Convection radiator design
 

Inspired by this idea.

Its a convective radiator, designed to fit on the side of PC case. I just cobbled it together in Truespace 3.2 as a kind of beginners exercise in 3D. Also because I might try and make one if I ever get round to doing the heatloss calculations…
I just like the idea of not using any fans :)

Looks like it came off the back of a refrigerator. If that setup in your link worked well, this should work better. What are those fins made out of?

A few guys used that approach :

http://clientes.netvisao.pt/carlo001/setup008.jpg

http://de.msnusers.com/_Secure/0PgCz...52548960001667

http://de.msnusers.com/_Secure/0SgAA...52648696800048

Seems to work.
An AC radiator can do the trick :

http://clientes.netvisao.pt/carlo001/__hr_rad1.jpg (of course, something smaller :D).

TerraMex: this is yours?
Did you make it? Does it work!?

The other pics look very similar to a commercial design here.
Do you know anything about that setup? how did you find those pics :confused:

This is of course a good thing

If you are serious have a check on EBay for Car radiators and such.. if you live near them you can pick it up for pratically nothing... I would have bought one ages ago (was marked up at £5) if the P&P wasn't so much, or they refused to send it b/c it was so heavy.

~ Boli

Hmmm. Car radiators/heatercores and A/C coils are designed for forced convection cooling, ie. with fans. Therefore the fins are very closely packed, and I would doubt that relying on natural convection to get the air through them would be very effective.
Thenagain, I suppose this is compensated somewhat by the increased surface area...

What I want is a radiator *designed* for natural convection :cool:
and that won't cost an arm and a leg (the cheapest of those Innovatek radiators is already 100€, and this one is just too small IMO)

Now if I could just get hold of some of that finned aluminium tubing...I wish :cry:

No , just some i saw around.
Might give you some ideas.

The second radiator is a good way to do that , it's made with aluminium pipes with large fins, which are cheap and just about any metal shop has access.
The rest was the manifolds (intake and exit).

Or , you could go with a Bong cooler (aka evaporation tower) , or similar. Search for Volenti's work that you'll find what I mean.

Not given up yet. (Finned aluminium pipe doesnt seem very easy to find, after trying a few of the local metal shops :( )

I've come up with a much simpler design, which could be assembled quite easily. It would also be less fragile, and so more portable :cool:

I'm hoping that if the two sheets are spaced apart sufficiently, the airflow will cool *both* sides of each, yielding roughly 0.7m2 surface area.
Of course, I could stack two of these radiators, or just use one on each side of the PC....

The "grey stuff" is aluminium sheet, probably 0.5mm thick. It would have to be clamped somehow onto the pipes.
Come to think of it I could use copper sheet, (although cost may be an issue) and just solder it to the pipes. :D
It could also be painted black to increase radiation.

I think you'll find that painting it black would decrease performance. At the sort of temperatures we tend to deal with radiation is a minimal factor. Nearly all your cooling would be via convection (either natural or forced air) and thus bare metal is better than paint due to having one less interface. The only reason you see black heatsinks is because they're anodising them anyway so colouring it black or not makes little difference to convection and can help radiation marginally.

I was thinking really thin coat, eg auto spray. You're probably right, though it wouldnt make much difference either way :rolleyes:

Based on this calculator I think I'd need quite a few "layers" of sheets. More like 8, rather than 4...
and even that wouldnt get me less than 10°C above ambient at 80W.
So... :confused:

...just stick some fins on, hehe :D

(watch out Zalman, Innovatek, whoever :D :D :D )

maybe I should just STFU before somebody else uses my idea, lol

With small enough copper tube, you could do away with the fins altogether. Have you considered just making a lattice of 1/4 inch copper pipes (with many parallel channels of course to preseve flow)?

Copper pipe is cheap and easy to work with.

side note : paiting it black wont diminuish it's performance if done properly. A thin layer will do.

A paint coating is an extra interface which will decrease performance no matter how thin it is. No matter how you try and justify it, having to conduct heat through an extra layer of material is always going to be less efficient, regardless of the actual material.
If you want to paint it for looks, go ahead, the difference isn't large. But don't kid yourself it's for performance reasons.

redleader that was pretty much what i was thinking of to start with:) then I did the sums and realised it'd take something like 50 to 100m of pipe ( :eek: ) to make something usefull :shrug:

Butcher, I'm not that certain that a few microns of paint would completely anihilate a hypothetical 5-15% gain in heatloss through radiation ? (yes, hypothetical :D )
Anyone know the thermal conductivity of paint offhand... ?

Pretty low. There's also the issue of conductivity across the paint-metal interface which is also low.
Also I'm curious as to how you get 5-15% of heat loss through radiation, given that a black radiator is going to absorb almost as much radiation as it emits, thus the net gain of it is somewhat minimal.

The radiator is assumed to be 10°C above ambient, so net heat flow is *from* radiator *to* environment.

Out of curiosity, I ran the figures through this

For one side only, at 40°C surface and 30°C amb, heatloss through radiation only:
ALUMINIUM SURFACE: 1.9W
MATT BLACK SURFACE: 8.5W

Black surface emits 6W more, which is 10% of 60W. (CQFD :) )

Edit(addition, whatever:)): Found a practical way of adding fins. Now at 1.95m2 surface area (twice as much as the - ahem - competition :D )

Comments, criticism, ideas and suggestions all welcome (apart from why the aluminium doesnt actually look like aluminium :mad: )

I can't understand the "Why".

Sure, I can understand the idea. But.....

Most computers are under/in a desk, with almost no natural airflow.
You CPU temps will be about 8 to 15 degrees higer as with forced air. (depending on the natural airflow)

If you want to go outside the case, a smaller rad with a fan at low speed will give you the same or better performence.

If you want to go for the "passive" radiator, why not look into the radiators design they use to heat your house?

They come in the right size, they have the right tubing, and they have years of development behind them!

reply to fhorst
 

You said.. I can't understand the "Why".

For the challenge of it, maybe. Or perhaps for more even more quiet than low RPM fans can provide.

I considered trying to build a passive radiator but decided just to go with what has worked for others. I'm using two 170mm fans @900 RPM on an '89 Camaro heatercore for push/pull. Not silent but close enough. Plenty of heat dissipation with low noise. My thanks to those who travelled the watercooling road before me and posted their knowledge online. I'm an imitator but a damn good one! :p

As for passive rads used to heat a house, could you recommend a specific type? All the ones I've seen are ancient, huge, and incredibly heavy compared to a PC. That's one way to get a lot of surface area I must admit... not something you want to bring to a LAN if you do that sort of thing.

A thin layer wont hurt performance, and guess what, black is a better heat radiator than most colours.
A "black body" is one that is a perfect radiator (and the perfect absorver),at a certain temperature. Actually the nature of the surface also makes a difference. Dull, matt black surfaces are the best radiators.
Shiny surfaces do not radiate so well.
Now apply that to a radiator which is hotter than ambient, then it will counter the new interface layer by being better radiator.
If you wish to say it wont really help that much, i'd say fair enough.
But it wont make it worse.

http://207.242.75.40/derbtech/color2.htm

Check the chart.
Things arent painted black of looks. Well, sometimes they are :D.
(http://www.howstuffworks.com/question32.htm)

Hmm it's a bit of a toss up, it will hurt convective heat flow (extra material with low thermal conductivity and extra interfaces) but the radiation, yes black is good at absorbing and dissipating by radiation. In standard fan/rad setups convective heat transfer is without a doubt way more important than "radiation." But in this particular design a radiative heat transfer may be more important although I doubt it. The only way radiation could be primary to convection is if there is no fluid flow around the radiator, but the heat will cause the air to rise and hence you have fluid flow around the radiator, meaning convective heat transfer is still probably the primary heat transfer occuring.

So, will painting it black help or hurt? Depends on a couple things, mainly temperature difference - and other radiative sources of heat. If you have a grow-op with some big grow lamps near your computer, the black surface will, absorb that radiative heat more quickly. Since radiative heat is a quartic function of temperature the difference in radiative source temperature and the object absorbing the temperature makes a huge difference. I doubt quite highly dissipating heat by radiation is going to make a lot of difference.

But will the painting hurt conduction...err convection - yes. A lot? No. But the thickness of the paint does make a huge difference. The temperature gradient in a thin poorly conducting material can be as steep as one in a thick highly conductive material.

I think the overall consensus is that paint doesn't hurt temps enough to justify ugly rads - personally mine's not painted, I'm ugly :cry: . This passive radiator does increase the % at which the heat is radiated but when we're talking < 50deg C the heat radiated isn't all that much (in comparison to 1000deg) Basically the hotter you get the more radiation takes over as the primary method of heat transfer - just due to the physics.

Hopefully this adds some light to your discussion.

Lol. Guess you just have to like designing things and see them come to life :)
Yeah it's a challenge, it's also not like there isn't a market for silent PC cooling...even if the products I've seen so far are slightly out of my budget, and not very portable either. (well, designed to sit on top of a desk probably- valid point there, although mine isn't under a desk ATM)
8-15°C more is still 13-20°C less than my old air heatsink, so I can live with that.
And if it doesnt work out, well I'll just settle for low speed fan+heatercore. Or maybe both, passive rad+forced rad ;)

As for the black paint, I'll be sure to do a before/after test and post the results :p

edit: too many smilies

Convection using any large rad will work, but what you have to realise is it wont work without some significat "temperature differential" factor. In other words you'll see higher temps. Now the rad design efficiency will affect this along with many other real world specifics, but without a decent air current, (fan or another form of artifically created convection going on you'll see coolant temps rise until the convection losses equal the heat input before it stablises. It would be very hard to pre-calculate exactly what temperature this would be on paper. I'm using a large JCB tractor rad, (about 20" X 20" X 2.5" with a 16" Kenlow fan @ 3v and it keeps my coolant at 2C to 3C above room temp, (at 15v it will keep near as damn it to room ambient, but everything ends up on the other side of the room, and my ears start to bleed :p ). If I turn the fan off the coolant will get to around 8C above ambient eventually before stablising.

Here's the rad and fan (I included small items, cd rom & my PC for size indication ;) ):-

http://www.zerofanzone.co.uk/projects/pc/giantrad.jpg

http://www.zerofanzone.co.uk/projects/pc/kenlowe0.jpg http://www.zerofanzone.co.uk/projects/pc/kenlowe1.jpg

Another topic you may want to look HERE at as it looks into ways of creating artificial convection without fans, although for a different reason..

bladerunner,

how loud is that fan at 3V. Does it have any difficulty starting up?

8-ball

so let me get this right.... a black object will radiate energy better than say a white or shiny chrome object... Is it not the other way around... and in "other way around"... I'm referring to a black object absorbing radiated energy ( light for instance) better. I can not see how the color can have any effect on an object's ability to radiate energy.

from first principles, the color of any object is but nothing more than a reference to it's ability to not absorb a particular set of frequencies in the EM spectrum. Thus an object to appear black in color, needs to absorb close to 100% of the energy in the complete visible spectrum that falls on it.