Case fan RPM's and diminishing returns

Control: management of fans, temp/rpm monitoring via soft/hardware

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
user587
Posts: 8
Joined: Mon Apr 13, 2015 5:59 pm

Case fan RPM's and diminishing returns

Post by user587 » Sat Apr 25, 2015 12:42 am

I have a question regarding cooling, and people around here seem pretty cluey so I thought I'd ask.

What I've been experimenting with is how much of a role case fans can play with GPU and CPU temps/fan speeds during gaming. I'm wondering if there's a point of maximum airflow and an incline of diminishing returns when it comes to case fan speeds/CFM.

My current system
Case: CM Storm Trooper - very spacious with a decent amount of ventiatlion
Case fans: 3x Scythe Kama Flow2 900rpm (2 front intake, 1 rear exhaust)
CPU: i7 4770 with CM Hyper 212 push/pull (2x Scythe Slipstream 1300rpm PWM)
GPU: Gigabyte 7970 OC (Windforce cooler)

My current cooling setup:
GPU fans: On curve, set to ramp up gradually from 33% @55°C, maxes out 41-45% @71-73°C.
Case fans: On curve based on GPU temp, 300rpm until 60°C, gradual increase to 900rpm at 70°C.

Tests done so far (max temps, fan rpms):
Baseline (fans ramping to 900rpm): GPU 72-73°C (2025rpm), CPU 47-55°C (940rpm), Motherboard 28-45°C
Fans at 300rpm static: GPU 76°C (2950rpm), CPU 48-62°C (1080rpm), Motherboard 34-53°C
Side panel off: GPU 70-71°C (1850rpm), CPU 40-52°C (880rpm), Motherboard 26-42°C

My question:
As you can see above, turning the fans from 300rpm to 900rpm made a huge difference in the temperatures and GPU/CPU fan speeds. Notable 7°C with the CPU, and 3°C and almost 33% rpms with the GPU. But when I take the side panel off, the differences are less dramatic. This leads to me to wonder, is taking the side panel off a good test to see if you have maximum airflow? Do you think using 1600rpm or 1900rpm fans would make a big difference to my temps?
If I have maxed out, I am surprised that just a few 900rpm fans could get me so close to the point of negligible returns. I'm generally wondering if there a point of saturation with case fans and case airflow, where adding more/faster/better case fans will have little or no effect on the GPU temperature. The way I see it, either putting in the 1600rpm should see some decent returns, or I'm scratching my head as to why the lowest end of the spectrum is more than sufficient. I do in fact have some 1600rpm fans (Scythe Slipstreams) I could put in, but I am thinking I might just sell them while they are still neatly in their packaging.

I hope people have found this nearly as interesting to read as it was to do. If anyone has experience they can add, feel free :D

cerbie
Posts: 46
Joined: Sat Dec 28, 2013 3:21 pm

Re: Case fan RPM's and diminishing returns

Post by cerbie » Sat Apr 25, 2015 9:53 am

Yes, there is a point of diminishing returns, often severely so. You have to exhaust a nearly-fixed number of Watts (cooler devices use less power, but the difference is not big). With a lower bound for cool air (a few degrees above room ambient), any given heatsink will only be able to reach up to a certain cooling efficiency. Once it gets close to that, more cool air, which keeps the delta-T higher, on average, will not help much more, because you're only really cooling air that's slightly warmed.

If your heatsink and fans can do 0.3 deg-C/W, and you pull 150W, the best you can get is a 45C rise, no matter how much cool air you force through its surroundings. Lower temp rises will require replacing the heatsink with a better one, once you start approaching that kind of performance.

Post Reply