Scythe Ultra Kaze 3000 experience

[m980424]

I did a little experience with this 3000rpm fan (120 x 120 x 38mm 45.90dBA / 133.60CFM / DC12V / 0.60A)

Case - Antec P1600
Fans - PSU 14cm low noise fan, CPU fan, GPU fan and rear fan (Kaze 3000)
Q6600 - Stock frequency, stock voltage, intel stock cooler (max speed)

I ran Prime95 Small FFT's and wait until the CPU temperature stabilized (56ºC), and then increased gradually the fan speed in 3/4 minutes breaks.

Speedfan - rpm - CPU - System

20% - ?500rpm - 56ºC - 32ºC
25% - 0700rpm - 55ºC - 32ºC
30% - 0900rpm - 53ºC - 31ºC
35% - 1100rpm - 52ºC - 30ºC
40% - 1300rpm - 51ºC - 30ºC
45% - 1475rpm - 50ºC - 30ºC
50% - 1650rpm - 50ºC - 29ºC
55% - 1800rpm - 50ºC - 29ºC
60% - 1950rpm - 50ºC - 29ºC
65% - 2125rpm - 51ºC - 29ºC
70% - 2250rpm - 51ºC - 29ºC
75% - 2375rpm - 51ºC - 29ºC
80% - 2500rpm - 51ºC - 30ºC
85% - 2625rpm - 52ºC - 30ºC
90% - 2725rpm - 52ºC - 30ºC
95% - 2825rpm - 52ºC - 30ºC

50% - 1650rpm - 50ºC - 29ºC

My question is:

Why did the CPU temperature increased above 2000rpm?
Was it because of the high shear stress in the cooler due to the high flux of air??

Thanks

[jessekopelman]

1 degree difference in a single run experiment of this type is not enough to draw any conclusions from. Chances are if you repeated the test, you may very well get a 1 degree swing in the other direction.

[jhhoffma]

Why did the CPU temperature increased above 2000rpm?
Was it because of the high shear stress in the cooler due to the high flux of air??

Most likely you have a short circuit in airflow. If your mobo CPU socket is too far away from your exhaust fan and PSU, it's possible intake air is going right out through the exhaust and not having enough time to be taken in and used by the CPU HSF when the exhaust fan is running at higher speeds.

Just a guess, I don't really know how your mobo is setup.

Question though, why go with such a thick and high-speed exhaust fan with the equipment you seem to be running? Your system temps seem quite good with that fan at it's lowest speed (though I can't imagine it's very quiet). You could get a better heatsink for your CPU and a lower speed exhaust fan and get very good results.

[m980424]

1 degree difference in a single run experiment of this type is not enough to draw any conclusions from. Chances are if you repeated the test, you may very well get a 1 degree swing in the other direction.

I increased the speed fan gradually until 95% and afterwards reduced immediately to 50%. The CPU temperature decreased 2ºC.

[jessekopelman]

See jhhoffma's post, seems a very reasonable theory. You'd be surprised how wide the area of effect of these large fans is. My case exhaust (120mm) seems to do just as much to exhaust my PSU as the PSU's own 80mm fan, despite the very indirect airflow (separate PSU compartment, so this is all happening from outside the case). You could test this theory by sticking up some sort of simple barrier/duct to make sure as much of the air being drawn by the fan as possible is coming through the CPU heatsink area.

[Farinorco]

Wow, that seems a weird result... but it could be completely logical.

When I saw it, the first thing I thoght was more or less the same thing that jhhoffma has said: the Kaze is stealing the air to the CPU heatsink fan. Specially since the Intel stock cooler, AFAIK, makes use of a fan which airflow is perpendicular to the rear exhaust one... indeed, it may be decreasing the temperature difference it should show even at the speeds showing an improvement. I'd bet that results of increasing the Kaze speed would be much better if the CPU heatsink airflow direction was the same than rear exhaust fan, and even more if the Kaze was that CPU heatsink fan, or the rear exhaust fan with a duct to the CPU heatsink and no other fan on it...

[m980424]

I replaced the intel stock cooler for a fanless scythe ninja rev.B and achieved the following results:

20% - ? rpm - 66ºC - 33ºC
25% - 700rpm - 60ºC - 32ºC
30% - 900rpm - 57ºC - 31ºC
35% - 1100rpm - 53ºC - 31ºC
40% - 1300rpm - 51ºC - 30ºC
45% - 1475rpm - 49ºC - 30ºC
50% - 1650rpm - 48ºC - 30ºC
55% - 1800rpm - 47ºC - 30ºC
60% - 1950rpm - 46ºC - 29ºC
65% - 2125rpm - 45ºC - 29ºC
70% - 2250rpm - 45ºC - 30ºC
75% - 2375rpm - 44ºC - 30ºC
80% - 2500rpm - 44ºC - 30ºC
85% - 2625rpm - 43ºC - 30ºC
90% - 2725rpm - 43ºC - 30ºC
95% - 2825rpm - 43ºC - 30ºC

The Ultra Kaze 3000 is only 0,8cm away from the ninja heatsink.

[Farinorco]

So here you have, some much more logical results:

Now the temperature is drastically decreasing with each speed increase of the Kaze at the lower speeds, and as the temperature is approaching to ambient, is needing more and more airflow to decrease even more.

I think a temp of ~55º is perfectly safe for a Quad*, so I personally would use the Kaze at ~900-1000rpm with the passive Ninja, or the Kaze at 700-800rpm with the Intel cooler**, whichever is quieter.

Or even better, make a duct to the passive Ninja to the Kaze to enforce the whole airflow to pass through the fins of the heatsink instead of going around it, and surely it will show the best results in terms of noise-performance. It doesn't seem difficult to do, and I think that lots of people here do so, so it's easy to find examples in this forum

*EDIT: at least if its a G0 revision with its thermal specification of 72ºC, the B3 has a thermal specification of 63ºC, I think.

**RE-EDIT: Nah, I don't know what the hell I was thinking when I said that (keeping the Intel stock cooler). If anything, add a fan to that Ninja, it would be a 2 fan system too (between exhaust and CPU cooler) but with a much better heatsink, probably a much better fan, and aligned with the exhaust airflow...