Less airflow equals cooler temps.....

[Bluefront]

The airflow out of a number of cases is wasted, as far as cooling is concerned. It by-passes the hot spots and blows right out the case without doing much good. You can improve the cooling of what airflow you have by careful design of the flow path, strategic fan placement, or various shaped ducts.

But what I'm suggesting here involves time.....the longer a particular volumn of air remains in the case, the better chance it has to pick up and remove any heat.Of course there is a limit to this.....go too slow and the effect is nullified.

This can be somewhat demoed in a automotive AC system using a thermometer. If you turn the fan speed to low, the output temp of the vents goes down, because the air remains in the cooling unit for a longer period of time before it is blown out the vents.

How can this effect be put to use in a computer? Maybe by a complicated airflow path for the exhaust, giving the airflow more time to pick up heat. I've been trying this out in my newest project which vents the exhaust out the top of the case. By careful placement of an upper exhaust deflector, which forces the exhaust to stay in the case relatively longer, I can get the temp of the components in the case to go down......while the temp of the exhaust goes up.

It seems to be working......finding the "sweet spot" is difficult, however.

[IsaacKuo]

This can be somewhat demoed in a automotive AC system using a thermometer. If you turn the fan speed to low, the output temp of the vents goes down, because the air remains in the cooling unit for a longer period of time before it is blown out the vents.

With an AC system, it's true that the air gets cooler the slower it flows--but the evaporator also gets cooler the slower the air flows!

For any given computer component, lower airflow will mean hotter temperatures. It's true that the air has more time to pick up heat, so it will be hotter. But it's also true that the component will be hotter--and that's the temperature we care about. No matter how fast or slow the air is flowing, the heatsink has the same area to transfer heat. The thing which changes is the temperature of the air contacting the heatsink. The hotter the air, the less heat gets transfered.

Maybe by a complicated airflow path for the exhaust, giving the airflow more time to pick up heat.

Now, this IS a potentially useful principle, and one which I use all the time. But the reason this principle can work is not what you're thinking. It's simply a matter of reusing airflow. My theory is that it can theoretically be better to "reuse" airflow by taking air that's used to cool a cool component and using the warmed air to cool a hot component. The traditional example of this is intake air cooling the hard drive and then being reused to cool all other components.

Taking this principle to the extreme, you'd think you'd want a maze-like pathway to cool every component in line. In practice, I've found that maze-like pathways lead to air restrictions and "turbulence" noise. I've gotten the best results with a simple "U-turn" airflow path.

I've been trying this out in my newest project which vents the exhaust out the top of the case. By careful placement of an upper exhaust deflector, which forces the exhaust to stay in the case relatively longer, I can get the temp of the components in the case to go down......while the temp of the exhaust goes up.

This sounds a little strange to me. There's only one exhaust, right? The same amount of heat should be going out the exhaust either way, theoretically. A deflector might increase the exhaust temperature indirectly by reducing overall exhaust airflow...

Hmm...at any rate swirling the air around within the case should help cool components however it's done.

[HammerSandwich]

Bluefront, I agree that you can more effectively utilize airflow with ducting, but I take issue with some of what you wrote. I'm going to start sounding like a broken record around here, but heat and temperature are not the same thing.

the longer a particular volumn of air remains in the case, the better chance it has to pick up and remove any heat.

Heat transfer is all about the cooling stream's mass flow times the temperature difference between it and the item being cooled. (The specific heat of your cooling medium applies as well, though that can be ignored if we assume air cooling.) If you need to cool X watts of CPU, the heatsink's temperature will climb until X watts transfer to the air. More air will give a lower temperature differential, which means better CPU cooling. Thinking about how much heat a given molecule of air picks up is misleading: there are simply too many molecules to visualize. [edit]I just googled this and found it's about 767 sextillion molecules in a cubic foot![/edit]

This can be somewhat demoed in a automotive AC system using a thermometer. If you turn the fan speed to low, the output temp of the vents goes down...

A temperature change does not imply a change in total heat when the airflow is altered as well. In your example, the air might be colder, but it will not cool the cabin as well as with a higher fan speed.

[jaganath]

Taking this principle to the extreme, you'd think you'd want a maze-like pathway to cool every component in line. In practice, I've found that maze-like pathways lead to air restrictions and "turbulence" noise.

Axial fans are rubbish when it comes to fighting backpressure, and that's what multiple obstacles and sharp turns in the airflow path represent. So to maximise the airflow (and hence cooling) from our axial fans, we want to reduce impedance in the airflow path, such as cutting out fan grills, etc. The "U-turn" arrangment works because the change in airflow direction is gradual, rather than sudden. Also convection at work etc.

[breunor]

A temperature change does not imply a change in total heat when the airflow is altered as well. In your example, the air might be colder, but it will not cool the cabin as well as with a higher fan speed.

Yup, say if reducing airflow drops teh temperature by 15%, but you have 40% less airflow due to it. The net result is less heat being removed from the case.

Bluefront's recent case designs are pretty sweet on maximizing the airflow/cooling with limited fan use/noise. I think he's worked more on maximizing the heat transfer of a fixed airflow more than trying to reduce airflow itself if that makes sense, at least that's the impression I've had.

[~El~Jefe~]

there is a huge difference between heat and temperature! remember your earth science class! (or not as I was 13 when i took it in 1986).

A bathtub at 105 degrees F is not that hot compared to a cpu at 140 degrees F, but the amount of heat in the tub is ENORMOUS!

your air conditioning example I feel is due to this. The temps go down if you let the stuff sit more and move less. however, the object of the AC is remove heat from the car not lower the temperature at a given point. The heat is moved best by a high flow of air so the air's energy can be lowered as fast as possible. Higher flow does this in an air system. There probably is a terminal point to this, but I dont think AC's do it to that degree. (eventually air speed causes friction!)

The other possibility is the chilling of the liquid in the ac unit increases its efficiency. Kind of like priming it. if you pull out the heat too fast, it will take longer to do its job. When I first start an ac, I always close the louvres (spelling?) so as to chill it before the room if it hasnt been on all year.

I could be wrong though.

As far as computer cases I find that almost 99% of the cases I have seen in the world, and maybe more than that, are starved for intake air and positive pressure. If your case has dust jammed in tiny holes around it where air can barely pass, and you clean your case out 2x a year, to me, thats the proof that you need more intake air. I have noticed that psu's ramp up with more out going airflow, and in general, higher ambient temps for the system on load when air is pushed out to quickly. However, if you had 2 120's on front and 1 on back with a psu that also uses a fan (mine doesnt anymore to prevent this situation) ... you will get much lower temps (in gaming rigs that I have fixed at least) than compared to more outgoing fans.

[Bluefront]

First off....I'm not an engineer in this particular field. I'm just a duffer, with a bunch of computer stuff, and the ability to mod these things easily. This particular wooden case I've been tweaking, is well suited for experimentation. All the exhaust exits fron the top of the case (with the exception of the PSU which has an independant air path). The case fan blows in through the heatsink, makes a 90 degree turn, and goes out the top. This fan is thermal controlled with a sensor on the base of the heatsink. Changes in airflow that I make are immediately noticeable by changes in the fan speed and the CPU temp.

There is a fan under the hard drives (which are mounted vertically), blowing upward between the two drives, and out the top. Some of this aiflow also blows upward through the open area between the drives and the CPU heatsink. This airflow helps the CPU airflow make it's 90 degree turn to exit the case. The other fan visible in the photo vents the video card area and has little effect on the other temps. This fan and the HD fan are fixed at 5V. The only fan that changes speed is the CPU fan.



Since is only takes seconds to change the top, while the computer is running, I can easily monitor temp and fan speed changes. So far I have made four different tops, plus I can adjust the angle of each different top quickly with small spacers (while the computer is running). Here's the top design that gives the best CPU temps, with the lowest CPU fan speed....



So what's happening here? Why does this top give the CPU a slightly lower temp and a lower fan speed than no top at all. Well I think it has to do with the back pressure on the HD fan airflow. This airflow is somewhat forced toward the CPU heatsink with the addition of a top, and provides slightly more cooling to the CPU, allowing the CPU fan to slow down. In effect the airflow from the HD fan spends more time in the case.....and causes the CPU to cool better. The HD temps don't seem to be affected by this particular top.

I'll admit the temp and fan speed changes are only slightly in a positive direction, but every little bit helps. In a case with really poor airflow, this time thing might prove more effective. And I do have a problem with ambient temps in this computer testing area.....the ambient varies maybe 2C as the central AC turns on/off. Even a 1C change in the ambient, changes the CPU temp and fan speed. So any results I achieve are somewhat suspect.....considering the temp/fan speed changes are so slight when I adjust the top.

This is a difficult experiment to prove....or even to duplicate. But it's something to consider when designing an airflow technique.

Oh....what does this have to do with noise you ask? With a top on it, the computer is quieter when sitting in front of it. So it is reasonable for me to make a top that doesn't worsen the temps, maybe improve temps....

[IsaacKuo]

So what's happening here? Why does this top give the CPU a slightly lower temp and a lower fan speed than no top at all. Well I think it has to do with the back pressure on the HD fan airflow. This airflow is somewhat forced toward the CPU heatsink with the addition of a top, and provides slightly more cooling to the CPU, allowing the CPU fan to slow down. In effect the airflow from the HD fan spends more time in the case.....and causes the CPU to cool better. The HD temps don't seem to be affected by this particular top.

Well, it is a good idea to reuse air which is slightly warmed by the hard drives to cool the CPU. The addition of that top would indeed tend to direct air from the hard drive intake fan toward the CPU heatsink--but this is an ugly situation because it's fighting the "natural" airflow direction of the CPU intake fan.

You're intuition that it's helpful for the HD heated air to spend more time in the case is correct--but as I noted this is simply an example of reusing slightly warmed air.

Thus, I repeat a suggestion I've made before--could you try flipping the CPU fan so that it exhausts air instead of sucking it in? That way, CPU warmed air would immediately exit out of the case. You could even close the top entirely and remove the hard drive intake fan.

[Bluefront]

Well this is mostly a positive pressure case......and using the rear fan as an intake makes it easy/possible to filter most of the incoming air. As it was designed there is really no way to cool everything in the case sucking out through the rear fan. Not without adding more fans that is. Right now the NB, the ram, all the board components benefit from ambient temp airflow. The bottom of the rear fan is not ducted, so there is plenty of airflow going over these components that hasn't passed through the CPU heatsink. I'm getting relatively cool temps on everything with this positive pressure airflow. I don't think this can be improved by going negative pressure.

FWIW.....I have a second unused Compaq computer framework, identical to the first one I used to make this case. I'm willing to take suggestions on a second wood case with different airflow. My only concern is that the intake air must be filtered.....most of it anyway.

[pcy]

Hi Bluefront,



The results you post are convincing. The reasons you are getting them are more difficuklt to figure...


I'm looking hard at those photos and trying to vizualize how the whole case is organized, how all the fans are positioned and how the airflow works.

And failing.... any chance of some more pics, or section diagrams?


Anyway, ther are some things I certain of, even so.


This has to be about airflow, and not about time. Cooling effect depends only on the temperature difference between the hot object and the coolant fluid (air). The speed of the air has no effect at all, except indirectly in that the slower the airflow the hotter that air gets, so the hot component also must heat up to maintain the temperature difference for the same heat dissipation.

But quite subtle changes in ducting can have significant effects on airflow. The two critiacal factors are the temperature and pressure of the air arriving at the input side of the CPU heatsink. Improving either (i.e increasing the input air pressure or reducing the input air temperature) will improve the CPU cooling.

Reducing input air temperature increases the heatsink-air temperature difference (for given airflow). Increasing input air pressure will increase airflow (for constant fan speed), which in turn reduces the average air temperature accross the entire heat transfer surface, again increasing the heatsink-air temperature difference. Total heat transfer is directly proportional to the heatsink-air temperature difference; so either improvement reduces CPU temp, or allows you to reduce the fan speed for constant CPU temp.


The problem we face is that airflow is hard to see and does not necessarily behave in a very intuitive way. Go talk to any aerodynamacist! The air velocity inside a computer case is (or should be, if you are using large fans at low voltage and large ducts) low, so that airflow resistance is dominated by viscosity effects and reduces with the 4th power of the linear dimensions of the air ducts. Irregularities and sharp bends create back pressure (as others have mentioned), and should be avoided; but these effects are generally proportional the the square of the air velicity, so they too almost vanish with sufficiently large area for the airflow.

So it's easy to see that the different covers could have significant effects, but harder to see what, exactly, each one does, and why.


Back to the request for more pics/ diagrams...





Peter

[pcy]

Hi Bluefront,



The results you post are convincing. The reasons you are getting them are more difficuklt to figure...


I'm looking hard at those photos and trying to vizualize how the whole case is organized, how all the fans are positioned and how the airflow works.

And failing.... any chance of some more pics, or section diagrams?


Anyway, ther are some things I certain of, even so.


This has to be about airflow, and not about time. Cooling effect depends only on the temperature difference between the hot object and the coolant fluid (air). The speed of the air has no effect at all, except indirectly in that the slower the airflow the hotter that air gets, so the hot component also must heat up to maintain the temperature difference for the same heat dissipation.

But quite subtle changes in ducting can have significant effects on airflow. The two critiacal factors are the temperature and pressure of the air arriving at the input side of the CPU heatsink. Improving either (i.e increasing the input air pressure or reducing the input air temperature) will improve the CPU cooling.

Reducing input air temperature increases the heatsink-air temperature difference (for given airflow). Increasing input air pressure will increase airflow (for constant fan speed), which in turn reduces the average air temperature accross the entire heat transfer surface, again increasing the heatsink-air temperature difference. Total heat transfer is directly proportional to the heatsink-air temperature difference; so either improvement reduces CPU temp, or allows you to reduce the fan speed for constant CPU temp.


The problem we face is that airflow is hard to see and does not necessarily behave in a very intuitive way. Go talk to any aerodynamacist! The air velocity inside a computer case is (or should be, if you are using large fans at low voltage and large ducts) low, so that airflow resistance is dominated by viscosity effects and reduces with the 4th power of the linear dimensions of the air ducts. Irregularities and sharp bends create back pressure (as others have mentioned), and should be avoided; but these effects are generally proportional the the square of the air velicity, so they too almost vanish with sufficiently large area for the airflow.

So it's easy to see that the different covers could have significant effects, but harder to see what, exactly, each one does, and why.


Back to the request for more pics/ diagrams...





Peter

[Bluefront]

I don't have a diagram.....it would have to be a complicated 3-D image that I could never manage. Here's a link to the whole wood case project. The fan blowing upward between the two hard drives, sucking air from the lower air chamber, proved to have a much bigger effect on the whole system than I suspected. If I ever make a similar design, I'll probably make better use of this factor....

FWIW.....locating the optical drive off to the extreme left side, mounted vertically, made this straight-through hard drive airflow, possible. I've never seen a case with this ability. It's something to work on.

[pcy]

Hi Bluefront,


Magnificent job.


That link helps. Let me see if I have it right. Air comes in through the bootom and exits through the top and out the back. All the air used to cool the CPU goes out the back. If that's so, the top increases internal air pressure which helps cool the CPU.

But if you close the top completely what happens. Three factors would be at work.

1. Internal air pressure and airflow through the CPU cooler would be increased, improving CPU cooling
2. Total aiflow through the case would be reduced, increasing the internal air temperature, which would detract from CPU cooling.
3. Totally blocking the top air exit might introduce turbulence with all sorts of effects.



Though very different you might like to look at the case I am designing, which should be on the market soon:
http://www.paq.ltd.uk

Very different in some ways, but I see the same ideas at work also.

There are two pics on the Information tab. The air comes in the front and passes over the HD. The case fans are internal - 3 x 120mm running at 5.5V. All the air exist either through the PSU or through the CPU cooler, which is ducted out the back. I'm now using a 140mm fan for the CPU.

It's also a straight through airpath. It also has a single fan doubling as the CPU fan and exhaust fan.




Peter

[jaganath]

The air comes in the front and passes over the HD. The case fans are internal - 3 x 120mm running at 5.5V. All the air exits either through the PSU or through the CPU cooler, which is ducted out the back. I'm now using a 140mm fan for the CPU.

It's also a straight through airpath. It also has a single fan doubling as the CPU fan and exhaust fan.

Is this not almost identical to Intel's BTX thermal case design?

[pcy]

Hi,

Ive never seen a BTX case in the flesh, and though I've searched the web I've only found one case that used internal fans for case ventilation the way the PaQ case does.

But I think that you're right that the BTX design exauasts the hot air from the output side of the CPU cooler straight out of the case. It's pretty obvious that you don't want to allow the hottest air to re-circualte...



Peter

[Bluefront]

Humm....interesting designs. With a longer case you can mount the fans right in the middle, which definately can give you an acoustic advantage. I've used this technique in a number of case designs. It does make for a much bigger case however.

And you say the exhaust has to go through a foam muffler? I'm afraid that sort of design would only be suitable for a low-power system. This wood case of mine is using a P4-3.4 which is about 95W.....not as hot as a Prescott, but plenty hot.

Not so obvious in my photos is the separate air chamber underneath the steel internal framework. It is about 20mm high, and covers the whole bottom of the case. This chamber has three openings (the photos only show two), which give a somewhat indirect path for air entry to the HD fan and the PSU fan.....makes it quieter. The PSU uses a 92mm Nexus....remotely mounted from the fanless PSU. It's inaudable except if you put your ear right to it.

link to psu mod

[pcy]

Hi,

The case is 7" by 18" by 22". That's about as small as I can make it and still get the three internal 120mm case fans in.


We've worked on the mufflers since those details were posted. The best I've had has been an X2 4800 plus 2 300GB HD and a fanless nVidia7300 running flat out (2 copies of Prime95) in 30C ambient.

The PSU fan was runnin c.800rpm, the case fans abut the same and the CPU fan at 950rpm was the loudest, but dropped to 650 on idle.

Overall I'm pretty pleased; and I'm hoping MikeC will accept one for review once I've got some spare...


But back to your case. Did I have the basic airflow right? I really think that increasing internal air pressure can do wonders for aiflow through a CPU heatsing that is ducted straight to the outside....



Peter

[breunor]

Both the lower fan blowing up into the case and the ducted cpu fan blowing into the case are providing positive pressure, and the only exhaust is at the top, aided by thermal flow. I'm not sure what else could be done to increase that flow really, without also increasing the case size to allow for smoother airflow.

[HammerSandwich]

But I think that you're right that the BTX design exauasts the hot air from the output side of the CPU cooler straight out of the case. It's pretty obvious that you don't want to allow the hottest air to re-circualte...

That's pretty obvious to everybody except the people making 120W CPUs. BTX feeds outside air to the CPU's HSF, and the exhaust dumps inside the case. Good for CPU cooling, but terrible for everything else.

[IsaacKuo]

...says the guy who put his watercooling CPU radiator in a BTX-like front intake location.

[pcy]

But I think that you're right that the BTX design exauasts the hot air from the output side of the CPU cooler straight out of the case. It's pretty obvious that you don't want to allow the hottest air to re-circualte...

That's pretty obvious to everybody except the people making 120W CPUs. BTX feeds outside air to the CPU's HSF, and the exhaust dumps inside the case. Good for CPU cooling, but terrible for everything else.

Then my case is not like the BTX design, because I do duct all teh hot air straight out the back.


Peter

[HammerSandwich]

...says the guy who put his watercooling CPU radiator in a BTX-like front intake location.

I plead guilty. OTOH, the HD and PSU do have a separate intake, and I could add the GPU to the WCing loop if needed. Come to think of it, no one's made an offer on the GPU block in my FS thread, so...