Getting good temps with reversed fan on HS

TheWesson · Post by **TheWesson** » Fri Jun 11, 2004 12:41 pm

I tried something recently which surprised me.

Background: I have a Vantec Aeroflow which is adapted to an 80mm fan. The 80mm fan blows the "wrong way" - sucking air from the heatsink. This assembly is enclosed in a duct expanding to a 120mm side blowhole. The duct has a 120mm L1A@5v exhausting at the blowhole.

Anyhow, the surprising thing was that I improved temps by 3C by adding some PUSH to the heatsink.

I did this by adding one 80mm L1A to the case (total intake = 1 low-flow Antec 120mm@7v, one 80mm@12v) and blocking all the case exhaust - taping up the little holes. Thus, the case can develop "positive pressure." (Indeed, a little breeze comes out of any hole.)

Temp deltas (case to heatsink) before: 22 C idle and 26C load.
Temp deltas after: 19C idle and 23C load.
No change in case temp from this.

Anyhow it seems as if doing this has almost eliminated the "reversal penalty", which seemed to be ~4C for this heatsink with this CPU in the past.

I'm getting a .25 estimated C/W rating which is not at all bad - like the Zalman 7000 at full power. There is no discernible fan motor noise from the whole setup, but some whooshing.

New theory: The "reversal penalty" comes principally from the fan trying to pull FROM resistance. If there is some push on the other end, this penalty is overcome.

the wesson

silvervarg · Post by **silvervarg** » Mon Jun 14, 2004 6:27 am

I think this is a classic practical example of positive ns negative case pressure. As you increase the case pressure your exhaust fans will push out more air and increasing the cooling.
The hard thing is to know what this translatest to in degrees celcius. For you it was ~3C.

The big question is: Why?
Do you really want to add lots of noise to gain 3C when you allready had great temps?
I would prefere to remove all the intake fans and possibly put a muffler on the exhaust fan to reduce noise even further.

peteamer · Post by **peteamer** » Mon Jun 14, 2004 7:32 am

TheWesson, can I just clarify, by :

Temp deltas (case to heatsink)

I take it you mean difference between case and heatsink temps?

And also :you've got a 120mm fan at the blow hole exhaust point and an 80mm fan right at the heatsink?

Are both nec.? what happens to temps if you remove the 120mm?

I'm thinking the temp might rise slightly ~2/3C but you would loose the noise of a fan right at the edge of the case, albiet a ~quiet one.

These are only curiosities, not critical, more analitical.

Also out of curiosity, if I'm right about the temp. deltas, what are your absolute CPU and Case temps?

Cheers and thanks for letting us know what you've found so far.

Regards

Pete

TheWesson · Post by **TheWesson** » Mon Jun 14, 2004 11:38 am

peteamer wrote:TheWesson, can I just clarify, by :

Temp deltas (case to heatsink)
I take it you mean difference between case and heatsink temps?

And also :you've got a 120mm fan at the blow hole exhaust point and an 80mm fan right at the heatsink?

Are both nec.? what happens to temps if you remove the 120mm?

I'm thinking the temp might rise slightly ~2/3C but you would loose the noise of a fan right at the edge of the case, albiet a ~quiet one.

These are only curiosities, not critical, more analitical.

Also out of curiosity, if I'm right about the temp. deltas, what are your absolute CPU and Case temps?

Cheers and thanks for letting us know what you've found so far.

Regards

Pete

Cheers.

When I had the 120mm main duct fan at 7v, it worked just fine without the 80mm. I think the converse - 80mm without the 120mm - is worse.

Case is 27. Ambient is 23C - slightly above 70F. CPU temp on load (90W CPU) is 50C with the setup below. I am constantly tweaking though - I had no low noise 120 mm intake so now I am substituting 2 80mm 7v.

the wesson

Post by **Rusty075** » Mon Jun 14, 2004 12:32 pm

What you've "discovered" is that fans have a P/Q curve.

As the air pressure diffential (P) increases, the airflow (Q) decreases. You can see the graphs for the L1A's here: Panasonic FBA08A Spec sheet

By blowing air into the case you've decreased (or even reversed) the pressure differential, resulting in more airflow out the exhaust. Running fans in series, as you are with the exhaust duct, has the effect of "steepening" the P/Q curve.

The downside is the noise:

You're now using twice as many fans, producing twice as much sound energy, to gain a trivial decrease in temperature.

A quieter solution may be to reduce the P passively, by opening up the intake as much as possible. Restricted intakes results in lower in-case air pressure.

TheWesson · Post by **TheWesson** » Mon Jun 14, 2004 4:25 pm

Yes and maybe what I've discovered is that trying to pull *from* impedance is worse than pushing *into* impedance.

Next Q:

What is better:

Fan -> Heatsink -> Fan -> Room
Fan x2 -> Heatsink -> Room
Heatsink -> Fan x2 -> Room

Fan x2 is two parallel fans.

I suspect the first may be best where the heatsink has high impedance. Plus, that option allows me to mount multiple slow-flow fans in the case to send air through my heatsink.

I should try option two: fans building pressure in the case but nothing (or very little) pulling from the HS. I suspect that will work pretty well too. It can be made quiet as well because I can bury case intake fans inside the case.

It also might allow me to dispense with the PS fan.

the wesson

PS twice the fans is only 3 dB more. That is nigh-trivial too.