Dual Fan CPU cooling

[jonhutson]

Hi All,

I recently decided to step in to the world of overclocking and messing with messing with ones PC.

Whilst sat at work one day I decided to make use of a few samples of P4 heatsink and fans and socket 370 heatinks lying around. In my usual wisdom.
I fancied doing something different.

On my system I previously had the standard boxed intel heatsink and fan that came with the 1.13GHz FCPGA2 256kB PIII that I bought when I built the system. It had long been on my mind to replace this, as from experiance, they always fail early and are likely to have an MTBF of 25,000 hours. A far cry from the 80,000 hour dual ball bearing fans I liked to deploy in to customers systems @ work.

What I had to hand was a spare passive copper skivved heatsink from Cooljag. A free sample that was never to get utilised. Also I had some P4 cooler samples, again all one's I was never going to use.

I decided that what I would do was to take the fans from the two P4 coolers and stick them on top of each other, then on top of the passive. It all looked to fit together nicely. Whilst the Fans are a slightly different height they were both 60 x 60mm.

Some time passed before I got round to doing it but one day I picked up to pair of fans, removed one of the headers and wired them in parrellel. Stacked them on top of each other and on to the heatsink.
First problem I found was the wire from the top fan kept hitting the blades of the lower. A quick application of plastic glue got the wires flush on the top fan and that was that.

I had a dual fan cold CPU.

The questions are. Is there real benefit to this turbine effect? And can it be taken further? Has anyone else deployed this set up and found it to be beneficial? Is there a detremental effect to either fan in speed balancing or will the natually balance?

One oddity is that yesterday MBM software told me fan speed was 5500rpm and today it's reading 11,250 rpm. Is this caused by a cpu speed sensor impedance mis-match? who knows.

I cranked my PC up and under normal conditions the CPU temp followed the ambient system temp, 27°C to 30°C after about 30 mins at full load (when normal CPU temp on the day was 30°C) it only rose to 34°C. These temps were also achieved after I clocked the CPU from 1.13GHz to 1.358 in the BIOS, where it is stable . But I wonder about taking this further with a few CPU voltage changes.

I look forwards to seeing your comments on this one.

Jon

[nixie]

From what I remember using two fans that spin the same way has little or no benifit as the turbulance from the first interferes with the second. However if the fans spin in opposite directions a slight increase in performance can be seen. If the passive coooler is as large as you say it may give better performance with either a 80mm or even 120mm high performance/high cfm fan and make less noise.

[The Coolest]

IMHO a duct would be MUCH MORE beneficial than stacking fans, adding a duct to the side of my case reduced CPU temps by upto 5-6C during full load (F@H). I put a 80mm hole in the side of the case drilled holes for the screws to mount the fan, and mounted the duct on the screws that were sticking out of the back, and duct taped it to the case, so it won't move, its a really nice mod, and can give u better performance, because it gives the CPU HSF a fresh air from outside the case

[deez]

From what I remember from an article 1 fan has to spin twice as fast as the other in order to get any benefit but dual fans wont do much. You would probably be better off with better case airflow. I've had temps go up as much as 10C just by changing the placement of a few fans. Also if you have both fans plugged into the fan header on the mobo it will likely burn out.

But your temps seem good so see how high you can push the CPU. Use prime95, SETI etc... for stability testing and let us know your results.

[jonhutson]

Neither fan will 'burn out' that's simply ohms law, perhaps if each fan was massive and drew a fat wedge of current then the load maybe to heavy for the header. But in this case both fans are pony.

Im using this as my heatsink. copper heastsink (http://forum.oc-forums.com/vb/showthread.php?threadid=101491)

[deez]

I meant that the mobo. header will be more likely to burn out not the fan. Its no problem to daisy-chain fans but I generally go from a 12V lead from the PSU. Better safe than sorry...no reason to replace a $100 mobo for a pair of $2 fans

The heatsink just looks like a generic copper shouldn't have problems there

[jonhutson]

The heatsinks original purpose was along with another one as a passive in a 1U system. Cooljag (http://www.cooljag.com) But it was all to late in the day to be bothering with dual PIII's so instead I'm now working on Dual Xeon's in 1U.

Jon