The Absolute Ultimate Air-Cooling Technique.

Bluefront · Post by **Bluefront** » Fri Apr 28, 2006 4:02 pm

It's for real. I have never made this claim before for any of my many projects involving low-noise, excellent cooling, moderate cost, mostly DIY. And it's not the case (an Antec Aria), nor is it the fan (a Globe), or the CPU (a moderate heat P4-2.8, not undervolted or under-clocked).

It's simply the ducted positive airflow, combined with an XP-120, and using only the one 120mm Globe fan for the whole case. The std PSU is removed from the case, replaced with a dc/dc converter, and using an external 12V power supply. The one fan blows in from the rear through a wide duct, the ambient temp air hitting the heatsink, the NB, the ram, the DC/DC converter, the Windbond chip, and everything else on the board. The duct is wide enough to overlap the video card (an nvidia 6200, runs about 44C). The top of the duct is made of a copper sheet, running ambient temps. A laptop drive (80gb Samsung SATA) is screwed to this cool-plate, and runs about 30C. After the airflow hits the various MB components, it exits the case through some stock vents, but mostly through a new vent constructed on a custom top.

Here's a side view of the thing as it is running right now (still in the tweeking stage.

These temp readings were taken at 22.3C ambient, with the one Globe fan at 5V, 760rpms. That's as low as I can go with a Fanmate.....due to be replaced with a better fan control system shortly. It idles about 31C, and after a long run using two instances of CPUBurn, here's the top temps....

It's the airflow gentleman, a cool flow of air directly hitting all the hot parts, and allowed to passively vent itself out the case. One fan at 5V doing everything. This complete project is posted in the Gallery, and my complete photo gallary is in Yahoo Photos. Some of the photos show a small fan, not necessary any longer and not used during the latest tests.

This airflow technique could be easily used to cool the hottest CPU, or with a different CPU and MB......could be the quietest setup you'll ever likely see. I'm open for criticism about these claims......but I'm totally convinced this is the way to design your quietest computer. Any number of cases could be modded to use this technique, not just an Aria. But I don't think a tower case would work as well.

IsaacKuo · Post by **IsaacKuo** » Sat Apr 29, 2006 3:00 am

Interesting! We've reached radically different conclusions about what the "ultimate" air-cooling technique is. You've gone with the ultimate "parallel" concept--fresh air from a single fan is ducted out to each component in proportion to how much cooling it needs.

I believe the ultimate technique is the "serial" concept. I believe the way to maximize the utilization of a single fan is to reuse the air as much as possible. My justification is that any air which is moderately warm leaving the case is wasted cooling potential.

So, I think the ultimate would be airflow which passes each component in turn. It should start with the component(s) which need the lowest temperatures and generate the least heat, and end with the component(s) which can tolerate the highest temperatures and generate the greatest heat. Assuming a low power GPU and an efficient PSU, I feel the ideal airflow path would be:

fresh air -> hard drive -> chipset/vga/case -> PSU -> CPU -> exhaust

I haven't yet implemented this exact scheme. However, I'd note that it's actually not very different from a traditional airflow plan. Also, for a typical PSU which isn't all that efficient, it makes more sense to split the airflow at the end so the PSU and CPU both directly exhaust out to the air. And that is the traditional airflow plan with a ducted CPU heatsink!

I am, like, totally serial.

Bluefront · Post by **Bluefront** » Sat Apr 29, 2006 3:43 am

Issac.....you'll have to demonstrate that your proposed system can cool better and quieter. Designs on paper are interesting, but with no guarantee they will work. This system I'm using has gone through at least four different duct designs, all very similiar, but with different results. And I think more tweaking will improve things.

Right now I'm stuck on a fan speed problem......the fanmate won't lower the speed any further, even though it could easily idle at less than 4V. Also the computer cannot read rpms lower than about 750 with this Globe fan, and I have booting difficulty.

Your idea about using airflow in a serial manner certainly works, but it makes for complicated ducting. My setup is simple.....one wide duct facing down. The heated air flows up and out the case, a totally natural, passive exhaust flow. Right now I can remove the top of this Aria, and the ducting works the same, and the temps hardly change.

Another interesting thing here.....I idle about 31C, system 32C, HD 28C. This is at 5V, 750rpm. If I increase the fanspeed to the max, about 1800rpms, the temps hardly change. So I think this system is running as cool as possible, no matter the airflow.

Some other people are going to have to try this technique to confirm my conclusion. It's relatively easy to pull it off.

IsaacKuo · Post by **IsaacKuo** » Sat Apr 29, 2006 4:23 am

I don't measure temps except with the "finger" method. The cheap mobo's I purchase never seem to really support all that fancy stuff.

So for me, "ultimate" simply means how quiet/silent the thing is. I've used the "serial" method to cool my 2.5Ghz Northwood workstation with a single 80mm fan at 5v. I don't have any hard numbers, but the fan I used pushes VERY little air. I chose it purely on the grounds of noise level.

I don't have anything hotter than a 2.5Ghz Northwood...and I don't have any video cards hotter than a Radeon 9250...so that's as far as I can get. (The power supply on this workstation is a cheap Enlight; I can't imagine it's very efficient so this system is producing maybe 20-30 percent more heat than it would with an efficient PSU.)

So, my only real hard conclusion is that I can cool my hottest computer adequately with a single very low flow 80mm fan. If you can get the parallel concept to work with a low flow 80mm fan, then I'll be quite impressed.

A single 80mm fan can in principle be used to cool a quieter computer than a single 120mm fan. Besides the possibility of producing less noise to begin with, the lower minimum airflow can mean less "whoosh". Also, an 80mm fan can be used with an intake and exhaust of half of the area. That can mean smaller portholes for noise to escape the case.

In any case, there's no way the "serial" concept is going to produce the coolest temperatures. It's not going to. It's not supposed to. It purposefully uses warm air to cool components which can afford to run hotter. That's never going to be able to produce the sort of temps that cool fresh air can.

It's not about producing the lowest temperatures. It's about producing adequate temperatures with the minimum amount of noise.

Bluefront · Post by **Bluefront** » Sat Apr 29, 2006 4:55 am

Well I always am concerned about temps, and measure everything carefully, using several methods. My theory is that using ambient temp airflow will yield the coolest temps, which in turn means the fan speed can be lowered. Right now I'm bottomed out on fan speed.

And here is the other difficulty comparing systems.....different components with different heat outputs. My power supply setup produces almost no internal case heat, as does the laptop drive. I am using a P4-2.8, with a relatively hot nvidia 6200, and a hot TV tuner card.

I've considered trying a smaller fan, but I don't know how it would perform with the necessary wide duct for this setup. If/when I get a different fan control system, using PWM, I might be able to idle at 4V or less......basically completely inaudible in a closed case. The intake muffler in this setup makes this setup almost inaudible at 5V......at an arm's length I cannot hear it right now.

jaganath · Post by **jaganath** » Sat Apr 29, 2006 5:13 am

Right now I'm stuck on a fan speed problem......the fanmate won't lower the speed any further, even though it could easily idle at less than 4V.

I find fanmate + resistor gets it below 5V. For the resistor I mean one of these:

http://www.quietpc.com/gb-en-gbp/products/casefanaccessories/resistors

I am using a P4-2.8, with a relatively hot nvidia 6200,

I thought the 6200 was too much for your power supply? Sorry if I missed the part where you changed to an uprated PSU.

IsaacKuo · Post by **IsaacKuo** » Sat Apr 29, 2006 5:13 am

I don't think there's a good reason to go to PWM. I'd be worried that it might even make the fan noise more noticeable due to "pulsing".

It's really impossible to truly compare our two systems. My main workstation has:

2.5Ghz Northwood Celeron, 1st gen Scythe Heatlane Zen heatsink (not as good as later Scythe towers) -- COOLER than P4-2.8

Radeon 7200 - This had a whiny 40mm fan which I removed and the puny heatsink underneath it is actively cooled with air exhausted from the PSU. --COOLER than nVidia6200

VIA chipset - not sure what you've got but the VIA chipset probably runs COOLER

I use a Western Digital Scorpio, although I found the noise completely unacceptable without a full enclosure. Thermally, it should still run about the SAME TEMP.

Last, but not least--the power supply. It's a cheap old Enlight 300 which I really don't think is very efficient. This runs HOTTER.

So who knows? At a guess I'd imagine our two systems are roughly in the same ballpark, but it's really impossible to say.

In any case, my system is probably louder than yours. That damnable Enlight makes bizarre whines and chirps. As you know, I'm really cheap--but I'm getting to the point where I'd be willing to spend a premium for a PSU if I knew it wouldn't make these strange noises.

I'm really afraid to risk spending the money, though. I just picked up a bunch of SPCR Clearance PSU's from MikeC, and the Coolmax Taurus 500 has coil whine so bad that even my wife can't stand it from across the room. When my wife can't stand it, that means it's BAD!

tay · Post by **tay** » Sat Apr 29, 2006 5:35 am

Nice build as usual bluefront. I love the padded ducts and the low cost high effectiveness solutions.

Here are questions to both you and Isaac.

- Due you think you get an advantage from using a small case when you are going with single fan systems? (for low powered systems)

- Can someone with a 80watt dual core and a 60watt video card (7600GT etc.) use such a system? (i suspect not). A followup to this is that can you cool such hot components in a smallish case as well? Or would you choose a larger case with 2 fans.

- Why are you so against putting 2 fans in at lets say 4v rather than just one?

IsaacKuo · Post by **IsaacKuo** » Sat Apr 29, 2006 6:19 am

tay wrote:- Due you think you get an advantage from using a small case when you are going with single fan systems? (for low powered systems)

No. I use small cases purely for style. It's really hard to work with, especially since I tend to use full size ATX mobos and PSUs (they can be a lot cheaper, with the right deals).

For example, one thing I'd like to do more of is moving the 120mm fan of a PSU to the outside of the PSU. Normally, a 120mm fan seriously blocks airflow out the back of the PSU. Simply moving the fan to the outside of the PSU opens up the airflow a lot. Also, moving it to the outside lets you soft-mount it. But obviously, this is only possible if you've got some room to work with between the PSU and the CPU heatsink--especially if the CPU heatsink is ducted to the rear!

- Can someone with a 80watt dual core and a 60watt video card (7600GT etc.) use such a system? (i suspect not).

I'll bet Bluefront's "parallel" method would work pretty well, although the case temps could be a bit warm. However, I suspect you could get away with less noise using two fans.

A pure "serial" method would NOT be a good idea. Even if you could figure out how to get the airflow to work, you'd either be directing CPU heated air to the GPU or GPU heated air to the CPU. Not a good idea either way!

A followup to this is that can you cool such hot components in a smallish case as well? Or would you choose a larger case with 2 fans.

You can cool anything you want in a smallish case, but it can be a real pain getting it to work well. In my Northwood workstation, I sacrificed all of the PCI slots in the name of compact stylishness.

If I were cooling your system, I'd go with a relatively traditional ATX case layout. It's actually a really good layout, with some changes here and there.

For a single fan strategy with hot components, I rather like the idea of an "inverse" parallel method. Bluefront's strategy is the intuitive one of ducting intake air at the main components. However, with a typical ATX case I think it's easier to do the opposite--duct exhaust air from the main components. That means the intake fan isn't ducted at all, it's just pushing air into the case creating positive pressure. Then each main component is fanless, each with individual exhaust ducts to the outside.

The real challenge with this strategy is the intake fan. Most cases have really restricted intakes, and many have extra holes behind the front bezel which need to be plugged up (otherwise air will recirculate through those holes into the front bezel where it gets re-sucked into the intake fan). In my latest build, I opened up the PCI slots for a big unrestrictive rear main intake--but I don't have a high power videocard. With a high power videocard, the GPU exhaust would be right next to the PCI slot intakes...I haven't thought up a good solution to this problem.

If you can figure out the intake fan, then the rest is pretty easy. The PSU simply needs its fan removed; it already vents directly outside. The CPU needs a tower heatsink and rear case duct--standard fare around here! The GPU needs something like an Ati silencer with the fan removed.

- Why are you so against putting 2 fans in at lets say 4v rather than just one?

There's nothing wrong with that, if you need that amount of airflow. But since the subject is the "ultimate" technique, the question is--how low can you go?

cAPSLOCK · Post by **cAPSLOCK** » Sat Apr 29, 2006 8:22 am

Still no dust filter though!

I agree that ducted positive pressure works really well, have a look at my pc (in the sig), that's basically what I have done, except the intake is at the front, which is less good from a noise perspective. I can also turn my intake fan down really low (subjectivly my case feels cood as opposed to hot), but in my setup it would make the gfx fan stop (they share a fanmate).

Bluefront · Post by **Bluefront** » Sat Apr 29, 2006 9:18 am

Heh....don't worry, an intake filtration system is in the design stage for this Aria. It'll be installed when I finish my custom top.

I am 100% certain this airflow system would handle the hottest Prescott.....not at 5V however. My problem is the DC/DC converter, which is only 200W, and the 12V power supply which is 180W. There are many ways to remove a PSU from the inside of a computer case. A std psu in a small case is an extreme limiting factor in what you can do. This converter is one solution.

I prefer the fewest number of fans.....which is one. Actually with this Aria, there is little need for a second fan unless you're using a 3.5" HD. Even then with more complicated ducting, a single fan might be enough. I've become a small case advocate, using a MATX board. No reason it cannot be as quiet and cool as a big case.

frankgehry · Post by **frankgehry** » Sat Apr 29, 2006 9:40 am

This really seems like btx, with the airflow reversed to fit the aria layout. In-line airflow cools the hot components with the same fan, but btx goes a bit further by providing airflow under the mb and optimizing component placement to maximize the cooling effects of an in-line airflow system.

diver · Post by **diver** » Sat Apr 29, 2006 9:43 am

The in ultimate air cooling is easy.

245

Two windows open, 45 MPH.

How long is your extension cord?

Bluefront · Post by **Bluefront** » Sat Apr 29, 2006 11:04 am

One big difference between this setup and BTX.....this setup uses natural convection current to exhaust the case. And you would be surprised how effective that is, combined with positive case pressure. Place your hand over the upper vent opening and you'd swear there was a fan right there blowing outward/upward.

And you know, I was thinking of cutting a big vent opening on the bottom of the case, in the CPU area. This would provide some passive heat output under the board.....at not much complication. Might do that yet....

ArcticOC · Post by **ArcticOC** » Sat Apr 29, 2006 11:05 am

Ducting is a nice thing, my system is also a duct-based solution.

The fan in the PSU is the only one except for the one in the NVSilencer.

Le_Gritche · Post by **Le_Gritche** » Sat Apr 29, 2006 11:20 am

jaganath wrote:I find fanmate + resistor gets it below 5V. For the resistor I mean one of these:

http://www.quietpc.com/gb-en-gbp/products/casefanaccessories/resistors

You mean they really sell a simple resistor soldered to a wire ! 2 pounds each !

Soon they will sell bottled water ...oh wait

I use the fanmate+resistor combo too, but for 2 f*king pounds these days you can probably buy a cheap chineese solder iron and the resistor

Bluefront · Post by **Bluefront** » Sat Apr 29, 2006 12:23 pm

The problem with a resistor plus a fanmate.....I'd end up with an analog voltage below 5V. And I doubt the Globe will start-up much below 5V. That's why I think a PWM solution would be better. I have used these Globe fans with PWM, and they handle PWM well, with no extra noise.

The last thing I want is a single-fan computer with a fan that doesn't always start.

jaganath · Post by **jaganath** » Sat Apr 29, 2006 1:11 pm

for 2 f*king pounds these days you can probably buy a cheap chineese solder iron and the resistor

I don't know if you're familiar with English pounds but you certainly can't buy a soldering iron and a resistor for 2 pounds. You could probably steal it for nothing, but I doubt theft is your habitual mode of acquiring technical equipment.

You mean they really sell a simple resistor soldered to a wire ! 2 pounds each !
Soon they will sell bottled water ...oh wait

Oh ho ho ho, yes, bottled water is such a con isn't it.....expect for the millions of people in the developing world, for whom it is their only source of clean drinking water.

Aris · Post by **Aris** » Sat Apr 29, 2006 1:43 pm

I noticed this technique a while ago. I also believe it to be the best air cooling technique, for sound as well as cooling.

was building a mini itx system in a very small mini itx enclosure. where the video card is situated over the motherboard (parrellel to it). i had an 80mm fan running at around 6volts at the front of the case that blew under the motherboard, across the top of it, directly at the cpu and northbridge, as well as over and under the passive video card (since it was parrellel to the mobo)

it worked very well. and was remarkably silent. i thought having a fan up in front of the case with nothing to block sound from the user would be loud no matter how fast the fan was. but its actually silent.

every time i build a system this is the cooling concept i have in mind when picking out components now.

thinking to myself "how can i get everything here to easily fit in the same airflow from a single fan".

BTX motherboards will help alot so that the hot side of a graphics card will be next to the CPU. so you can cool both with the same fan rather easily. with standard ATX motherboards you either need a riser card to move the video card over the cpu region, or a heatpipe cooler to move the heat into the cpu region.

i'm also a huge proponent of DC/DC ATX PSU's. specially with cpu's becomming more and more energy concious, its becomming very easy to use these nifty little devices for just about any system.

my only major issue with them though, is the ease to find good 200+ watt external AC Converters. they can be found. but require alot of modification to work with the DC/DC psu's. also not many of them are fanless, which is another issue. so far the best one ive found are rebadged Delta 220w 12v AC/DC Converters. Dell has some, so does gateway. but their hard to come by, and like i said require some modification to get them to work.

Bluefront · Post by **Bluefront** » Sun Apr 30, 2006 5:09 am

One of the main components of this technique, is the passive vent on the top of the case. It has reduced the temps drastically over the non-vented stock top. I have been experimenting with different designs for a vented top, one which lets out the air, but keeps the sound contained. Here's my latest design....

The new top is a piece of thin plywood, covered with felt, and fitted with a vent from a Braun bathroon vent fan. Inside the cover is an extra sound deflection panel, also felt-covered. It allows the entire area of the vent to be used, while deflecting most of the sound, since it is located right over the intake vent area....which produces the most sound. This has proved to be the most effective top vent design thus far.....looks nice, and doesn't affect the temps.

justblair · Post by **justblair** » Sun Apr 30, 2006 5:32 am

Nice work Bluefront.

I think that this case will give all who see it a few good ideas on building ducts. Not sure that everyone would be able to implement the whole scheme. Its good that we have pioneers on the site working at the extremes.

Loads of innovative ideas in here. I am sure that we will start to see some of them appearing in other members builds over the not too distant future.

Keep up the good work.... I wonder how long it will take before you unveil the next "Absolute Ultimate Air-Cooling Technique"

Bluefront · Post by **Bluefront** » Sun Apr 30, 2006 7:44 am

Thanks....it's difficult to get a good view of this intake duct once it is in place. Here's a view of it showing the over-lap of the nvidia card. I bought this particular card for this project. It's a half-height card (makes the install easier), and it has a heatsink with the fins orientated to take advantage of the duct airflow (from the top downward). Here's a closeup....

I tried this card in a different computer without the duct, but with a fan blowing toward it from about 2" away. With this setup the nvidia runs about 8C cooler.....usually in the low 40's. Watching full-screen video raised the temp to around 48C. I'm new with nvidia cards.....but these temps seem pretty good compared to different posts I've read on card temps.

And let me add this about "the next absolute ultimate"........there won't be anything better, unless a better heatsink is designed similar to the XP120. It has about the maximum number of heat-pipes that will fit over a CPU. It's size is about the maximum that will fit a MB. This heatsink design is perfect for this airflow. The current rage of tall tower-type heatsinks probably would never work like this design, because the airflow through a tall heatsink by-passes the MB.

The single fan output of this design uses ambient temp air to cool everything, something next to impossible with a heatsink like a Ninja. Combine this airflow with a fan control that will go way down to maybe 3-4 volts, and automatically increase speed slightly under load to maybe 5-6 volts in the case of a P4-2.8, and you'll have a mostly silent computer, that could easily handle the hottest CPU.

IMHO.....If I find a fan control system that can do that (without breaking the bank), minor tweaking is the only thing that can improve this design. There is some bling left to do, like a LED temp read-out on the front of the case, maybe a different top design for the case. But the basic design of the airflow will remain unchanged.

Le_Gritche · Post by **Le_Gritche** » Sun Apr 30, 2006 5:39 pm

jaganath wrote:I don't know if you're familiar with English pounds but you certainly can't buy a soldering iron and a resistor for 2 pounds.

I bought that 4.5â‚¬ soldering iron in an electronic store three blocks away.
You are right, I checked and that's not 2 pounds, but 3.1 pounds. Still a bargain compared to the price of that soldered resistor.
The store clerk even gave me some soldering alloy and the 0.08â‚¬ resistor for free to compensate for their being out of stock of potentiometer that day.

Back on topic, I find it hard to comment on your air-cooling technique Bluefront.
You probably did what was the best for your computer, but you don't have a common setup : SFF case, laptop drive and DC/DC converter as a PSU.
In the gallery topic you seemed to have difficulties to cool your 3"5 HDD without an additional cooler. That's a serious drawback even though you solved it by using a laptop drive and a coper sheet. (BTW in what kind of store did you find that sheet ?)
That's why I'm using something similar to IsaacKuo, to ensure my suspended 3"5 drive gets fresh air, as it is the most temperature challenged device in my computer.
I put my ATX PSU out of my case though (I couldn't make a fresh air-duct for it in my old ATX case), and I enjoy the free room above the CPU cooler, as well as the ability to put a fan there, just like you did. But it's exhaust rather than intake and I didn't duct anything, as I still use a CPU fan with a 100 ohm potentiometer + 68 ohm resistor combo for convenience.

IsaacKuo · Post by **IsaacKuo** » Sun Apr 30, 2006 5:49 pm

Le_Gritche wrote:That's why I'm using something similar to IsaacKuo, to ensure my suspended 3"5 drive gets fresh air, as it is the most temperature challenged device in my computer.

Thanks for the plug, but Bluefront is in fact the ultimate master of the basic technique you're refering to--suspending 3.5" hard drives within an intake tube. It's only recently that he started seriously playing with 2.5" drives (which is actually more my style).

I simply happened to be the one to reply to your request for advice, with my ghetto-flavor version of Bluefront's traditionally favored technique. He's experimented with many more variations on that theme than I'll ever do (since I tend to prefer 2.5" drives anyway).

Bluefront · Post by **Bluefront** » Sun Apr 30, 2006 7:22 pm

Le_Gritche.....the thing here that I refer to as "ultimate" is the airflow, not the whole project, certainly not the Aria case. The case really needs to be about 3" deeper. That way a std 3.5" drive could be incorporated into the intake airflow path....easily. Of course when you add more heat to the picture, you'll probably need more airflow.

While I was waiting for my laptop drive to arrive, I pretty much had a std Maxtor drive in the Aria, running at a good temp, using a second fan. But a laptop drive is quieter, and runs cooler....no doubt about it. That copper sheet came from a hobby store.....one that sells various small metal pieces for different crafts.

There are several ways to add a std PSU into this project.....but again the Aria case is just slightly too small. I do think I might fit the guts of the newest 300w SFF Fortron inside this case....tight fit though. And the airflow would have to be modified. It would also fit outside the case, maybe incorporated into a custom top. But the DC/DC converter is a good match to this Aria project. If I ever need more PSU power, I might change the thing around...but I like the result so far as is.

Ren · Post by **Ren** » Mon May 01, 2006 1:03 pm

You dont happen to have the old Aria PSU do you? I need a replacement and I've already voided my warranty by doing the Nexus fan swap. I'd love to buy it off you.

My only other option is somehow making my own PSU similar to yours

Bluefront · Post by **Bluefront** » Mon May 01, 2006 2:34 pm

Yeah I have it.....PM me with an offer. Trade maybe?

nici · Post by **nici** » Mon May 01, 2006 2:52 pm

Very nice system as usual

Makes me want to build an inexpensive, simple system

jaganath wrote: Oh ho ho ho, yes, bottled water is such a con isn't it.....expect for the millions of people in the developing world, for whom it is their only source of clean drinking water.

Bottled water is an extremely flammable subject..

Anyone willing to start a thread before the snowball starts rolling?

Bluefront · Post by **Bluefront** » Sun May 07, 2006 2:36 am

Another top design has tweaked this project to an even cooler and quieter level. I constructed a new slanted top....it ends up about 2.5" taller at the rear, allowing a wide vent opening the width of the case. I think this vent method is the quietest as any exhaust noise goes out the back. This initial design is nice, but can be improved (not enough bling).

I am really surprised virtually nobody has called me out on the claim made in the title of this thread. Just where are you negative pressure fanboys? Or you Ninja users?

adobian · Post by **adobian** » Mon May 29, 2006 12:07 pm

Shouldn't you have the vent in and vent out on opposite side ?