CFM ventillation requirements and best way to control fan speed?

[citsacras]

I'm getting ready to replace a socket 939 system with an Opteron 170. The old system is in an ancient (in computer years) Lian-Li case and has only a single rear 80 or 92mm fan. Obviously, this is not ideal.

When I build my new system with a 45nm Intel chip, I want to also replace my aging Lian-Li case with something using front + rear 120mm fans. I am leaning toward a Lancool PC-K10B or maybe Sonata III. The most demanding thing I do on my system is gaming, but keeping the noise and power consumption low is an equal priority. I currently use a 9600GT with an Accelero heatsink. I haven't decided yet whether to go with an E8500 or E8600 CPU and would like to overclock if it doesn't impact stability, heat, and noise too much. 4ghz would make me smile, but if I can't hit it without having a wind tunnel under my desk, I'll stick with lower.

So to get to the point, how much CFM capacity do I need to look for in my case fans? The Lian-Li I am looking at has 2 front 12cm fans and 1 rear fan.

Secondly, how should I control the fan speeds? I'm not really sure if I should be using a software or hardware solution. Ideally, I'd like to be able to hook my fans up to the motherboard headers and dynamically and independently control their speed with a BIOS or software profile. Can I do this with an Asus P5Q Deluxe motherboard which has 3 "CHA_FAN#" headers? Or do I need an external hardware solution like the panels that mount in a drive bay and manually control the fan speeds?

Any other advice that will help me build a better, quieter rig is appreciated as well.

TIA,
Jonesy

 

[RallyMaster]

For 120mm fans, I'd say the max before it gets to be loud is around 50 CFM. I would go with cheap but good Yate Loon D12SL-12s for that purpose at $3.60 each sometimes at jab-tech.com. Controlling the fan speed depends on your motherboard to be real honest. I have only a single Arctic Cooling 120mm fan hooked up onto my motherboard on my Opteron 165 setup and I just leave it on 90% which runs it at 1200RPM. I put a D12SL-12 up front to cool off my hard drives and provide a bit of air to my GPU and NB. Speedfan gives me control on two mobo fan headers on my Abit KN8 motherboards and two on my MSI P43 Neo3-F. A lot of the folks prefer the hardware fan controller route but if you want to save some dough, might as well get quiet fans from the start.

 

[BonzaiDuck]

Question: Take a metric ruler and see if there's enough "flat-space" around that 80mm fan-hole in the back of your Lian-Li to cut a 92 or 120mm fan hole -- concentrically or offset for the center of the original hole. I mean ''Is there enough room back there?"

We had an old PC-Power-&-Cooling "plain-vanilla" $60 Mid-Tower ATX case with an 80mm port, and it at least had ample room to widen the hole to 92mm.

IF SO: Do you have a Dremel and cutoff wheels? Or a Radio-Shack $10 "nibbler" tool? At best, if you can neatly mod your case to fit a 120mm fan, it will save you the ducats you'd otherwise spend on a case fitted for a 120mm exhaust.

Here's some wisdom from the ol' Bonzai-Duck-t-ster-Meister.

My E8400 processor is OC'd to ~ 3.6 Ghz -- up from the stock 3.0 Ghz speed. It's 79.7F -- ~80F room-ambient here at the moment, and I'm stress-testing with PRIME95 Blend Test now. The two core temperatures, are, of course, maybe 5C lower than they would show under small-FFT's [CPU-intensive] testing -- right now, about 46C/46C. I'm using a Noctua NH-U12P heatpipe cooler -- a notch better than a TR-Ultra-120-Extreme.

I don't even have a 120mm fan hanging directly on the cooler! Instead, a flexible "accordian-type" ThermalRight 120x120mm duct ($6.95 at Sidewinder Computers) fits to the rear 120x38mm exhaust fan, drawing air exclusively through the Noctua.

Those temperatures leave a LOT of wiggle room per processor longevity, additional over-clocking, thermal throttling -- peace of mind.

I think my thermally-controlled exhaust-fan alone is pulling between 40 and 60 CFM, but at rpm between 1900 and 2,500. Frankly -- I don't hear it -- if at all, it's white-noise from turbulence, and pretty quiet. I discovered the other day that I'd set the motherboard threshold temperature at which the fan moves toward it's 2,600 rpm top-end too high, and it was spinning at 900 rpm with the idle temperatures virtually the same as with a 100% Duty-cycle setting.

Another point -- my strategy -- contrary to myths and popular belief in these forums -- is to pressurize the case by intake CFMs in excess of exhaust CFMs, or at least equal to or greater than. Then, the idea is to direct the intake air filling the larger volume of the case through narrow, ducted apertures (the tower-type heatpipe coolers and their closely-grouped fins) and across the hotter GPU and motherboard components, such that -- and I emphasize here -- the air passed through these narrow apertures is exhausted immediately from the case.

SHEESH!! I haven't even finished my "ducting" project on my latest build at the moment, and you can see my load temperatures under Blend, while I'm not exaggerating the extrapolated and previously measure load temperatures under small-FFTs. I almost don't want to bother ducting the motherboard, but it's a cheap approach and will only improve cooling.

So . . . . I'm not so sure there's a hard and fast rule for CASE CFM throughput. If I could get 60 intake, I'd make sure it was exhausting between 40 and 60. If I could get 100 CFM intake . . . . you can pretty much assume how I'd do it by applying the rule. The real question: the velocity and pressure of air flowing through the heatsink fins and past the motherboard components.