WoodenPC: Passively air cooled DIY HTPC

garacs1 · Post by **garacs1** » Mon Jan 22, 2007 3:08 pm

Hi, just wanted to get your comments on my ongoing project, an HTPC with a DIY wooden case (not exactly wood, but pressed carton... just the same

) All the case will be completely passively air cooled, not even one tiny fan will be used to cool the cpu or whatsoever component. Here's the specs:

Celeron 2,6GHz s.478
512MB DDR PC3200
3GB notebook HD
P4-ITX motherboard, mini-ITX factor
shuttle case 90W fanless and frameless psu (not installed yet)

To passively cool the cpu, I am using a Thermalrock 3-heatpipe 80mm-fan heatsink (Silent Torch, p/n RC-PM0020), with the fan removed. The motherboard has integrated VGA connector (SiS chipset), LAN and 5.1-capable audio.

Here's some pics of the almost final results, case is almost finished by now.

Here you can see also the shuttle psu, in it's final position. Picture taken just to take measures.

Pics with case standing. The case will assume this position when finished:

Here's a "mad-shot": I overclocked temporarily the cpu via cpucool, just to see if it was possible... You can see idle temps in the screen shot:

The temperatures: with case still opened, reproducing DiVX (a 700MB DiVX taken from a fanless NAS I build myself using an old fanless P55 200MHz motherboard/cpu insinde a pressed carton 2-stories DIY case) the CPU temp never went above 60Â°C, while closing the case (how it should be and was planned to be...) the temperature drops down to 54Â°C max. CPU usage is 45-65% average during DiVX reproduction. Needless to say, no sound comes from the HTPC, except for the slight "hum" from the laptop 3GB hd, which I intend to make noiseless further on.

Please tell me what you think of my passively woodenPC HTPC. Any comments are appreciated.

Regards.

Garacs1

jaganath · Post by **jaganath** » Mon Jan 22, 2007 3:36 pm

What a fantastic build. It certainly makes a change from yet another P180/150 with Seasonic PSU and a Ninja! Even more impressive is the fact that the CPU you have chosen is quite hot, yet the temps are very reasonable! Of course the top of the case will be left open, to allow hot air to escape upwards?

True fanless PCs have sort of gone out of fashion at SPCR, which in some ways is a shame because it's never been so easy, what with 35W X2's and the new E4300 from Intel.

garacs1 · Post by **garacs1** » Mon Jan 22, 2007 4:09 pm

jaganath wrote:What a fantastic build. It certainly makes a change from yet another P180/150 with Seasonic PSU and a Ninja! Even more impressive is the fact that the CPU you have chosen is quite hot, yet the temps are very reasonable! Of course the top of the case will be left open, to allow hot air to escape upwards?

True fanless PCs have sort of gone out of fashion at SPCR, which in some ways is a shame because it's never been so easy, what with 35W X2's and the new E4300 from Intel.

Thanks for your comments

I cannot get my hand on a P180/150 for the moment, here in Italy they are not so easy to find. But even if I were able to, I wouldn't just the same. I pretty don't like much that case, or just cases in general. I am of the minimalist side: small, nifty, almost invisible. Yet powerful and well cooled. I am fond of watercooling (I am an Italian distributor and online reseller of watercooling products) but for this project in particular, I just wanted an hassle-free HTPC. Liquid cooling is, imho, the best cooling you can get for any PC, as simple as that (I repeat, is only my personal opinion, don't want to open flames...) but a passively cooled HTPC was on my mind since ages.

The CPU is hot, as you say. Some 62.6W are coming from that CPU at full throttle... The Thermalrock heatsink seems to be handling them pretty well, though.

As you managed to figure out, the upper part will be opened, but not as wide open as you see now. Right on top of the heatsink heatpiped body I'll create a small 80mmx80mm funnel, creating a somewhat "suction" effect for the hot air coming from the fins. All around this "funnel" it will be closed, except for small vent holes for the rest of the components.

The idea behind this project is the Venturi effect: fluid acceleration caused by a costriction in the forced-fluid pathway. The fresh air is coming from below, it will be channeled with some small panels positioned at a 30Â° angle respect to the direction of the air path, directly to and trough the fins of the heatsink. Air passing through the fins will be forced to accellerate, due to the restriction, thus cooling the fins a little bit faster, just to be finally exhausted by the top of the case, through a constant-area section diameter funnel, right above and directly connected to the fins.

To get that effect, of course, a front panel will be applied to the case. And the lower panels creating the restriction must be still made and put in place.

BTW, I chose the Celeron 2,6GHz S.478 Northwood Core CPU because it was lying around on my desk... right aside of the mini-ITX Insight P4-ITX motherboard...

Garacs1

McBanjo · Post by **McBanjo** » Mon Jan 22, 2007 4:16 pm

A totaly passive computer is technically very impressive. But with a fan at 5V you get so much bether result and yet it's silent. So it's more a bragging right than needed thing

But impressive build non the less. Nice to see a woodcase

My next fully positiv airflow will be ready in 1-2 weeks

garacs1 · Post by **garacs1** » Mon Jan 22, 2007 4:21 pm

McBanjo wrote:A totaly passive computer is technically very impressive. But with a fan at 5V you get so much bether result and yet it's silent. So it's more a bragging right than needed thing

But impressive build non the less. Nice to see a woodcase
My next fully positiv airflow will be ready in 1-2 weeks

I know a 5-Volted fan does a pretty good job, but that's not my point. I didn't plan a passively air cooled case just because I didn't have around a fan to downvolt. It's the concept that's behind passive cooling that tickled my mind. I studied hard and came up with this project. Since it's working perfectly well without fans, I prefer a no-fan HTPC than a 5V-fan HTPC... no pun intended

I was not looking for the best results at all... I could have built a liquid cooled HTPC for that, which I already did...

jaganath · Post by **jaganath** » Mon Jan 22, 2007 4:32 pm

A totally passive computer is technically very impressive. But with a fan at 5V you get so much better result and yet it's silent. So it's more a bragging right than needed thing

Well, technically anything with moving parts cannot be silent (ie no pressure vibrations in the air). But yes, depending on your hard drive noise a fan at 5V might be inaudible (which is a different thing). The thing is that Garacs has made a fanless PC without spending a lot of money; this disproves the conventional thinking that a fanless PC has to be expensive; in fact with a cooler CPU (for example undervolted Sempron) it would probably be trivially easy to make one.

Garacs, I look forward to hearing the results of the venturi tunnel.

garacs1 · Post by **garacs1** » Mon Jan 22, 2007 4:37 pm

BTW, if some of you want to see all the pics I took for this project, I'm writing a worklog on my Italian Forum, Freemodding.it. Here's the link:

[F.M.C. 2006] WoodenPC - Passively AirCooled Wooden HTPC

The case is taking part on a FreeModding Contest, due for February 19, 2007 (that's what the [F.M.C. 2006] stands for).

There you'll find pics of the various steps, ideas, errors I made creating this project. Have a nice read.

Garacs1

McBanjo · Post by **McBanjo** » Mon Jan 22, 2007 4:42 pm

I didn't plan a passively air cooled case just because I didn't have around a fan to downvolt.

hihi you didn't?

The more challanging the more fun. As I stated, it's very impressive build. And I'll look forward to that venturi tunnel.

[quote"jaganath"]Well, technically anything with moving parts cannot be silent [/quote]
I stand corrected

garacs1 · Post by **garacs1** » Mon Jan 22, 2007 4:43 pm

jaganath wrote:
A totally passive computer is technically very impressive. But with a fan at 5V you get so much better result and yet it's silent. So it's more a bragging right than needed thing
Well, technically anything with moving parts cannot be silent (ie no pressure vibrations in the air). But yes, depending on your hard drive noise a fan at 5V might be inaudible (which is a different thing). The thing is that Garacs has made a fanless PC without spending a lot of money; this disproves the conventional thinking that a fanless PC has to be expensive; in fact with a cooler CPU (for example undervolted Sempron) it would probably be trivially easy to make one.

Garacs, I look forward to hearing the results of the venturi tunnel.

You bet I'll post results here too, jaganath!

For the costs involved: 150â‚¬ circa for the MB, 50â‚¬ for the cpu (both of them bought some 2 years ago, prices given are actual, though: I spent a lot more when I bought them...), 30â‚¬ for the 512MB stick, and 30â‚¬ for the Thermalrock heatsink. That's 260â‚¬ in total. Add some 10â‚¬ for the pressed carton panel, some elbow-grease (priceless...), and there you have it. For less than 300â‚¬ (I added also a 15â‚¬ usb bluetooth mini-mouse, and do not use a keyboard) I am building myself a fanless demi-powerfull HTPC. Oh, btw, the Windows XP PRO installation will be replaced with a linux GEEKBox distribution. Just to keep software costs down to zero...

BillTodd · Post by **BillTodd** » Tue Jan 23, 2007 5:35 am

Nice

But, please be careful operating that PSU without the covers !

I susgest you look into HDD enclosures, since the HDD will be the noisiest thing in the box.

garacs1 · Post by **garacs1** » Tue Jan 23, 2007 6:02 am

BillTodd wrote:Nice

But, please be careful operating that PSU without the covers !

I susgest you look into HDD enclosures, since the HDD will be the noisiest thing in the box.

Hi BillTodd, thanks for the advice

I will be careful, do not worry.

For the HD, I preferr not using HD enclosures, since they only would take space. I rather will "encapsulate" the hd in some foam or other damping material of some sort.

Thanks for passing by

klankymen · Post by **klankymen** » Tue Jan 23, 2007 9:41 am

just wondering, are you running the OS off the USB-Stick?

Or off the Hard-drive? If one, than why are you using the other for?

By the way great build, haven't seen something like this for a while (since that australian who's name I cant remember had a computer attached to the back of his desk

)

EndoSteel · Post by **EndoSteel** » Tue Jan 23, 2007 12:59 pm

garacs1
Why use an Infinity? Ninja outperforms it by ~5C in convection mode.

P.S. Good job anyway

.

TMM · Post by **TMM** » Tue Jan 23, 2007 1:46 pm

Nice build, looks... quiet

EndoSteel wrote:garacs1
Why use an Infinity? Ninja outperforms it by ~5C in convection mode.

P.S. Good job anyway .

Its not a Scythe, its a Thermalrock

garacs1 · Post by **garacs1** » Tue Jan 23, 2007 4:55 pm

klankymen wrote:just wondering, are you running the OS off the USB-Stick?

Or off the Hard-drive? If one, than why are you using the other for?

By the way great build, haven't seen something like this for a while (since that australian who's name I cant remember had a computer attached to the back of his desk )

The USB stick you see is the bluetooth usb receiver for the mouse. The OS is running from the small 3,3GB laptop drive which will be included inside the case, right underneath the motherboard: that's why I used standoffs that high.

Thanks for your positive comments, I appreciate them very much!

Stay tuned, you'll soon see the upcoming steps.

Garacs1

garacs1 · Post by **garacs1** » Tue Jan 23, 2007 4:56 pm

EndoSteel wrote:garacs1
Why use an Infinity? Ninja outperforms it by ~5C in convection mode.

P.S. Good job anyway .

Thanks for your comments, but as TMM replied, the heatsink I am using is a Thermalrock. It's good nonetheless

garacs1 · Post by **garacs1** » Tue Jan 23, 2007 4:58 pm

TMM wrote:Nice build, looks... quiet

EndoSteel wrote:garacs1
Why use an Infinity? Ninja outperforms it by ~5C in convection mode.

P.S. Good job anyway .
Its not a Scythe, its a Thermalrock

Thanks TMM for your comment

It IS quiet indeed

The project is almost finished by now, so stay tuned to this thread, you'll see the final steps very soon...

Garacs1

garacs1 · Post by **garacs1** » Thu Jan 25, 2007 3:06 pm

Here I am again with the new pics and steps for the WoodenPC.

I inserted a panel inside the case to separate the motherboard zone from the heatsink-fins zone. This panel also helps me fitting the PSU inside the case. I'll later use some small-diameter screws to fix the center panel to the side panels. Here are the pics:

The panel, cut to measure to fit the heatpipes:

How the panel fits through the heatpipes:

The inside panel fitted, notice it is shorter than the side panels, to permit airflow:

This image shows how the inside panel creates two separate chambers:

And finally, this is how the psu will be placed inside the case:

As you can see, I used another smaller vertical panel to separate the pcb of the psu from the aluminum fins of the heatsink: this is to avoid short-circuits, obviously, but also to funnel fresh air better to the fins.

Stay tuned for new updates, shortly

Garacs1

flyingsherpa · Post by **flyingsherpa** » Thu Jan 25, 2007 4:04 pm

garacs1 wrote:
The idea behind this project is the Venturi effect: fluid acceleration caused by a costriction in the forced-fluid pathway. The fresh air is coming from below, it will be channeled with some small panels positioned at a 30Â° angle respect to the direction of the air path, directly to and trough the fins of the heatsink. Air passing through the fins will be forced to accellerate, due to the restriction, thus cooling the fins a little bit faster, just to be finally exhausted by the top of the case, through a constant-area section diameter funnel, right above and directly connected to the fins.

Garacs1

i wish you luck, but i don't think that will work very well. as you state, the venturi effect is caused by restricting a forced-fluid flow, and thermo-buoyant flow just doesn't have much 'force' behind it (really pressure). but if it works fine passively right now, it doesn't really matter if you don't get much venturi effect. i'll be interested to see your results

McBanjo · Post by **McBanjo** » Thu Jan 25, 2007 4:33 pm

Can you run the computer safely without the panel? Would be intresting to see the effects in numbers, primary temperatures then

garacs1 · Post by **garacs1** » Thu Jan 25, 2007 4:58 pm

flyingsherpa wrote:
i wish you luck, but i don't think that will work very well. as you state, the venturi effect is caused by restricting a forced-fluid flow, and thermo-buoyant flow just doesn't have much 'force' behind it (really pressure). but if it works fine passively right now, it doesn't really matter if you don't get much venturi effect. i'll be interested to see your results

Hi flyingsherpa, and thank you for your comments. I know I misused the terms "Venturi Effect", but that is what is nearest to the effect I would like to obtain with my build. To be precise, and you name it, it is barely a thermobouyant based rig: natural convection will make hot air rise, sucking from the lower opened side fresh air. The Venturi effect I wanted to create is still a principle this case is bounded to, even if natural convection of air will not cause any real acceleration to the air flow. For that, I should put a fan in the lower side of the case, pushing air upwards. But it makes no sense to me since I wanted a fanless pc case....

garacs1 · Post by **garacs1** » Thu Jan 25, 2007 5:01 pm

McBanjo wrote:Can you run the computer safely without the panel? Would be intresting to see the effects in numbers, primary temperatures then

Well, I made some tests yesterday night, *without* panel. CPU temps during DiVX play never went above 55Â°C, with room temperature @ 24Â°C constantly. I'll repeat tests with and without inside panel this weekend, taking pictures and screenshots. What buzzes me is the hd noise, now... gotta find a way to mute it definitively

Anvar · Post by **Anvar** » Fri Jan 26, 2007 1:07 am

garacs1 wrote:Well, I made some tests yesterday night, *without* panel. CPU temps during DiVX play never went above 55Â°C, with room temperature @ 24Â°C constantly. I'll repeat tests with and without inside panel this weekend, taking pictures and screenshots. What buzzes me is the hd noise, now... gotta find a way to mute it definitively

A good sharp hit with a hammer will do that.

Or lots and lots of foam, not as fun though.

albatros_la · Post by **albatros_la** » Fri Jan 26, 2007 4:46 am

Foam will stop the noise, maybe, but it'll raise hd temperature for sure! I think the only big lack in your project is the casing material. It's a good thing to use pressed carton between heatsink and psu, cause it insulates those heat sources, but for external casing I suggest to use aluminium: in that way heat would be dissipated also through the vessel.

BillTodd · Post by **BillTodd** » Fri Jan 26, 2007 4:54 am

The idea behind this project is the Venturi effect:

I think it may be more accurate to describe it as a chimney effect:

i.e. the heated air is lighter, so is pushed up by the cooler air above it. The forces involved are tiny but, if the air is constrained in a long tube, it will accelerate to a reasonable velocity and its momentum will help overcome the drag caused by the closely spaced heat-sink fins.

I've used the effect in my yet-to-be-finished VESA-ITX project

The above heat-sink is far more effective in the tube body below

Here's a slightly larger example
http://www.solar-e.com/secondary/solarthermalenergy.htm

albatros_la · Post by **albatros_la** » Fri Jan 26, 2007 9:27 am

BillTodd: lovely job!

Suggestion: a couple of heat pipes properly soldered would make possible to use the whole aluminium vessel as an heat sink.

jaganath · Post by **jaganath** » Fri Jan 26, 2007 10:11 am

The above heat-sink is far more effective in the tube body below

So you mean there is actually a measurable decrease in temps when the heatsink is enclosed?

BillTodd · Post by **BillTodd** » Fri Jan 26, 2007 11:01 am

jaganath wrote:
The above heat-sink is far more effective in the tube body below
So you mean there is actually a measurable decrease in temps when the heatsink is enclosed?

Well, perhaps far more is an exaggeration

It certainly works well in the enclosure.

albatros_la wrote:BillTodd: lovely job!

Suggestion: a couple of heat pipes properly soldered would make possible to use the whole aluminium vessel as an heat sink.

Yeah, could do, but it doesn't really need it.

EndoSteel · Post by **EndoSteel** » Sat Jan 27, 2007 2:51 am

albatros_la
I suggest to use aluminium: in that way heat would be dissipated also through the vessel.
Ð