it's not too difficult to do a fan swap.

frenchie · Post by **frenchie** » Thu Oct 09, 2008 4:39 am

+1 for Arctic Cooling PWM fans

I had a bad experience with a nexus PWM fan... it had a bad whine.

cpemma · Post by **cpemma** » Fri Oct 10, 2008 2:00 pm

I've just swapped the Scythe SY1225SL12M that came with my Ninja for a Scythe Kama DFS1225L-PWM. At idle it's silent at 500RPM. The fastest I can get it running is 1090RPM (cores at 55C) and the noise is noticeably less than that from the non-PWM fan, which used to spin at 1140RPM at that same temperature.

I think it's a keeper.

GreatScot · Post by **GreatScot** » Fri Oct 10, 2008 6:33 pm

Thanks for the benefit of experience, folks!

I'll give the non-PWM Scythe a go, set at about a medium speed with a FanMate, and if that makes me grumble, I'll give the Arctic Cooler or Kama PWM fans a try.

I expect that the S-Flex will be just fine... but hey, if I can't obsess about minor hardware things, where's the fun in that???

Merci beaucoup, once again.

-Scot

rdibley · Post by **rdibley** » Fri Oct 10, 2008 7:09 pm

After thinking about my comments earlier, I'm having doubts about what I said. There probably isn't a controller onboard the fan. It just isn't practical to do that. Anybody know what kind of circuit is in the fan?

pony-tail · Post by **pony-tail** » Fri Oct 10, 2008 7:40 pm

The Arctic cooling PWM fan I pulled apart had a small I.C. connected to the PWM signal input a couple of small surface mount transistors and some other components . It appears to me that it reads the pwm signal the switches the field windings on and off according to the signal . Normal brushless DC motors switch the fields on and of on a tachometer signal generated by an internal generator ( this emulates the brushes and commutator in a standard electric motor eg. the starter motor in a car )
link to a pwm motor speed circuit : http://www.discovercircuits.com/H-Corner/PWM.htm
link to wikipedia : http://en.wikipedia.org/wiki/Pulse-width_modulation
hope this helps !

cpemma · Post by **cpemma** » Sat Oct 11, 2008 11:46 am

pony-tail wrote:The Arctic cooling PWM fan I pulled apart had a small I.C. connected to the PWM signal input a couple of small surface mount transistors and some other components . It appears to me that it reads the pwm signal the switches the field windings on and off according to the signal . Normal brushless DC motors switch the fields on and of on a tachometer signal generated by an internal generator ( this emulates the brushes and commutator in a standard electric motor eg. the starter motor in a car )

A 4-pin PWM fan also works fine on a 3-pin header or from a DC power source, so it's safe to assume it contains the same Hall-effect sensor that steers current to the appropriate coils as the hub rotates in a traditional PC fan. The main addition is an extra transistor switch to chop that coil current into variable-width pulses, controlled by similar but lower-power pulses from the motherboard header, so governing the speed.

By putting this current-amplifier switch inside the fan, the "other" fan electronics are able to get an uninterrupted DC supply.

This article shows simplified electronics in 3- and 4-wire fans.