GOAL - Air cooling with no fans (edit: GOAL reached)

Ecological issues around computing. This is an experimental forum.

Moderators: Ralf Hutter, Lawrence Lee

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Mon Feb 22, 2010 3:43 pm

Yet another saving: It's not well known fact that dropping a PC component temperature by 10 degrees DOUBLES it's age. I dropper my CPU's full load temp by 20 degrees (40C now, 69C before) so i suppose i can run it 4 times longer at full speed. Not that i ever do it...

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Tue Feb 23, 2010 2:48 am

I think the chimney idea is a little impractical for me. I'd prefer something like the old Zalman TNN500 case, if I had to go fanless. To bad Zalman ceased development. Otherwise, I'd rather take my chances with higher temps.

Silverstone RV01, RV02 & FT02 claim the "stack effect", wonder how these compare to a stock P183 fanless.

xan_user
*Lifetime Patron*
Posts: 2269
Joined: Sun May 21, 2006 9:09 am
Location: Northern California.

Post by xan_user » Tue Feb 23, 2010 8:04 am

maalitehdas wrote:Yet another saving: It's not well known fact that dropping a PC component temperature by 10 degrees DOUBLES it's age. I dropper my CPU's full load temp by 20 degrees (40C now, 69C before) so i suppose i can run it 4 times longer at full speed. Not that i ever do it...
As wiki would say "citation needed"
If i run my cpu on liquid nitrogen it will really last 500 years? :shock:

lm
Friend of SPCR
Posts: 1251
Joined: Wed Dec 17, 2003 6:14 am
Location: Finland

Post by lm » Tue Feb 23, 2010 8:31 am

How did you measure power use of all those components? RAM etc

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Tue Feb 23, 2010 8:59 am

I have no way to measure them so I googled the common rates of consumption for each part. I was a bit surprised of how much power an optical drive might need in its worst. SPCR have tested few with an average of 15,7W (full speed) but they were not too fast back then. I suppose it's the laser there which is hungry.

My next report will be the undervolted system's temperatures without the chimney - i'm pretty sure i don't need it any more after these new voltages. I'm a bit worried about PWM temperatures in advance, Abit is known to have high temps in power management even thou it's digital.
Last edited by maalitehdas on Mon Mar 01, 2010 11:49 am, edited 3 times in total.

lm
Friend of SPCR
Posts: 1251
Joined: Wed Dec 17, 2003 6:14 am
Location: Finland

Post by lm » Tue Feb 23, 2010 9:14 am

maalitehdas wrote:I was a bit surprised of how much power an optical drive needs. I suppose it's the laser there which is hungry.
I doubt it. Probably rather needing to move a heavy read/write head around and spin a far from perfectly balanced heavy disk.

Hard drives have much lighter read/write heads that have shorter range of movement, and their platters are very well balanced and light.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Tue Feb 23, 2010 11:05 am

lm wrote:
maalitehdas wrote:I was a bit surprised of how much power an optical drive needs. I suppose it's the laser there which is hungry.
I doubt it. Probably rather needing to move a heavy read/write head around and spin a far from perfectly balanced heavy disk.

Hard drives have much lighter read/write heads that have shorter range of movement, and their platters are very well balanced and light.
At least a terrible sound is a proof for that :D I suppose you're right, laser beam is too thin to consume much power. Could be under 1W.

swivelguy2
Posts: 404
Joined: Sun Jul 04, 2004 9:18 pm
Location: Illinois, USA

Post by swivelguy2 » Tue Feb 23, 2010 11:06 am

maalitehdas wrote: I suppose you're right, laser beam is too thin to consume much power. Could be under 1W.
Well under 1W. I suspect the laser in optical drives would consume 10-20 mW to read and 50-100 mW to write.

But that is neither here nor there. I'm looking forward to some pics of your chimney setup!

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Tue Feb 23, 2010 12:17 pm

Image Image
I'm going to decorate it later (maybe with some photography or paint), i'll just look at it for a while to know what would be nicest way to cover it.
The back exhaust vent is totally covered with opacite tape. There's about 20cm between top of the tube and roof.
Latest test results E8400 @ 1,2GHz-1,8GHz 0,8V - DDR2 800 @ 1,8V wrote: Temperatures (C) for each component. Ambient at 21,0C
IDLE with chimney / LOAD with chimney / LOAD wo. chimney
MB 37 / 37 / 39
CPU 36 / 43 / 50
Core1 41 / 47 / 53
Core2 37 / 44 / 50
PWM 53 / 56 / 64
GPU 55 / 60 / 65
1,2m x 12cm chimney on top of P182 affect (C): MB -2 CPU -7 PWM -8 GPU -5
That's still half of what I reached in my overclocked test (with 3,6GHz), and still more than I expected. The chimney really works, it's not only for the looks :P

OddSilence
Posts: 65
Joined: Wed Feb 03, 2010 5:03 pm
Location: Portugal, EU

Post by OddSilence » Tue Feb 23, 2010 8:01 pm

Very nice idea and experiment.

I have an idea for a hipothetical adaptation of your system - without any fan, although I can't recommend you to do it, because it might be dangerous for you or for your computer, ... (electric shock for you, high temps and fire for system and house... :shock: ):

:idea: Your system, taking out the fan of the PSu and instead of the fan, connecting the chimney (or a second chimney) to the exhaust of the PSu.

I don't expect you to try it, i'm just talking about it, wondering if it would work...

Do you think it would work?
it's not only for the looks :P
:lol:


Not much important but in terms of looks, the worse for me is the tape connecting the chimney to the case.


I imagine it won't be easy to move your computer around.
Now imagine if the chimney was made of brick-and-cement !!! :lol:


P.S.: Why don't you use a "kill-a-watt" device to measure the watts of the entire system?

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Wed Feb 24, 2010 11:44 am

I'll consider your idea of ducting PSU to chimney too, thanks. I think it would work, since temperature drops are that good with other components. When aiming for passive, man should go all the way, right?

Unfortunately I don't own a kill-a-watt, maybe some day i'll get it.

OddSilence
Posts: 65
Joined: Wed Feb 03, 2010 5:03 pm
Location: Portugal, EU

Post by OddSilence » Wed Feb 24, 2010 1:23 pm

I thought a bit more about the "chimney on the psu":
Your psu is a 12cmFanAtBottom one, so it is built assuming an airflow against its components (namely it's heatsinks), not through them. I think a 8cmFanAtBack PSu would be more appropriate for attempting it.

And although i would find very interesting to see the results of your system with the second chimney , I ask you really, don't do it,
it presents too many hazards/risks... (specially if you leave your computer on without "adult supervision" !! :shock: )

About the "kill-a-watt", when/if you decide to buy one, be carefull because I first asked in one store and they said they could order one for 70€!!! :shock:
I decided to look elsewhere, after some time i found one for 16€ and ended up buying one for ~12€... :D (not saying they have the same quality, but it is good enough for what i wanted)

cordis
Posts: 1082
Joined: Thu Jan 15, 2009 10:56 pm
Location: San Jose

if truly fanless

Post by cordis » Wed Feb 24, 2010 2:15 pm

So if the system really has no fans (except the psu), would it make sense to just take off the side panels and let as much air through as possible? I suppose you might get some more noise from the hard drives, although with some suspension you could get rid of that, for the most part. I guess with a giant chimney, going for full convection cooling would make it necessary to keep the panels on to channel the air flow. But wouldn't pulling off all the panels be a little more practical? Or is that too Skeleton-y?

OddSilence
Posts: 65
Joined: Wed Feb 03, 2010 5:03 pm
Location: Portugal, EU

Post by OddSilence » Wed Feb 24, 2010 4:15 pm

@Cordis: Obviously only maalitehdas can answer for himself,
but I think it wouldn't be good to open the side panels (wouldn't it loose the channel efect? and make the chimney pull air from outside of the case instead of through the components, loosing it's efectiveness?)
But it might be good to open the bottom (or opening a big hole on the bottom), although I wouldn't do it because i hate cutting metal, and I think it would be worse in terms of dust collection...

lm
Friend of SPCR
Posts: 1251
Joined: Wed Dec 17, 2003 6:14 am
Location: Finland

Post by lm » Wed Feb 24, 2010 9:49 pm

maalitehdas wrote: Unfortunately I don't own a kill-a-watt, maybe some day i'll get it.
I got something similar from Clas Ohlson for something between 15 and 20 eur.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Re: if truly fanless

Post by maalitehdas » Sun Feb 28, 2010 3:52 pm

cordis wrote:So if the system really has no fans (except the psu), would it make sense to just take off the side panels and let as much air through as possible? I suppose you might get some more noise from the hard drives, although with some suspension you could get rid of that, for the most part. I guess with a giant chimney, going for full convection cooling would make it necessary to keep the panels on to channel the air flow. But wouldn't pulling off all the panels be a little more practical? Or is that too Skeleton-y?
That would have a clearly negative effect, both in HDD noise and reducing the airflow given by chimney. The whole meaning of case is to boost and direct the airflow, but i know there are few cases on the market which does exactly the opposite.

lorn
Posts: 18
Joined: Fri Jan 11, 2008 3:07 pm

Post by lorn » Sun Feb 28, 2010 4:42 pm

maalitehdas wrote:UNDERVOLTING DOESN'T SLOW YOUR PC and doesn't cause any risks
While investigating PHC on Linux, a while ago, I read that it could in fact damage your hardware. I can't find that article/mail anymore but I remember that the explanation involved some component turning into a current sink and getting fried in the process and that the experiment was reproduced on two motherboards.

So thread lightly. If anyone can dig up that story please post a link, thanks!

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Post by ces » Sun Feb 28, 2010 4:53 pm

maalitehdas

The Socratic dialog is too subtle for me. I am only sort of getting what you are saying.

Would you be kind enough to explain exactly what you did and on what exact system in a single message? And what you are now doing?

Some pictures would certainly be nice. Seems like you are doing something noteworthy. But I am just having trouble following what exactly it is.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Mon Mar 01, 2010 11:38 am

ces wrote:The Socratic dialog is too subtle for me. I am only sort of getting what you are saying.

Would you be kind enough to explain exactly what you did and on what exact system in a single message? And what you are now doing?

Some pictures would certainly be nice. Seems like you are doing something noteworthy. But I am just having trouble following what exactly it is.
Thanks for your question ces. It really might get a bit confusing with all the posting every time I change something. My next goal is to remove PSU fan. It's already switched to Scythe s-flex 12cm and it's the only one still running with about 5V. Since it's not cooling any other components (PSU in its own chamber), my cooling may be called passive but my system isn't totally passive because of that. Here's a short review how I got to this point, more detailed information can be found on my general gallery post.

1. P182 and 5 fans in positive pressure setup, overclocked
Image
- CPU E8400 @ 3,6GHz, voltage 1,125V-1,225V
- MEM DDR2 @ 667MHz, voltage 1,9V
- CPU fan 12cm noctua @ 120rpm-640rpm
- SYS fan 12cm nexus @ 360rpm-640rpm
- CASE fans 3x 12cm nexus @480rpm-640rpm
- temperature (STRESS TEST average) 52,3C

DONE HERE: removed all the case fans

2. P182 with 2 fans in neutral pressure setup, overclocked
Image
- CPU E8400 @ 3,6GHz, voltage 1,125V-1,225V
- MEM DDR2 @ 667MHz, voltage 1,9V
- CPU fan 12cm noctua @ 150rpm-640rpm
- SYS fan 12cm nexus @ 360rpm-640rpm
- temperature (STRESS TEST average) 60,7C

DONE HERE: Overclocked my memory. Added a 1,2m x 12cm chimney on top of the case.

3. P182 with 2 fans and a chimney, neutral pressure setup, overclocked
(same image + chimney on top)
- CPU E8400 @ 3,6GHz, voltage 1,125V-1,225V
- MEM DDR2 @ 800MHz, voltage 1,9V
- CPU fan 12cm noctua @ 150rpm-640rpm
- SYS fan 12cm nexus @ 360rpm-640rpm
- temperature (STRESS TEST average) 53,7C

DONE HERE: Removed all fans from upper chamber. Modded PSU by changing original ADDA fan to Scythe S-flex. Underclocked CPU. Blocked back exhaust vent and some other minor holes high in the case with transparent tape.

4. P182 with chimney only, neutral pressure, underclocked
Image
- CPU E8400 @ 1,8GHz, voltage 1,125V-1,225V
- MEM DDR2 @ 800MHz, voltage 1,9V
- temperature (STRESS TEST average) 57,5C

DONE HERE: Undervolted CPU and MEM to minimum possible

5. P182 with chimney only, neutral pressure, underclocked, undervolted
(same image)
- CPU E8400 @ 1,8GHz, voltage 0,8V
- MEM DDR2 @ 667MHz, voltage 1,8V
- temperature (STRESS TEST average) 46,1C
(for comparison: build 5. without a chimney, average stress temperature 50,9)

TO BE DONE: remove PSU fan, find a solution for cooling it otherwise. If that's not possible, change to originally fanless PSU to achieve total passive system.

TO BE DONE AFTER THAT: See how much performance can be achieved with a totally passive chimney solution, setting maximum temperature to some reasonable level (CPU max 65C, GPU max 75C, PWM max 85C maybe?) If you have a passive build or only PSU fan spinning like me, please visit this thread (in "silent front"). If you're interested in more precise temps / fan speeds, please dig my general gallery post and this post from beginning.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Final look

Post by maalitehdas » Tue May 25, 2010 10:47 pm

Final looks
Image
P182 is located behind the TV, you can see the chimney painted to same color with the wall. The system has been running almost 4 months now so now I can really say it's working - and the TV is the noisiest part in the photo (with coil whin and electric hum)

Details of my home theater:
Computer: P182, original fans replaced with a chimney (natural airflow)
Image from inside the case in this post
Full specs of all the components in this post
Link to this post (image of the whole setup)
Television: Sony KDL42W4500
Stereo Amplifier: Pioneer A-676
Surround Amplifier: cheap Logitech
Front stereo Speakers: Gradient 1.3
Front middle speaker: cheap logitech
Back L/R speakers: Forvoice 7.7
Subwoofer: cheap Logitech, rarely used




As you can see, not too much of boxes and nothing too expensive exept the speakers, but they were all bought as used. No antenna either, TV image and internet are coming via optical fiber. It's nice to have a traditional stereo amplifier for computing and listening music - Now I turn on a whole surround system only when i'm watching movies with surround sound. May I call my system green?
Last edited by maalitehdas on Thu May 27, 2010 2:39 am, edited 1 time in total.

OddSilence
Posts: 65
Joined: Wed Feb 03, 2010 5:03 pm
Location: Portugal, EU

Post by OddSilence » Wed May 26, 2010 8:03 am

Great, looks very nice (the chimney looks better painted, obviously :) )

I'm glad it works well (i agree that the fact that it is still working is a good proof) and that you achieved your goals.

8)

KayDat
Posts: 222
Joined: Wed Aug 15, 2007 7:37 pm
Location: Melbourne, Victoria, Australia

Post by KayDat » Wed May 26, 2010 3:29 pm

Fascinating idea...perhaps I should try it myself. =]
Where on earth did you get such a massive cardboard tube though?
Also, just to note, technically I think your setup would be considered negative pressure, since the chimney causes air to be drawn in through air pressure difference. If you had fans blowing into the case at the same rate as air is being drawn out of the chimney, then it would be considered neutral.
But still, great achievement on a passive setup.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Wed May 26, 2010 5:52 pm

Where on earth did you get such a massive cardboard tube though?
I asked for tube in a local hardware stores selling floor carpet. The tube inside the carpet roll was rubbish for them, so I just got one for free and cut the lenght (1,2m) with a saw.
technically I think your setup would be considered negative pressure, since the chimney causes air to be drawn in through air pressure difference.
You're right about the sucking effect but the pressure is (and must be) the same inside the case and in the room. Intake vents must also be big enough to maintain the neutral pressure. If the pressure goes negative, the chimney starts to suck slower or even stop sucking, and the cooling wouldn't work.

Some additional physics of heat and pressure:

In houses with electric air contidioning and a fireplace with chimney a pressure difference causes a true problem sometimes - the smoke comes inwards. It happens when the pressure inside is smaller than outside. Same thing might happen even without electric air contitioning when weather is bad. At some point the pressure grows big enough (either by heat or enough of replacement air) and the chimney starts to work normally.

In computer cases the heat first causes positive pressure (expanded air) which tries to get out from the case wherever it can (pressure inside the case want's to achieve balance with pressure outside the case). Since the air is expanded, it's also lighter - so first direction is always upwards. If there are holes located higher and lower, the positive pressured air is exhausting from the higher one. After the balance in pressure is achieved, the flow of lighter air still goes on and tries to cause negative pressure. Replacement air starts to fill the gap (pressure ourside the case want's to achieve balance with pressure inside the case). Eventually all this physics cause a neutral pressure into the case (as long as the vents are big enough) and the heat remains the only thing that keeps the progress running. In all passive cooled systems this is the situation.

To maintain difference in pressure (in not sealed system) some force must be used all the time. Difference in pressure is againts the laws of nature.

In a chimney system positive pressure (caused by heat) is needed to start the flow - and the longer the pipe, the bigger the needed pressure. That's because there's always a natural pressure difference in the low and high ends of the pipe and that difference must be beaten. After the flow starts the system balances itself to neutral pressure. I suppose there is an optimal length for the chimney too, I don't know that. However please don't experience with this build with too long pipe cause your case temperature might rise too high before the flow is able to start.

A question:
Has anyone ever tried a passive water cooling system? (without an active pump). It would be nice to know if natural convection also works with water in practice, especially if it works even better. I've also heard about a case totally filled with liquid, how's that?

KayDat
Posts: 222
Joined: Wed Aug 15, 2007 7:37 pm
Location: Melbourne, Victoria, Australia

Post by KayDat » Thu May 27, 2010 1:34 am

Most immersion systems I've seen consist of a fish tank full of cooking oil. For the average enthusiast, cooking oil is probably the most cost effective dielectric liquid you can have access to.

Edit: Wikipedia has an approximation equation for stack effect flow rates, which you can use to find height if you solve for h instead.
Image
Assuming worse-case discharge constant Q of 0.65 (Not sure what this is, I just plugged in a value that wiki suggested), flow area A of 0.01131m² (cross sectional area of 120mm fan vent), average case temperature Ti of 313.15K (40°C/104°F) and room temp Te of 293.15K (20°C/68°F), with a desired flow rate Q of 0.011389m²/s (~24CFM or 41m³/h, the airflow of a Noctua NF-S12 running at 600RPM), you would need a chimney that is ~1.79M (~5.87') high.
If we apply the equation to your computer, again assuming computer case and room temperatures to be 40°C and 20°C respectively, you would get around 34m³/h (~20CFM) flowing up the chimney. Of course, the greater the temperature difference, the higher the airflow.
Last edited by KayDat on Thu May 27, 2010 3:03 am, edited 5 times in total.

Klusu
Posts: 196
Joined: Thu Jul 24, 2008 10:57 am
Location: Riga

Post by Klusu » Thu May 27, 2010 2:24 am

maalitehdas wrote: the pressure is (and must be) the same inside the case and in the room..
Wrong. If it was, there would be no reason for the air to enter the case.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Thu May 27, 2010 3:24 am

Klusu wrote:
maalitehdas wrote: the pressure is (and must be) the same inside the case and in the room..
Wrong. If it was, there would be no reason for the air to enter the case.
You're right - but when the replacement air enters the case from big intake vents the pressure is soon balanced. Then the chimney causes negative pressure again, then it balances again and so on... In the end, the difference in pressure is so small inside the case that I suppose you can call it neutral in practice. The sucking effect in chimney setup is caused by lower air pressure in top of the room (near roof) compared to higher air pressure around the intake vents. The heat causes this flow to go faster.

Somewhere in my case, amid all this airflow, must be a Neutral Pressure Plane (NPP). Above this theoretical plane, the air pressure is slightly positive compared to the room air pressure and is trying to force its way out of the case. Below the plane it is slightly negative and the case is trying to draw air in. The location of the NPP can constantly change in response to changing conditions (for example heat rise inside the case). When NPP is located in the chimney, my case is slightly negative pressured. When it's in the case, there are both positive and negative pressures there. When it's in the level of intake vents, my system is positive presurred. Anywhere it might be, the pressure can be called neutral in my opinion.

For KayDat: Thanks for the formula. Total height from the intake vents to top of the tube is 150cm in my system, so I suppose I'm pretty close to that much of airflow. Nice to know.

I also edited the last picture post, now there are links for inside picture and system specs too.
Last edited by maalitehdas on Thu May 27, 2010 3:52 am, edited 1 time in total.

Klusu
Posts: 196
Joined: Thu Jul 24, 2008 10:57 am
Location: Riga

Post by Klusu » Thu May 27, 2010 3:48 am

maalitehdas wrote:lower air pressure in top of the room
Wrong again (although it is a little lower, like it is just 1/3 on the top of Everest)

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Thu May 27, 2010 4:00 am

Klusu wrote:
maalitehdas wrote:lower air pressure in top of the room
Wrong again (although it is a little lower, like it is just 1/3 on the top of Everest)
Please explain how you see the situation and where you think the NPP is?
In my opinion: When the computer is off, I suppose it's exactly 75cm down in the tube. When there's 46 degrees celsius inside the case, I suppose it's close to the level of the heat sources.

Klusu
Posts: 196
Joined: Thu Jul 24, 2008 10:57 am
Location: Riga

Post by Klusu » Thu May 27, 2010 4:20 am

The chimney creates some positive pressure in it's top and negative in it's bottom. The case can be considered part of the chimney. When off (and cold), the same pressure everywhere.
But the idea is good. Some time ago I was thinking about making an intake tube from outside. Would not work in summer.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Fri Aug 20, 2010 7:47 am

Hi all, I just found a wattage meter in our library! They seem to borrow a lot of things among books nowadays... I borrowed it and got some true watt readings for my chimney-PC

PC at idle (windows+startup programs only): 73W
PC at normal use (browsing, office work etc): 78W
PC at 100% burn of CPU+MEM+PSU+cache+HDD: 98W

I have no idea if these readings are any green. I checked our laptop also (Compac 6820s): It took 30W-57W with maximum screen brightness

Watching HD-movies with big screen and great sound isn't green:
My PC+Sony 42"+Surround 5.1 (big speakers) together takes 275-295W

Post Reply