Undervoltings vs. Reduced Capacity Fan

bgavin · Post by **bgavin** » Sat May 10, 2008 9:24 am

Is there a preference between undervolting a more vigorous fan vs. using a lesser fan?

Example: Panaflo 120x38 fans. One model spins at 2750, another at 2100 rpm. I can undervolt the faster fan down to 2100 rpm, but I don't know which fan is the better performer.

If the same, then why doesn't everybody simply buy the faster/fastest fan, and undervolt as needed?

I realize there is a point of no return, i.e. insufficient voltage to start the fan. It would be silly to buy a 2750 rpm fan, when 900 rpm is the desired result.

VanWaGuy · Post by **VanWaGuy** » Sat May 10, 2008 11:41 am

You can see in the very recent Scythe review that basically what you are saying works, although when looking at the details, one of the fans performed a little better than the others at the same speed.

I think one factor would be convenience. If you know you want 900 rpm or 1200 rpm, then buying one you can just plug in at 12 volts is just easier. On the other hand, the higher RPM fan undervolted does give you more flexibility. If you later need more RPM, you could undervolt a little less.

bgavin · Post by **bgavin** » Sat May 10, 2008 12:38 pm

I completely agree with buying the right fan for the job, if known in advance.

I'm testing the Zipang cooler on a hot running Pentium 530 (prescott) board. The stock fan is woefully inadequate at cooling the VRM, so I'm experimenting. The Panaflo 120x38 model 12U does the job, but at 2750 rpm with a lot of racket. At 1200 rpm, the Evercool RSF-14 Red Scorpion is a compromise between the Panaflo and Stock.

I'm curious to know how undervolting the thicker (120x38) Panaflo down to 1200 compares to the RSF-14 (140x20) fan designed to run at 1200 rpm.

Bluefront · Post by **Bluefront** » Sat May 10, 2008 3:33 pm

I don't think there's any way to know the exact answers to those sorts of questions. You'd really have to find somebody with those two fans who has done that sort of testing.

I usually try to bench test the board and cooler on a warm day.....then try to determine how many rpms are needed to keep the cpu at a comfortable temperature during bench-marking. Most 120mm fans push about the same airflow at similar rpms. So if the fan I'm testing is maxed out (for instance), and the cpu temp is ok, I try to use a fan that can run about 25% faster if necessary.

I'm new using 140mm fans......so trying to compare one to a 120mm fan with regard to airflow is somewhat guess-work right now.

Cistron · Post by **Cistron** » Sat May 10, 2008 3:45 pm

Bluefront wrote:I'm new using 140mm fans......so trying to compare one to a 120mm fan with regard to airflow is somewhat guess-work right now.

If the airflow is proporational to the fan surface area, this would imply a ~35% difference.

Bluefront · Post by **Bluefront** » Sun May 11, 2008 12:55 am

Yeah but..... There are so many variables in this situation that a paper/spec comparison may be way off. I suspect an actual hands-on comparison, would be your best bet for now anyway. There are few 140mm fans available right now, and less actual reviews/tests.

I suspect I have more different 140s than anybody....

2 different 140x20mm Areocool models.

2 140x25mm Yate Loons

1 140x25mm Scythe.

1 140x25.. Evercool.

VanWaGuy · Post by **VanWaGuy** » Sun May 11, 2008 7:48 am

Cistron, did you subtract out the hub area?

Cistron · Post by **Cistron** » Sun May 11, 2008 3:34 pm

VanWaGuy wrote:Cistron, did you subtract out the hub area?

No, my calculation was as easy as 70^2/60^2=1.36 (pi cancels out anyway). I suppose the bigger the fan the better, since the hub can stay at a similarly small size. However, it also means there is less air-pressure as this is related to the rpms (if I read correctly somewhere around here).

dhanson865 · Post by **dhanson865** » Mon May 12, 2008 5:25 am

Cistron wrote:
VanWaGuy wrote:Cistron, did you subtract out the hub area?
No, my calculation was as easy as 70^2/60^2=1.36 (pi cancels out anyway). I suppose the bigger the fan the better, since the hub can stay at a similarly small size.

That just depends on how lazy/cheap the manufacturer is. Hub sizes do vary from fan to fan (see Scythe Slipstream vs Nexus 120mm) but there is no guarantee that the bigger fan will have the same size hub.

Anyway the main reason I buy higher speed fans is flexibility/headroom for higher temps as mentioned above but occasionally you get the situation where the choice is:

Nexus 120mm 1000RPM or so $15
Yate Loon 120mm 1350RPM $3

S&H costs often prevent you from getting such comparisons on a regular basis but when the opportunity exists for me to buy in combination with something else to save on shipping I'm plenty happy saving a few dollars and slowing down the cheaper fan.

Felger Carbon · Post by **Felger Carbon** » Mon May 12, 2008 6:01 am

Cistron wrote:I suppose the bigger the fan the better, since the hub can stay at a similarly small size. However, it also means there is less air-pressure as this is related to the rpms (if I read correctly somewhere around here).

For a given fan, the pressure is proportional to the RPM squared. Double the RPM, four times the pressure. For a fan of a different size, you'd want to look at the fan blade tip velocities, not RPM, to make pressure comparisons. And of course, fan blade geometries (if different), get into the act...

jaganath · Post by **jaganath** » Mon May 12, 2008 6:27 am

For a fan of a different size....

according to fanlaws pressure varies with the square of the diameter. of course a larger fan rotating at the same speed as a smaller fan will also have a higher tip velocity, so this is not entirely unrelated.

bgavin · Post by **bgavin** » Mon May 12, 2008 6:46 am

Felger Carbon wrote:For a given fan, the pressure is proportional to the RPM squared. Double the RPM, four times the pressure.

Is this rule of thumb reliable enough to calculate static pressure when undervolting and reducing the rpm?

If so, this would let me accurately compare pressures from a hot fan such as the Panaflo 120x38-12H undervolted to 1200 rpm, with another fan running at 1200.

It is my understanding that sufficient head pressure is required for high impedance loads such as Zipang or SI-128.