somethings been bothering me about the way you test PWM fans. just by undervolting them, theyre not running as they should be run, or as most people would run them... at 12v, but with pwm control.

would it be impossible to test them at various percentages on pwm control from either a motherboard header with speedfan, or a true pwm fan controller?

i think this information would be more relevant to more people.

I have thought about that myself. I myself don't use speed fan or headers to control fans. I just set them at a specific voltage and let them run.

Is it possible to get a motherboard to set a header to run at different reproducible "pulse width" settings so that you can get it to run different fans all at the same reproducible pulse width settings? That it seems is what would be required to do this.

I would expect that doing this would not generate results that differ from the current SPCR testing methodology if you are talking about the noise generated by fan blades and air current.

It seems like it is possible it might generate different results if you are talking about motor noise. If there is no motor noise, it seems like it shouldn't make any difference. Does that make sense to you?

yep, thats what i was talking about.

This is an issue that arose as PWM fan control for CPU fans on motherboards became prevalent. Now, almost all current motherboards use PWM for the CPU fan header. So it makes sense for us to test PWM fans with a PWM fan controller.

There are 2 basic ways to do this --

1) test for SPL at specific RPM
2) test for RPM at specific SPL

The "% duty" cycle setting simply does not work, as fans have different maximum speeds, so % duty does not give as any clear reference point. The problem is that we have not found any motherboard fan control systems that allow us to dial in RPM or simply set the speed as with a variable resistor. They are invariably tied to CPU temperature and offer varying slopes (or relationships) between fan speed and CPU temp, with some difficult to set fixed speed options.

I went so far as to obtain an evaluation PWM fan controller board (to showcase the fan control chip) from Analog Devices. This board, tied to a PC and driven with dedicated software, has all the flexibility and power to control multiple fans with both WPM and voltage control. Alas, its fatal flaw is a noisy embedded miniature high speed blower fan that is used to cool couple of critical parts -- a transistor and a resistor. The noise makes it impossible for us to use in our fan testing. It would be folly to risk burning out the board after all the effort to get it in hand (it took well over 6 mos of bulldog determination), so I have a project (yet another!! ) to affix the right size/type of passive heatsink that would keep these components cool enough. So far, I have not found anything quite appropriate... tho admittedly, this task is not high on my priorities.

But, I have yet to find any clear evidence that a PWM fan sounds any different when run at the same speed by PWM or voltage control. It is my experience that poor implementation of PWM control can make any fan (PWM or not) sound worse than voltage control. The undesirable result is usually increased buzzing, ticking. The big theoretical advantage of PWM is the ability to run a fan slower than with voltage control (and reduced power consumption) but if this comes at the expense of extra machine noises, few will accept it. I would like to believe that PWM controllers on mobos are designed to work well will the vast majority of PWM fans that come on heatsinks... and so far, this belief has not been challenged by first-hand experience.

So to get back to the original poster's complaint, I think our voltage set speeds do tell what a PWM sounds like at different speeds, and that is the most important thing -- to get a good sense of the overall character of the fan noise at different speeds. Users only have mobo-embedded PWM control for their fans, and that, hopefully, will provide a range of speeds similar to our voltage set speeds.

So that is where those absurdly low Scythe PWM speeds come from.

It would be nice to get a subjective assessment of how well the Scythe's behave at those absurdly low speeds.