How To: Properly Plan And Pick Parts For An Air-Cooled PC, Part 2

[Arbie]

I finally found a very good way to cool a hot CPU in the "PSU Base-Mounted; CPU Cooler Vertical; Rear & Top Vents" configuration you describe. The only important difference is that my Coolermaster CM-690 case has two top fan ports instead of the one in your example. I put an exhaust fan in the rearmost of these top panel ports, augmenting the rear panel exhaust fan. I put an intake fan in the foremost top port; there it blows cool air down to the DRAM and the intake area of the CPU fan. Both top panel fans were 120mm Scythe Kama Flex with a PWM control range of roughly 300-1600 RPM. These have fluid bearings for long life in a horizontal position where sleeve-bearing fans would fail early.

The key to success was the Akasa AK-CB002 PWM splitter ($4 at PerformancePCs). This lets you control up to three PWM fans from the single CPU fan PWM (4-pin) header on the motherboard. It has a large Molex connector so that all the fan power comes directly from your PSU. I used this PWM splitter to power the CPU fan and the two top panel fans. The rest of the setup is as you show: a front panel and bottom panel intake, and rear panel exhaust. These three fans run full speed at 1100-1200 RPM.

In normal operation the system fan / air noise is audible, but unobtrusive. Under heavy load the CPU fan spins up and brings the top panel intake and exhaust fans with it, providing plenty of air even for an overclocked quad-core. And this arrangment maintains positive pressure because there is always significantly more intake than exhaust capability. This prevents dust buildup on the internals. Of course all the intake fans have filters, and I chose a good CPU cooler. If a hot graphics card needs even more air, add a side-panel intake fan at 1100-1200 RPM.

The above scheme is reliable, relatively cheap, and almost fool-proof, using only the PWM CPU fan control found on any quality mobo. It will give you a quiet but thermally very capable machine.

 

[Arbie]

I will add that if your case does not have two top ports (vents), but does have a side panel port, you can use that as the PWM-driven intake. So this scheme can be applied to most of the more modern (bottom PSU) enthusiast-class cases. Also, I did try having both top panel PWM fans be exhausts but that was definitely inferior. It may seem strange to blow air down from the top, but in this configuration it works well.

 

[RedJaron]

I also remember watching a program on NatGeo on submarines. Apparently modern day subs use propellors with a large number of thin blades. This basically avoids pressure diferences between water in front of the blade (in the direction it's spinning), thus prevents excessive low pressure in the region through which a propellor has passed. This avoids small bubbles of steam from building up and then popping, causing noise.

Seeing that air is a fluid too, i wonder if a similar principal will apply to it. Of course, there wont be any air bubbles popping in air, but i'm wondering if turbulence has any role to play, since i assume more blades=less turbulence. not sure though.

The bubbles are called cavitation, and yes they make noise. You may have noticed that many companies make a marketing point out of nine-blade fans over seven blade varieties. This is a similar principle. More blade edges closer together ( up to a point, anyway, ) generally makes for smoother airflow and less noise.

 

[RedJaron]

I read years ago that thinner the layer of thermal paste, the better.

Well...not THIN as in barely there thin but just enough to make lettering opaque, not completely covered.

I thought the paste is just to fill in cracks and imperfections on the heat sink and the surface of the processor...

That is correct. Metals in the CPU heat spreader and the cooler base have better thermal conduction than the paste itself, but the paste has better conduction than air. If you have tiny air pockets between the CPU and the cooler, they act as insulators and heat builds up. If you have too much paste, then heat isn't transferred to the cooler as well as it could be, again resulting in minor insulating characteristics.

Ideally you want just enough paste to bridge any tiny gaps between the heat spreader and the cooler, then let the cooler do its job.

 

[RedJaron]

There are a few things I dispute.

For the negative pressure case:

Amplification of natural convection

While the top fan blowing upward does indeed support (rather than oppose) natural convection, this is not a trait of a negative pressure setup. This is a trait of proper fan orientation (cover in the last article). Negative pressure means fewer particles available to transfer heat to. Thus negative pressure opposes natural convection. Conversely, positive pressure with a more particles available to transfer heat to supports natural convection. A good example of a properly oriented positive pressure system is the Silverstone Raven 2/ Fortress FT02.

Someone already commented that convection and conduction are not the same thing. Convection is simply the movement of molecules within the fluid. Conduction is the actual heat transfer to those molecules. Less pressure actually means less resistance to the molecules to be able to move about, so yes, negative pressure can help convection.

As for opposing the conduction itself, that's debatable. Yes, you may have fewer particles to conduct the thermal energy, but then again Boyle's Law states that the remaining gas will suffer a temperature drop due to the drop in pressure. Newton's law of cooling states the rate of heat loss is proportional to the temperature differences between the two bodies, so cooler air will help cool the internals faster. The reverse is also true in that higher pressures will increase conductive particles but also increase ambient temperature.

However I don't feel the urge to do all the math to see if the an ever so slight difference in air temperature will offset the marginal conduction difference of gas density. Let's be realistic, we're not talking about huge differences in pressure or temperature here. I'd be shocked if you could sustain even a 1 psi difference in either a positive or negative pressure inside a computer case, let alone a 1°F difference. You get more difference than that in 2000 feet of altitude, so why aren't computers in Denver burning out every day?

Compatible with all direct heat exhaust graphics cards (the ones that blow dissipated heat out of a vent built into the shroud)

Again, the more of a vacuum you create, the fewer the particles available, the worse the heat transfer. As a side note, fans are noisier in a vacuum as well, but I'll let you look into that if you're interested.

And again, we're not talking huge vacuums here either. I'd also be interested in how you figure a fan is noisier in a vacuum considering there are fewer particles to carry the waves. Consider how sound is conducted not only faster but clearer when conducted through liquids and solids.

Augments the cooling performance of downward-facing CPU heat sink fans

You'll have to explain this one to me, but I suspect this is another fan orientation issue rather than a positive vs negative airflow issue.

It's easier to direct airflow with a negative setup than a positive one, particularly within irregular spaces. Downward CPU coolers basically recirculate whatever air is around them. A negative pressure setup will help evacuate that air, especially in the diagram they presented where the two exhaust fans are quite close to the CPU.

Without a dust filter, dust gathers inside the case

Even with a dust filter, you have to understand that most cases aren't air tight. With negative pressure air will be drawn in through every crack, bringing dust with it.

And even with positive pressure, some dust will build up inside. I think you're taking this point a little too literal.


As for the rest of your arguments, the images in the article are what they're using as an example for their statements because those types of cases with those fan configurations are about the most common available to the average consumer. Keep in mind, this article series isn't intended for the computer expert who uses the latest custom cases. This is meant for the beginner enthusiast who is more likely than not using fairly standard parts, meaning a case with fan mounts on the front, top, and back and maybe on the side. You can't "put the air right where you need it" if your case can't mount a fan there ( and I don't think the beginning enthusiast is willing to cut up and mod a case to do so. ) I love my Raven 3 case, sure, but most newcomers aren't willing to spend more than $60 on a case, let alone $150+.

 

[mapesdhs]

[citation][nom]Erelyes[/nom]An OK article. I was surprised the "line" method of applying paste is not explained, as this is what is recommended by many manufacturers (http://www.arcticsilver.com/pdf/appmeth/int/vl/intel_app_method_vertical_line_v1.1.pdf). ...[/citation]

In this case, not so, and your link proves the point, the line method is only for certain models of CPU,
and that link refers to Intel chips. Some Intel chips though are best done using the dot method
described here which is indeed also the method recommended by AS for Athlon64 X2 CPUs. In some
cases, AS suggests the surface spread method; it varies.

For this article though, the correct method was used.

Ian.

 

[RedJaron]

I also think you are giving a little too much credit to convection. A 50rpm change in fan speed achieves more than convection every could. Convection helps passively cooled systems; in anything with a fan, convection is overshadowed.

Yes, a fan can overpower the direction of convection. But why have them fight when you can use them together?

 

[RedJaron]

Good article, but negative pressure ? Anyone commenting on this can simply disregard it.
You want to force cool air in and pull hot air out this is not negative pressure

Care to back that up on why a negative pressure system won't work?

Seriously people, stop treating positive/negative like you've got a hermetically sealed case with 10 psi pressure change. With all the differences in case sizes, shapes, fan mounting, fan speeds and CFM, how can you say one or the other is superior in every situation?

Basically, it goes something like this: Positive pressure is great if you've got lots of fan mounting options to direct the air exactly where you need it ( as in right over cooling fins. ) Just make sure the back end of that directed airflow has a ready exit or you can develop eddies that keep hot air recirculating in the case. Negative flows can be better if you've got a stuffed case where all the components make a lot of hard angles for air to flow around. The downside here is that air will take the shortest, most direct route toward the vacuum, which may or may not have it pass over heat sinks.

But really, no one uses a pure positive or negative cooling system anyway. Everyone will have both intake and exhaust fans, so that even if you have a net positive or negative pressure, you can still route the air just as you need it without any problems.

 

[ojas]

[citation][nom]RedJaron[/nom]And again, we're not talking huge vacuums here either. I'd also be interested in how you figure a fan is noisier in a vacuum considering there are fewer particles to carry the waves. Consider how sound is conducted not only faster but clearer when conducted through liquids and solids.[/citation]

While i see sense in most of your points, i think JP has a point here. I don't know if he's thinking what i'm thinking, but what i think he means is that:

A fan is noisier in a vacuum because of the whine it emits because of a pressure change. Go to your bathroom, leave the door open and turn the exhaust fan on. Note the noise. Now close the door. See what i mean? The fan seems to struggle to spin (unless your door has enough space to let air in)...and it creates more noise...i don't exactly know why, but i think it's a pressure thing. You suddenly lower the pressure inside the bathroom and the exhaust fan has to fight the atmospheric pressure, which it sort of fails at. Makes more noise too, but that's because of the stress on the motor i think.
Even though you have less particles, but the fan frame also makes more noise now because it's picking up increased vibrations from the fan motor. That adds to your contact surface with the particles.

In the computer's case, this vibration would be transmitted to case body too, which is generally steel, resulting in that loudspeaker effect or whatever you want to call it. Diaphragm effect seems more appropriate...

Cooler master specs their 90 CFM 120mm fans as creating 3.04 mm H2O of pressure. I really can't make much sense of that, i wish it was in atm, easier to get an idea. Will have to google or look in a book for mm H20 to mmHg conversion...regardless, doesn't seem too big a number.

 

[mmaatt747]

Nice article. It made me realize that I used too much thermal paste when I recently built my new computer. However, the temps of my CPU are well withing the acceptable range even when playing BF3 on high settings.

However, I do intend to overclock my i5 2500k soon so I may remove the CPU cooler and reapply with a better thermal paste (and proper amount) than what I had used in the beginning.

 

[TopGun]

I think I became more confused from this article and the reactions to it in the comments.

I have an Antec 300 and an Antec 600 (one of which I plan to use for my next build). I have a ZALMAN CNPS9900MAX-B heatsink. I see that most orientations have the heatsink blowing out the back of the case...is there a specific reason for this? Wouldn't it be more "natural" to position it to blow out the top of the case? Maybe out the back helps with directing the airflow better because you have the natural heat rising that still goes out the top? Maybe out the back draws more cool air from the front than it would from the bottom (directly above the GPU)?

I also have no idea about whether to try for a positive or negative airflow. I'm guessing that with the 2 120mm intake fans and the 120mm rear + 140mm top (1 200mm top in the case of the 600) it would be negative pressure? I was considering adding a 120mm side case intake fan to help with cooling the GPU and possibly modding in a bottom 120mm intake fan between the PSU and the hard drive cage. I'm guessing this would create a positive pressure airflow? Would this be overkill? From the article it seems like negative pressure is recommended...so I'm confused about what I should do.

One last thing...there are tabs about 1/2" above the floor of the case that the PSU will rest on. Will the approximately 1/2" gap between the floor and the PSU allow enough air so I can position the PSU face down or should I mod the case (I have no issues with cutting it up) with a cutout and filter under the PSU (or should I just face the PSU up drawing in (likely?) warmer air from under the GPU)? Will either of the face up or face down options with no bottom hole cut out contribute more to a negative pressure airflow?

Am I making too big a deal out of this and it probably doesn't matter much what I do?

 

[RedJaron]

A mm H2O is the same as a kg / m^2, or 9.81 Pascals, so 3.04 mm H2O is less than .005 psi ( basically, you can breathe with more force, admittedly the fan can sustain that pressure longer than you can. )

Decent point about the bathroom fan, ojas, and well illustrated. But I'm not sure it carries over. Between the drywall, topping compound, paint, and sealant, I'd wager a bathroom is better sealed than a computer case and the fan much more powerful too. Between all the seams in the case, venting mesh, etc, I don't think you could reasonably create a vacuum that would starve a case fan and make it work against itself ( unless you have every fan as an exhaust, in which case you've got bigger problems than just extra fan noise. )

 

[ojas]

[citation][nom]RedJaron[/nom]A mm H2O is the same as a kg / m^2, or 9.81 Pascals, so 3.04 mm H2O is less than .005 psi ( basically, you can breathe with more force, admittedly the fan can sustain that pressure longer than you can. )Decent point about the bathroom fan, ojas, and well illustrated. But I'm not sure it carries over. Between the drywall, topping compound, paint, and sealant, I'd wager a bathroom is better sealed than a computer case and the fan much more powerful too. Between all the seams in the case, venting mesh, etc, I don't think you could reasonably create a vacuum that would starve a case fan and make it work against itself ( unless you have every fan as an exhaust, in which case you've got bigger problems than just extra fan noise. )[/citation]

Ah...thank's for the pressure conversion, been too lazy to look it up. That's tiny though

I agree with you, a bathroom is much better sealed than a computer case, and the fan is far larger and more powerful. I just wanted to draw a comparison

You're probably right though, a variation of 0.005 psi shouldn't really matter too much...and even if does make a difference, perhaps the same argument would apply to having positive pressure, because then you'd be making the intake fan fight against the higher pressure inside and become noisier...

which would make the entire subject of this "this pressure creates noise" discussion invalid, because now it seems to me that it should apply to both, so neither would be better. lol.

I'm still interested to know what happens in case of neutral pressure.

 

[RedJaron]

One last thing...there are tabs about 1/2" above the floor of the case that the PSU will rest on. Will the approximately 1/2" gap between the floor and the PSU allow enough air so I can position the PSU face down or should I mod the case (I have no issues with cutting it up) with a cutout and filter under the PSU (or should I just face the PSU up drawing in (likely?) warmer air from under the GPU)? Will either of the face up or face down options with no bottom hole cut out contribute more to a negative pressure airflow?

If I understand this correctly, the PSU sits on tabs, raising it off the floor ot the case, but still enclosing the whole thing in the case? There might be enough clearance beneath it for proper fan breathing, but you're still running into the problem of feeding warmer air within the case into the PSU to cool it down. Now if your system doesn't run very hot anyway, that may not be a big deal.

But if you're willing to mod the case, I'd give the PSU a vent through the bottom of the case and then give it a proper stand so it sits at least an inch off the floor to let the PSU breathe from below.

But please, please people get your computers off the carpet. Seriously, am I reading some of these questions right? Did someone really ask if an intake fan could breathe through carpet? Even if it could, that's asking for a lot of dust, lint, hair, and carpet fibers to be sucked into your case, not to mention overtaxing your fan and risking static buildup.

Get your machines off the carpet, either on a desk or on a non-conducting stand.

 

[TopGun]

If I understand this correctly, the PSU sits on tabs, raising it off the floor ot the case, but still enclosing the whole thing in the case?

That is correct.

There might be enough clearance beneath it for proper fan breathing, but you're still running into the problem of feeding warmer air within the case into the PSU to cool it down. Now if your system doesn't run very hot anyway, that may not be a big deal.

I doubt my case would run "very hot"...either way I doubt it will be a "big deal"...I'm just looking to optimize.

But if you're willing to mod the case, I'd give the PSU a vent through the bottom of the case and then give it a proper stand so it sits at least an inch off the floor to let the PSU breathe from below.

Get your machines off the carpet, either on a desk or on a non-conducting stand.

Ya, I've always put my PCs on some kind of hard surface (usually a wooden board or desktop). I'm definitely willing to mod the case, it's just a matter of is the extra time/effort/cost worth whatever benefits (if any) of adding a vent to the bottom of the case?

Any insight on my other questions?

 

[Arbie]

For those who are confused: All discussions of "negative pressure" are moot, as I think spentshells was saying. Far offsetting any theoretical advantage it may possibly have is that fact that IT WILL CRAP UP ALL THE COMPONENTS IN YOUR CASE. Dust *will* be sucked in, and once you have a layer of that on everything, local heat transfer takes a nosedive.

You want positive pressure. You get this by having appreciably more intake capability than exhaust. That will give you PP even in a mesh case. Start out with at least one more intake fan than exhaust fans (including the graphics card exhaust if it dumps outside the case). *Then* decide where you want the flow, how much noise is OK, how much to spend etc. You can juggle fan speeds, sizes, and positions - but always maintain more overall intake capability than exhaust, at least under normal operating conditions.

BTW this is another advantage of the PWM arrangement I described earlier. It does not affect pressure balance even as it compensates for load.

 

[RedJaron]

I doubt my case would run "very hot"...either way I doubt it will be a "big deal"...I'm just looking to optimize.

"Very hot" is a subjective term, and really I shouldn't have said that. It really depends on what kind of components your system uses, how they're configured ( mainly whether you overclock or not, ) and what kind of ventilation they get ( largely dependent on your case. )

A lot of Tom's heat reviews are done with overclocked hardware, which not everyone does. However, overclocked or not, most new video cards generate a lot of heat, and passing that hot air through the PSU is not the best idea. I can't say for certain whether it will prematurely wear down the PSU. Then again, if you're not planning on keeping the system, or particularly the PSU, for very long ( as in less than a year or two, ) then maybe you don't care if it wears a little fast. But if you don't have to do it, I wouldn't, just to be on the safe side.

To answer your other question, I believe the reason the most CPU coolers point toward the back and not the top is because until somewhat recently, cases didn't have vents on the top. Now they're fairly available, but the basic consumer case still vents out the back only. Also, Intel has square mounting brackets so if you want to point it up, you can. Sadly AMD mounts can't do this, though some coolers have brackets that can mount both directions on an AMD board.

 

[RedJaron]

Arbie, every computer, regardless of net positive or negative pressure, will accumulate dust on the inside. Yes, a net negative pressure in the case can attract extra dust around mesh vents and panel seams, but it won't appreciably deposit dust on the internals any more than a positive system would do.

All computers ought to be regularly cleaned out inside with canned air or some gentle vacuuming ( meaning a dustbuster, not a shop vac. ) Personally I do this about every six months.

But rather than worrying about negative and positive pressure, a better way to prevent dust in your computer is basic housekeeping. Get your computer off the floor and away from the carpet, change your furnace air filters regularly, vacuum your carpets, etc. This will make a much bigger difference in dust accumulation than any fan configuration.

And no, you DON'T need to start with an additional intake fan for positive pressure. With the huge variaty of fan size, speed, and efficiency, you can easily get net positive pressure with fewer intake than exhaust fans.

 

[zodiacfml]

i've been an enthusiast for case intakes and exhaust combinations but found that the most annoying is the dust. aside from dust causing bad cooling performance, it also increases the frequency of cleaning the system which is to me is downtime for anything you do.

i discovered that the negative and positive case pressure is not really of a concern for cooling but dust control. let me explain.

intakes is only useful to cool parts directly with ambient air such as Harddisks, video cards, or cpu. you still need more exhausts more than your intakes to prevent heat buildup.
the problem with intakes is they suck in dusts. using a filter doesn't control dusts, it just blocks bigger dusts and lets in finer dusts. you could make a thicker filter but it wouldn't suck air enough.

for dust control, never use intakes only exhausts. seal all holes of a case on all its sides except for the hard drive side intake, a fine filter would do great. clear tape or sealants will do.
also, the type of case should have holes at its bottom for it is where the air will come in.

this setup works since dust would be sucked only from the bottom but wouldn't reach the computer parts due to lack of air pressure. with this, dust only settles at the bottom of the case, and table/floor.
i still haven't cleaned my almost 3 year old video card also with the help of its auto fan speed control. motherboard is almost clean as new. the cpu heatsink and fan was cleaned for the first time a month ago.

 

[f-14]

the excessive sized sythe 10mm fans was an interesting discovery for the accelero passive heat sinks, i will have to try switching an aero slot cooler for this option.
[citation][nom]mjmjpfaff[/nom]I would have liked to seen silverstone's air penetrater fans in here. Just to see how they stack up against other fans.[/citation]
SILVERSTONE AP141 139mm Case Fan
Bearing Type: Fluid Dynamic
RPM: 1500 rpm@12V / 1000 rpm@7V / 700 rpm@5V
Air Flow: 28.26 CFM (5V) / 41.09 CFM (7V) / 64.34 CFM (12V)
Noise Level: 18 dBA (5V) / 20.9 dBA (7V) / 30.1 dBA (12V)
Model #: AP141
Item #: N82E16835220046
http://www.newegg.com/Product/Product.aspx?Item=N82E16835220038

IMO not a very good choice the 120mm 120mm
Bearing Type
Fluid Dynamic
RPM
1500 RPM
Air Flow
35.36 CFM
Noise Level
22.4 dBA

also not a very good choice, seems silverstone is still learning how to make fans.
my shopping search for fans is CFM then Noise then Price. only time i deviate is when i have a special need for a certain type of fan, like when i used sythes 10mm thick fans for tight spaces or axial fans for tight confined area higher needs cooling.

 

[mjmjpfaff]

the excessive sized sythe 10mm fans was an interesting discovery for the accelero passive heat sinks, i will have to try switching an aero slot cooler for this option.

SILVERSTONE AP141 139mm Case Fan
Bearing Type: Fluid Dynamic
RPM: 1500 rpm@12V / 1000 rpm@7V / 700 rpm@5V
Air Flow: 28.26 CFM (5V) / 41.09 CFM (7V) / 64.34 CFM (12V)
Noise Level: 18 dBA (5V) / 20.9 dBA (7V) / 30.1 dBA (12V)
Model #: AP141
Item #: N82E16835220046
http://www.newegg.com/Product/Product.aspx?Item=N82E16835220038

IMO not a very good choice the 120mm 120mm
Bearing Type
Fluid Dynamic
RPM
1500 RPM
Air Flow
35.36 CFM
Noise Level
22.4 dBA

also not a very good choice, seems silverstone is still learning how to make fans.
my shopping search for fans is CFM then Noise then Price. only time i deviate is when i have a special need for a certain type of fan, like when i used sythes 10mm thick fans for tight spaces or axial fans for tight confined area higher needs cooling.

thy direct air very well- http://www.youtube.com/watch?v=8m8fC809TK0

 

[RedJaron]

Zodiac, I can't disagree with that enough.

intakes is only useful to cool parts directly with ambient air such as Harddisks, video cards, or cpu.

Um, that's incorrect on multiple points. First, all air cooling is due to lower air temperatures, so I have no idea where you're trying to go with this. Second, you mention air intakes are only good at cooling HDDs, CPUs, and GPUs. Well, considering those items contribute to the vast majority of heat in a computer ( the only other thing that comes to mind it the mboard VRM, ) why is that a bad thing?

But more to the point, air intakes are not only "useful to cool parts directly with ambient air." Intakes also serve an important purpose of directing airflow around in the case.

you still need more exhausts more than your intakes to prevent heat buildup.

Care to explain this? Heat doesn't magically stay inside the case. Fluids flow from high pressure to low pressure, regardless of positive/negative sources. If there's a vacuum, air will rush in to fill it. If there's overpressure, air will rush away to equalize. Whether you're trying to expel hot air or pull in cool air, either way will work.

the problem with intakes is they suck in dusts. using a filter doesn't control dusts, it just blocks bigger dusts and lets in finer dusts. you could make a thicker filter but it wouldn't suck air enough.

Intake fans don't mystically suck in dust. Dust is in the air, period. Whether it's pulled in by a fan or pulled in by the vacuum in your case, dust will enter.

for dust control, never use intakes only exhausts. seal all holes of a case on all its sides except for the hard drive side intake, a fine filter would do great. clear tape or sealants will do.

Right, so the next time you need to replace or upgrade a part, you have to clear off all the tape and sealant? Worst idea I've heard all month. And didn't you just say that fine filters wouldn't let the case breathe properly?

also, the type of case should have holes at its bottom for it is where the air will come in.

Wait, didn't you just recommend sealing all holes, vents, and seams except on the front of the case?

this setup works since dust would be sucked only from the bottom but wouldn't reach the computer parts due to lack of air pressure. with this, dust only settles at the bottom of the case, and table/floor.

If you've sealed up the entire case, are using all fans as exhaust, and you're still lacking air pressure, that means you have poor air flow through the case, meaning poor cooling.


Cases are designed with vents in certain locations for very good reasons and suddenly sealing them up is not a good idea for a healthy and cool computer system. I'll say it again. If you're concerned about dust in your computer system, proper housecleaning has a much larger impact than fan configuration.

 

[Guest]

your positive pressure analysis...
I found to be completely wrong and infact completely opposite from my real world results and experience
another uninformative inspired toms article

 

[JoeMomma]

My layout works well.

http://xyzcoord.webs.com/images/airflow.jpg

http://xyzcoord.webs.com/media/airflow.cortby.pdf

I have the case on its side with a 24 inch monitor on the top front of the case.
There is a 120 mm horizontal intake left front(bottom).
It blows through to the left (under) of drives to the GPU.
Behind the monitor is a 140 mm vertical intake on the top left rear blowing down onto the GeForce.
GPU: GeForce 470 has a 80 mm exhaust fan with a vented card slot.
Before I improved my case, the GeForce would crash when overheated.
There is a 80 mm vertical intake fan top center.
Under the fan is a telescoping tube to a 92 mm CPU cooler with 4 copper heat pipes.
CPU: Core 2 Duo E8400 3.0 to 3.6 GHz
There is a 120 mm exhaust in the rear and a vented card slot below it.
The 800 W PSU has a 120 mm bottom exhaust fan (now sideways).
I used clear package tape to seal all of the holes and slots in the metal case.
All of the intake fans have washable foam filter grilles.

The case is pressurized and I addressed cable management by folding them up to make an air dam between the pressurized GPU cell at the left bottom of the case. The cables are wedged between the end of the graphics card and the back of the drives and top of the case. The CPU is getting twin supercharged charged fresh air from the tube. The PSU is exhausting the blow-by pressure from the CPU, GPU cell and the front intake.

I fiddled around a few times before I arrives on this setup a few months ago. The twin chamber air dam of wires trick dropped both my CPU and GPU temperatures by 10 degrees Celsius average across the board. At idle or while gaming. I get 47 CPU - 50 GPU in Windows (at 3 GHz core#0, 607 MHz CUDA). While gaming I get 60 CPU, 70 GPU max (at 625 MHz CUDA). It is quieter and cooler after using my engineering skills to tinker one step at a time. -10 deg ain't bad.

 

[JoeMomma]

I turned my top fan over to blow up and out, in the direction of convection.
I put washers between the fan and the removable side case to cure a rubbing noise.
The gap between the washers and the fan was sealed with black electric tape.
I scrubbed out all of the heat sinks, the CPU and inside of the GPU, etc. using an old soft tooth brush. Then I just had to air blast the leavin's.

Admitted: There was a just enough of crud on the front of the GeForce heat sink to affect my previous temperature readings. I clean the air filters on my case weekly and dust the inside monthly. The card was not even very dusty when I opened the case, but when I un-clipped the lid on the card I saw a little cotton candy dust bunny on the leading edge of a beautiful 5 heat pipe heat sink. There was not a lot at first but looking closer the little there was was in just the wrong place.

Gaming-HIGH
Before: CPU = 65 C. Before: GPU = 70C. Fan DOWN in
After : CPU = 55 C. After : GPU = 60C. CLEAN GPU + Fan UP out

Normal-IDLE
Before: CPU = 50 C. Before: GPU = 60C. Fan DOWN in
After : CPU = 45 C. After : GPU = 50C. CLEAN GPU + Fan UP out