SP 120, get PWM or 3 pin for Rad

[felang]

I have a 240mm rad with the stock XSPC fans that came with my kit currently. They are standard 3 pin fans and I power them both (using 3 pin splitter) off the CPU 3 pin connector from my motherboard (Asrock Z77 Extreme 4, has both 4 pin PWM and standard 3 pin cpu fan connectors). I have auto fan controlled according to CPU temp configured in Bios, and it works fine through the 3 pin connector.

I decided recently to replace the fans with Corsair SP120 perfomance edition fans, these are available as both 3 pin and PWM.

What´s the advantage of getting PWM fans when 3 pin fans are also auto controlled by my mobo? Am I missing something?

 

[tracerbullet]

I don't think there ends up being much difference, from what I've seen some / many motherboards control the 3 pin header using a method that basically amounts to PWM.

The general idea seems to be go with the best fan you can afford, based on noise or airflow or whatever your requirement is. If you have a few contenders then more often than not PWM gets recommended.

Supposedly a 3 pin fan at low speeds can "click", but I've yet to hear it or read about anyone that has had it become a problem, only read that it can happen. If you're looking at the Corsair SP fans then maybe lack of noise isn't your #1 priority and the cheaper option (of the 3 /4 pin choice) is better.

Hope that helps... would like to see someone who really knows for sure chime in.

 

[felang]

I don't think there ends up being much difference, from what I've seen some / many motherboards control the 3 pin header using a method that basically amounts to PWM.

The general idea seems to be go with the best fan you can afford, based on noise or airflow or whatever your requirement is. If you have a few contenders then more often than not PWM gets recommended.

Supposedly a 3 pin fan at low speeds can "click", but I've yet to hear it or read about anyone that has had it become a problem, only read that it can happen. If you're looking at the Corsair SP fans then maybe lack of noise isn't your #1 priority and the cheaper option (of the 3 /4 pin choice) is better.

Hope that helps... would like to see someone who really knows for sure chime in.

Yeah you hit the nail on the head... I am looking for something quieter, as long as everything stays under 70 degrees with low noise I'm happy. I'd rather go with the performance edition instead of quiet edition cause I figure I could always lower them to around 1000 rpm using the bios fan control anyway.

Also, although you can get them in pairs, Im not really sure if they include the necessary splitter to be able to run them off the same header.

I would like to know if there's an actual difference between PWM and 3 pin though? Maybe greater available rpm range for PWM? I'll probably just wait for either of them to go on sale, which seems to happen like once a month on newegg.

 

[aigomorla]

ur mixing apples with oranges..

PWM.. 4 pin and regular 3 pin are whole different beasts.

The 3 pin can be controled via motherboard diode temp or software, but its based mostly based on a temp diode on the motherboard.

Now we know how notorious these diodes are in being inaccurate.

The PWM goes on LOAD.
The whole theory behind it is.. your fans do not need to be spinning at value X all the time. The only time it needs to be at value X is when cpu load is Y.

As the Y changes the value X reported tells the PWM to either spin faster or slower though pulse.

What is a better system?

PWM... basically if ur not overclocking.. and your not doing a nuclear reaction on your cpu, pwm wins..

If your overclocking.... your idle temps are massive... with overclocking temps massive... well pwm doesnt help u very much when your fans are already set to 100%.


Basically if your an overclocker... u have high vcore voltages... high idle temps.... u really dont want your fan on low.. or ever to go on low.
Even after a long processor test run... you still want your fan on high for at least 10 min to bring average temp on the sink back down.

PWM wouldnt do that... u would have to recool your processor with low fans which could take longer.

So if ur a avid overclocker.. u probably want to avoid PWM.. or get a PWN fan controller and set your own ranges.
If your not a avid overclocker... PWM would save your ears a few dB. But that can also be ahighly variable by the fans you use.

 

[felang]

ur mixing apples with oranges..

PWM.. 4 pin and regular 3 pin are whole different beasts.

The 3 pin can be controled via motherboard diode temp or software, but its based mostly based on a temp diode on the motherboard.

Now we know how notorious these diodes are in being inaccurate.

The PWM goes on LOAD.
The whole theory behind it is.. your fans do not need to be spinning at value X all the time. The only time it needs to be at value X is when cpu load is Y.

As the Y changes the value X reported tells the PWM to either spin faster or slower though pulse.

What is a better system?

PWM... basically if ur not overclocking.. and your not doing a nuclear reaction on your cpu, pwm wins..

If your overclocking.... your idle temps are massive... with overclocking temps massive... well pwm doesnt help u very much when your fans are already set to 100%.


Basically if your an overclocker... u have high vcore voltages... high idle temps.... u really dont want your fan on low.. or ever to go on low.
Even after a long processor test run... you still want your fan on high for at least 10 min to bring average temp on the sink back down.

PWM wouldnt do that... u would have to recool your processor with low fans which could take longer.

So if ur a avid overclocker.. u probably want to avoid PWM.. or get a PWN fan controller and set your own ranges.
If your not a avid overclocker... PWM would save your ears a few dB. But that can also be ahighly variable by the fans you use.

I had no idea that PWM was based on cpu load only, always thought rpm was increased or decreased in response to temps only. The motherboards I´ve had usually vary PWM fans´ rpm ramp according to a "target temp" set in the bios, but I guess this could just be a parameter that adjusts how aggresively rpms are increased or decreased according to a predefined ramp.

The thing is I know Aigomorla definitely knows what he´s talking about as he´s helped me before. So basically in my situation (CPU and GPU in the same water cooling loop) I figure I´d be better served by 3 pin fans because fan rpm (actually voltage) would increase according to CPU temp.

For example, even if CPU load is low but GPU load is high enough to increase the whole wc loop temp and in turn CPU temp, rpm´s should increase in accordance to the predefined cpu temp/rpm ramp. PWM fans would just keep spinning according to CPU load without taking actual temps into account, right?

 

[Bryf50]

ur mixing apples with oranges..

PWM.. 4 pin and regular 3 pin are whole different beasts.

The 3 pin can be controled via motherboard diode temp or software, but its based mostly based on a temp diode on the motherboard.

Now we know how notorious these diodes are in being inaccurate.

The PWM goes on LOAD.
The whole theory behind it is.. your fans do not need to be spinning at value X all the time. The only time it needs to be at value X is when cpu load is Y.

As the Y changes the value X reported tells the PWM to either spin faster or slower though pulse.

What is a better system?

PWM... basically if ur not overclocking.. and your not doing a nuclear reaction on your cpu, pwm wins..

If your overclocking.... your idle temps are massive... with overclocking temps massive... well pwm doesnt help u very much when your fans are already set to 100%.


Basically if your an overclocker... u have high vcore voltages... high idle temps.... u really dont want your fan on low.. or ever to go on low.
Even after a long processor test run... you still want your fan on high for at least 10 min to bring average temp on the sink back down.

PWM wouldnt do that... u would have to recool your processor with low fans which could take longer.

So if ur a avid overclocker.. u probably want to avoid PWM.. or get a PWN fan controller and set your own ranges.
If your not a avid overclocker... PWM would save your ears a few dB. But that can also be ahighly variable by the fans you use.

You probably have more experience than I do but this doesn't really sound right. As far as I can tell on my Asrock board all fans are controlled by temperature regardless of PWM or conventional. Maybe on some boards that is the way it is but there's no technical reason why it has to be that way.

Really the only difference is how the speed is controlled. PWM is controlled by a pulse on that extra wire. Basically different RPM depending on the width of the pulse. Conventional fans are simply controlled by adjusting the voltage ie. at lower voltages the fan spins slower. One benefit is that PWM fans are always running at 12v which is supposed to be better for the life of the motor. Also PWM allows for more predictable and precise control as fan motors behave differently with voltage control and don't linearly lower RPM with voltage.

 

[aigomorla]

You probably have more experience than I do but this doesn't really sound right. As far as I can tell on my Asrock board all fans are controlled by temperature regardless of PWM or conventional. Maybe on some boards that is the way it is but there's no technical reason why it has to be that way.

Really the only difference is how the speed is controlled. PWM is controlled by a pulse on that extra wire. Basically different RPM depending on the width of the pulse. Conventional fans are simply controlled by adjusting the voltage ie. at lower voltages the fan spins slower. One benefit is that PWM fans are always running at 12v which is supposed to be better for the life of the motor. Also PWM allows for more predictable and precise control as fan motors behave differently with voltage control and don't linearly lower RPM with voltage.

that 4th wire is the pulse modulation cable.
And yes you are correct in that it pulses however the rate at which it pulses is based on the load of the processor.

Example:
http://s125.photobucket.com/user/aigomorla/media/Nadeshiko/MOL012.mp4.html

As the load increases you can hear my pwm fans really blast.... cuz my cpu load was at 100... and then it falls back to idle values... and then back to 100, and then back to idle.

As for pwm being better in life... false... no way.
You will have a fan last longer at 7V then on PWM any day of the week.
Less current => always wins in durability... think of reving an engine.. whats better reving it at 3k rpms constant or reving it 6k in pulses..

The one problem i have with pwm is that the fans arent reliable at all compared to the non pwm ones. (my personal experience hence why i got rid of all my pwm fans.)
However the qty in pwm fans tested is relatively small... but i have had more pwm fans fail on me then non pwm fans.

Another downside with PWM is if your system hangs... sometimes your fan will freeze...
hang'd system with frozen fans -> heat being dumped into system by LC pump -> That spells disaster for a LCer like me whose fans are on a radiator cooling coolant.

On a Air sink... maybe not so much... a locked processor means a idle processor... and most high end heat sinks can keep a processor in idle in check.

 

[Automaticman]

Really the only difference is how the speed is controlled. PWM is controlled by a pulse on that extra wire. Basically different RPM depending on the width of the pulse. Conventional fans are simply controlled by adjusting the voltage ie. at lower voltages the fan spins slower. One benefit is that PWM fans are always running at 12v which is supposed to be better for the life of the motor. Also PWM allows for more predictable and precise control as fan motors behave differently with voltage control and don't linearly lower RPM with voltage.

This is correct. Do not confuse the method that the speed of the fan is physically controlled with the data used to determine how fast the fan should spin. The speed of a PWM fan (or 3 pin fan) can be manually set to a certain speed, set to increase based on the temp o fthe CPU, or the GPU, or...whatever. That part all comes down to the software doing the controlling (ie how you have Speedfan or similar SW set up to adjust the fan speed).

The main advantage PWM fans have is , as Bryf50 said, that have more fine grained control of a PWM fan. Another good thing is that, because they see 12v all the time, you can start the fan at a very low speed if you want, while a 3 pin voltage controlled fan might not start spinning until a higher speed/voltage is applied.

I can't really comment on the longevity part, I've never looked into it. I would expect the motor portion of a fan spinning at full speed to last longer that a fan spinning at 50% speed due to the fact that electric motors actually draw less current the faster they spin, but then the bearings may go out faster...

The big thing to look for is what your motherboard can control. most boards only have a couple PWM fan ports, and not all of the 3 pin ports may even offer voltage control at all. My Asus p8Z68 has 1 PWM port, 2 s pin ports that offer voltage control but are slaved together, and a third 3 pin port with no control at all.


*aigomorla your fans are probably being controlled based on CPU temperature, which can rise and fall very quickly. As soon as you load your CPU, the core temp will start to rise. The software (or BIOS) watching those temps quickly increase the fan speed to compensate.

 

[Mister_Fix]

ur mixing apples with oranges..

PWM.. 4 pin and regular 3 pin are whole different beasts.

The 3 pin can be controled via motherboard diode temp or software, but its based mostly based on a temp diode on the motherboard.

Now we know how notorious these diodes are in being inaccurate.

The PWM goes on LOAD.
The whole theory behind it is.. your fans do not need to be spinning at value X all the time. The only time it needs to be at value X is when cpu load is Y.

As the Y changes the value X reported tells the PWM to either spin faster or slower though pulse.

What is a better system?

PWM... basically if ur not overclocking.. and your not doing a nuclear reaction on your cpu, pwm wins..

If your overclocking.... your idle temps are massive... with overclocking temps massive... well pwm doesnt help u very much when your fans are already set to 100%.


Basically if your an overclocker... u have high vcore voltages... high idle temps.... u really dont want your fan on low.. or ever to go on low.
Even after a long processor test run... you still want your fan on high for at least 10 min to bring average temp on the sink back down.

PWM wouldnt do that... u would have to recool your processor with low fans which could take longer.

So if ur a avid overclocker.. u probably want to avoid PWM.. or get a PWN fan controller and set your own ranges.
If your not a avid overclocker... PWM would save your ears a few dB. But that can also be ahighly variable by the fans you use.

You. i like you.

 

[BonzaiDuck]

Oh -- I like Aigo, too, but I mildly disagree about this.

Maybe he COULD be correct about the lifespan of fans as PWM. But I've been running PWM fans to cool my CPU in tandem with exhaust for about four years non-stop. I like the PWM option because I can actually control fans according to CPU temperature from a single motherboard (PWM) port, while powering them directly from the PSU -- using a $10 Swiftech PWM splitter.

I find myself using both 3-pin and PWM fans in my strategy, because certain fans can only be had as 3-pin. so exclusively -- my intake fans all use 3-pin ports; my CPU and exhaust fan use a single CPU_FAN PWM port.

Generally I'm more likely to spend extra for higher MTBF PWM fans than for 3-pin units, but whichever it is -- I avoid sleeve bearings.

While it's true that some motherboards (and those I use) will thermally control either a 3-pin or PWM fan from the main CPU an CHA ports, you need to keep track of fan amperage if you run more than 1 3-pin fan from a single port, and you either do your own soldering to wire them in parallel, or you buy an add-on controller for them -- in which case, they're powered and controlled separate from the mobo. I don't like either of those options, but I've done it both ways.

Otherwise, you'll use up all your mobo fan-ports if you use a sufficient number of 3-pin units. Basically, I REALLY like the idea of eliminating power-draw from the motherboard ports if I can CONTROL the fans from the motherboard.

Generally, with no more than five fans in the case, you don't even need the splitter, and you can at least monitor all the fans connected to the mobo ports. If I can reduce the number of fans I need in the case, I'll pick PWM every time, and I'll pay extra for premium fans.

 

[Bryf50]

I'll say this clearer than I did last time. PWM has nothing to do with the method the speed is determined. Its just another way to set the speed of a motor. I've never seen a CPU fan set its speed based on processor load and I don't think that would be a very good way to determine fan speed.

 

[BonzaiDuck]

I'll say this clearer than I did last time. PWM has nothing to do with the method the speed is determined. Its just another way to set the speed of a motor. I've never seen a CPU fan set its speed based on processor load and I don't think that would be a very good way to determine fan speed.

Aigo may have had either a senior moment or he forgot his coffee. When he suggested controlling fans according to CPU "load," he meant according to CPU temperature, and the most likely cause of variation in CPU temperature is CPU load or usage.

I guess my major point of contention was the thought that overclocking leads to high idle temperatures, so you might want to run low-amperage fans at full bore. But with EIST and other features, idle CPU temperatures under good overclocks won't be that much higher: mine, for instance, are about 38 to 39C right now on this (sig) machine. That number might go up or down if I trim the base fan speed in BIOS or my fan-control software, but my exhaust fan is an AP-30 with top-end 4,250 RPM, and it's running now at ~1,450 to 1,500. I think the "floor" I set for these temperatures is around 30%.

It's the loaded conditions and those temperatures needing attention for setting the upper boundary of fan speeds, and it doesn't need to be 100%. I think my AP-30 is set to top out at 3,400RPM with CPU core temperatures >= 70C, and the curve is pretty steep starting at 50+C.

ADDENDUM: I decided to "up" my CPU_FAN setting -- connecting to both CPU fan and exhaust fan. The "beefy" exhaust fan -- a GT AP-30 -- had been running at 1400 to 1500 RPM or something between 20 and 30%. I pushed up the floor (maybe to 40+%?) so the AP-30 is showing 1,800 RPM, and my two highest core temperatures at idle dropped about 3 to 4C.

It's all a balance between noise, your fear of noise (when it might not really increase), and the temperatures. For me -- this time -- I didn't compromise any noise for temperature improvement, even for running the fans a tad faster at idle CPU speeds/temperatures.

 

[aigomorla]

2013!!!

pwm has evolved from those days.

If you look at the older bios's it went on load option of a 4 pinner.
The pwm value was determined by cpu load value.

Auto was a setting for CPU heat values.
Voltage was well voltage...
PWM had 2 options.. One was CPU temp, while the other was based on CPU LOAD values in BIOS.

You could not set a 3 pin fan on LOAD.... while on a PWM you could.
There as no need to get a PWM FAN unless u wanted that CPU LOAD feature, as a 3 pin fan could do variable on cpu temp.
Also when i wrote that statement, PWM fans were NOT as reliable as today. Boards only had 1 PWM header which typically was dedicated to the CPU because they primary used PWM fans on CPU heat sinks.
(this doesnt apply anymore, and now we have boards with multipul PWM headers, and PWM has changed terms for more pulse width of the fans, so they could indeed be clocked lower then a traditional 3 pin fan.)


Like i showed in my video... as the cpu ramp'd up in load, the PWM value would increase in junction to the load values more so then heat values.
If it was on heat values, the change would not have been so drastic due to a LC system keeping the temps on the cpu as stable as possible, and the holding potential of heat being higher then traditional air sinks.

that 4th wire is the pulse modulation cable.
And yes you are correct in that it pulses however the rate at which it pulses is based on the load of the processor.

Example:
http://s125.photobucket.com/user/aigomorla/media/Nadeshiko/MOL012.mp4.html

 

[BonzaiDuck]

I won't contest any of what Aigo says. I built my 2007 system with 3-pin fans -- period. I DID use mobo thermal control features, but I don't remember having any GUI software to do it, and it probably involved only the CPU fan exclusively. I'd used Speedfan earlier -- see below -- but chose not to use it for the Q6600 system.

2011? That was "new age." I could use the GUI software; I could use PWM fans -- for size requirements that were available in PWM.

All that time -- from about 2004 through 2011, I'd been searching for thermal fan control -- on the motherboard, with aftermarket fan-controllers -- any possibility. I actually had SpeedFan working for me on a Springdale motherboard and a Northwood processor. But the add-on controllers either cost more than I wanted to spend, they were part of a "hardware scam" and a rip-off, or just not practical for my needs.

These days (2011 and forward -- to my personal experience) -- It's there. So I'm still surprised to see threads or posts from folks who are naïve about either BIOS features or the proprietary motherboard software.

Take for instance Ai Suite. You don't need "Turbo EVO" or "Digi+ VRM," or "EPU." YOu can USE the sensor-recorder feature. But you only need to use "Fan Xpert" once -- set it and forget it. Or occasionally adjust it. Once done, you don't need to "run" Ai Suite or the fan Xpert feature: the settings take care of themselves. And frankly, you could do it all within the system BIOS. I think ASUS called their "fan-curve" software by different names for different gen boards and different models. But it all has the same essential features.

And as I also said earlier, with these PWM splitters, you only need at most three PWM fan ports and essentially only one, controlling strings of fans from the motherboard without drawing power from it.

So -- yeah! PWM has EEE-VOLVED! And it's about time, I say! About time!

2013!!!

pwm has evolved from those days.

If you look at the older bios's it went on load option of a 4 pinner.
The pwm value was determined by cpu load value.

Auto was a setting for CPU heat values.
Voltage was well voltage...
PWM had 2 options.. One was CPU temp, while the other was based on CPU LOAD values in BIOS.

You could not set a 3 pin fan on LOAD.... while on a PWM you could.
There as no need to get a PWM FAN unless u wanted that CPU LOAD feature, as a 3 pin fan could do variable on cpu temp.
Also when i wrote that statement, PWM fans were NOT as reliable as today. Boards only had 1 PWM header which typically was dedicated to the CPU because they primary used PWM fans on CPU heat sinks.
(this doesnt apply anymore, and now we have boards with multipul PWM headers, and PWM has changed terms for more pulse width of the fans, so they could indeed be clocked lower then a traditional 3 pin fan.)


Like i showed in my video... as the cpu ramp'd up in load, the PWM value would increase in junction to the load values more so then heat values.
If it was on heat values, the change would not have been so drastic due to a LC system keeping the temps on the cpu as stable as possible, and the holding potential of heat being higher then traditional air sinks.