Notes on system fan placement and use.

The following points summarize how air flows through and around fans. It's a summary from an article I read by NMB engineering:

0. One rear fan produces the best overall case cooling effect. Combine that with a fresh air duct to the CPU for added effect

1. Fan intake is smooth and laminar in flow

2. Fan exhaust is turbulant

3. Turbulant air has up to twice the heat dissipation ability as laminar flow

4. The turbulant air only extends a few centimeters from the exhaust face.

5. Up to 90% of air flow can be lost due to recirculation (churning) and care must be taken eliminate recycling hot air.

6. Baffles or ducts can be used to block recirculation of air.

7. Air flow will always take the path of least resistance.

8. An exhaust vent should always be at least 50% larger than the fan.

9. Dust accumulation can significantly block cooling so intake fans should have filters.

10. The case should have a slight positive pressure to prevent dust from entering components.

11. Components with critical cooling requirements should be placed close to air inlets. High temperature components should be placed near outlets.

12. Hot air flowing through a fan can decrease its life time by 2/3. It is better to have intake fans and avoid exhaust fans.

13. Avoid placing obstructions in high air velocity areas (like fan filters too close to the intake) this will reduce the noise generated and increase intake velocity. Filters should be mounted at least 2cm from the fan surface.

14. Use vibration isolators on all fans to eliminate mechanical transfer of noise to the case which then acts as a nice amplifier.

15. Use structural reniforcements to control enclosure resonant frequencies. Use sound absorbing material inside the case to reduce noise.

16. Mount a fan on an interior surface rather than an exterior one.

17. Obstructions near fan intakes create more noise than the same obstructions placed near fan exhaust (eliminate intake vents holes and replace with fan guards).

18. Two or more fans in parallel produce less noise than one fan producing the same CFM.

19. Two fans in series increase pressure but only increase CFM marginally in a low pressure environment (like a PC case).

20. Improving air flow within the enclosure can significantly improve cooling. (making sure there are no dead spots and use baffles to redirect air to places that need cooling). Tie up wires, place hot objects above cool ones.

21. Each fan added to a system also increases heat. Six, 5 watt fans produce 30 watts of heat that needs to be exhausted. Minimize fan use. In some cases the heat from an additional fan might be greater than its cooling effect.

22. Each additional fan increases noise, heat, cost and may disrupt air flow.

23. Experience shows that an empty enclose reduces air flow by 5 to 20%. A densely packed enclosure reduces air flow by 60% or more.

24. Finger guards produce small losses and small increases in noise so use them.

25. Fan adapters cause CFM loss. The more sloped the sides the greater the air flow loss. In theory a 10% slope produces minimal loss for an exhaust. But a 20% slope is fine for air intake. (ie: to go from 120mm to an 80mm HSF the duct must be 8" long for no air flow loss...at 5" you see a 50% loss).

Interesting.

If turbulent air is better for most heatsinks than laminar air, this would explain the "reversal penalty."

In which case, if you wish the HS fan to "suck" not "blow", one should disrupt the air coming into the heatsink or going through the heatsink. A wire grill might help do this, or similar concept.

the wesson

By the way, I vote for stickey! (with credit to NMB naturally as appropriate.)

the wesson

Definetly should read something like this before mounting all your fans.

Sticky? I'm up for it.

Supporting the Sticky Movement

BTW, nice thoughts on air disruption, TheWesson, but I think to disrupt air to the point where it makes a difference would require more than a grill in the way. I'm kinda leaning toward impossible with the fan in front of it pulling towards and through the heatsink

Thanatos_2005 said:

Supporting the Sticky Movement

BTW, nice thoughts on air disruption, TheWesson, but I think to disrupt air to the point where it makes a difference would require more than a grill in the way. I'm kinda leaning toward impossible with the fan in front of it pulling towards and through the heatsink

Click to expand...

Yeah ... perhaps ...

You made me think of the kind of loose-weave metal filter/mesh that people have on the fans that live over the stove. Hopefully both not a big CFM loss and also plenty of turbulence.

Another factor reducing efficiency of "suck" mode could be the lack of a counter-flow setup. Ideally, a cooling stream hits the less-warm parts of the hot surface first, and then proceeds to the warmer parts. Thus, the cooling stream can pick up more heat along its entire course, rather than hitting the hottest part first, and then being as hot as (or hotter than) the extremities of the fins, and not picking up any more heat there.

In "suck" mode, the air stream will probably be hitting the hottest parts first. Not so good.

the wesson

sticky!

Eh, what the hell. STICKY

Can you believe that one rear fan produces the best average cooling for a PC case? Read here and find out:

http://www.short-media.com/review.php?r=230


Though I bet if you add a fresh air duct your temps get even better.

i vote sticky! CRAP MAN! that stuff is EXTREMELY useful! i very much appricate this thread...ESPECIALLY NUMBER 21! wow! i never anticipated something like that.

10. The case should have a slight positive pressure to prevent dust from entering components.


does that mean the intake fans should be set to a higher speed than exhaust?

Hmm, and what about water cooling, kinda needs to modify the whole 1 fan is enough theory. I wonder what NMB would say on exhausting heatercores, etc and how to properly exhaust when using those. Is it best to blow instead of suck to get more turbulent flow? so confused when not in average PCs. What about external fans blowing on gpu? Would these fans be good because the produce much more turbulent flow on GPU, or bad because they disrupt airflow? questions questions

r1 said:

10. The case should have a slight positive pressure to prevent dust from entering components.

does that mean the intake fans should be set to a higher speed than exhaust?

Click to expand...

It means the total input CFM should be a bit larger than the total output CFMs. If all the fans have the same CFM rating, then yes at least one intake should have a higher speed. Normally people have different kinds of fans so you have to figure out what the total in and out is.

To do that, take a fan's maximum rated CFM and divide it by its maximum rated RPMs then multiply that result by its current RPMs. This gives you the current CFM flowing for that fan. Add all the intakes and subtract all the outputs to get the final case CFMs. If that number is negative, slow down some outputs or speed up some inputs until the equation gives you a net positive inflow of air. I guess 20-40 CFM would be a good positive number. Remember that fan speed varies by plus/minus 5% so a bit extra on the positive side will make sure the intake is always more. Don't forget to include the PSU fan in the calculation.

BUT, note #0. One rear exhaust fan cools a case best overall.

So I think a good configuration would include a fresh air intake duct with fans mounted on both sides of the duct AND a filter to prevent dust accumulation (dust dramatically kills HS cooling ability). Those intake fans could then provide the positive pressure while the rear output fan would be set slightly lower.

Reversing that configuration and producing an exhaust effect without the rear exhaust fan might be just as effective with a reduction of one fan.

I like a fan blowing on my HDs but it clearly isn't necessary for proper operation and probably just adds to the case temperatures because of the fan's own motor heat.

@/2ct!< said:

Hmm, and what about water cooling, kinda needs to modify the whole 1 fan is enough theory. I wonder what NMB would say on exhausting heatercores, etc and how to properly exhaust when using those. Is it best to blow instead of suck to get more turbulent flow? so confused when not in average PCs. What about external fans blowing on gpu? Would these fans be good because the produce much more turbulent flow on GPU, or bad because they disrupt airflow? questions questions

Click to expand...

Water cooling may or may not use fans, but those are external anyway and so remove heat in a separate process. Here we are talking about case fans removing heat.

A sucking HSF exhausting to the outside might well be just as effective as a single rear exhaust fan. It certainly produces cooler case temps. Of course reduced HS turbulance and increased air temperature entering the HS make it hard for sucking to beat blowing at HS cooling.

As far as external GPU fans, the tests showed that more fans make case cooling worse not better. Of course if your GPU needs more cooling to work then case temperatures aren't the only consideration.

orion456 said:

The following points summarize how air flows through and around fans. It's a summary from an article I read by NMB engineering:

0. One rear fan produces the best overall case cooling effect.

Click to expand...

don't do this man. one web site conducts a short experiment which shows one rear fan doing as well or a tiny bit better than other setups (under some circumstances) and all of a sudden it's gospel to you?

this is the same experiment which was getting (at best) 40C mobo temps, mind you.

Maybe they had huge cable tangles or some dang thing. Anyhow, one result is not gospel.

I think most people here would disagree from their experience - a single rear case fan is not the best possible solution.

the wesson

TheWesson said:

don't do this man. one web site conducts a short experiment which shows one rear fan doing as well or a tiny bit better than other setups (under some circumstances) and all of a sudden it's gospel to you?

this is the same experiment which was getting (at best) 40C mobo temps, mind you.

Maybe they had huge cable tangles or some dang thing. Anyhow, one result is not gospel.

I think most people here would disagree from their experience - a single rear case fan is not the best possible solution.

the wesson

Click to expand...

Hardly gospel, but it was a pretty interesting test setup. The result simply says one fan was best over all. It didn't produce the best cooling in most cases but it sure came close enough in every case (with lots of static intakes available). As usual, experimental results often runs counter to what everyone thinks so I don't much worry about what everyone would say, only what the experiment reveals.

This result mirrors my own experiments which showed that a single rear exhaust worked very well at removing case heat (combined with a PSU exhaust fan of course) and my hunch that all those extra fans just add heat with very little gain. In fact more than a few times adding extra fans to my case actually increased temps, which surprised me. In fact I wrote about my experience on this forum when someone mentioned they got higher temps when they added two intake fans. I had been planning to add fans too but when I tried it out, my temps when up as well. Those extra fans are now just empty fan bays in my case.

I do prefer to add an extra intake fan blowing on my HDs and trying to make case pressure positive. However, I am satisified that the extra intake fan makes no difference to CPU cooling at all.

And I note that my Dell machine uses a single rear exhaust fan with a shroud over the unfaned HS. The result is extremely quiet and a pretty cool case and HS as well, which also surprised me (although it is only a 1.7 gig celery). Not that Dell is any cooling champ, but still the idea works for them too.

So yep, I think the result is pretty valid and if more people checked carefully they would find all those extra fans are pretty much doing little but making noise. Less fans, more static filtered intakes and a fresh air CPU duct produce very good results that easily rival and often surpass the 10 fan jobs.

I currently have two machines, one with a HSF exhaust vent and two filtered front intake fans on low blowing over my HDs. The rear fan holes are filtered but open. The other machine has one rear exhaust fan on medium speed, one stock intel HSF sucking on a duct and a third filtered fan blowing into the duct which overcomes the resistance caused by the filter. Both machines (p4c's, 3.6 ghz & 3.0 ghz) have temps lower than 42c on full load. So one exhaust works well.

But to appease your gospel concerns I removed the underlining and added in the need for a fresh air intake.

ah no offense meant.

the biggest gain I got when first modding the case was to open up the front airway, with one exhaust fan.

after that, 1 intake fan does something. But not strikingly much (2 or 3 C?). And a 2nd intake hardly helped.

But I think the point should be not that "1 exhaust fan is best", but instead that "additional fans aren't necessarily useful" - depending on placement, airflow, etc.

Physically, increasing CFM through case by 100% should drop case deltaC by 50%. But extra fans won't always increase CFM throughput, or get the air to where it's needed ...

Maybe in the experiment you linked to, the single rear exhaust fan was just the best way of getting air in the region by the CPU + mobo sensor. Doesn't mean single exhaust fan is best for case cooling.

...

By the way, most 'standard' case fans are trivial in wattage. Like 2 watts. It's the Delta monsters which are 12 or 20 watts.

the wesson

Vote sticky 04!
However, I would like to see much of this stuff tested - Generally the oc.com community does a much better job at testing things then anywhere else that I know of.

12. Hot air flowing through a fan can decrease its life time by 2/3. It is better to have intake fans and avoid exhaust fans.

Click to expand...

Wow. Now that would be a kick in the teeth. I wonder how true that is, especially if you have enough airflow to keep the air in the case from getting to hot to begin with.