| sardonx |
| I searched but couldn't find anything. Which metal has the best heat conductivity and heat dissapation? I need to build a heatsink. Anyone know? |
|
|
| damonpip |
| Diamond, silver, copper. I cannot find the link that proves this at the minute though. |
|
|
| AMT-freak |
After copper, the list continues with aluminium.
Cheap, easy to work with, non-magnetic, ... There is a reason why everybody keeps using it ;) |
|
|
| Magura |
| quote: | Originally posted by damonpip
Diamond, silver, copper. |
A truth with modifications.
For heat conductivity, this is correct.
For dissipation, it is not correct.
An anodized black aluminium surface will have far better dissipation properties than any of the above, for the simple reason that you can dye it black. Any of the other materials you would have to paint black.
Magura:) |
|
|
| sardonx |
that was a quick response.. thank you.
Aluminum is not a problem to get :) |
|
|
| Bill Fitzpatrick |
| Copper has better conductivity but poorer dissapation than alum. |
|
|
| sardonx |
| so ideally the chips should be mounted to a copper bar (of appropriate dimensions) and the fins made out of black anodized aluminum...? |
|
|
| breguetphile |
| It needs to be said that just because one material is greater than another does not mean it is appropriate for the job. The reason aluminum heatsinks are used so frequently is because they are best 99% of the time. While you did ask for best dissipation and heat conductivity, I'd keep in mind cost and ease of use. |
|
|
| Sch3mat1c |
Saphire is also pretty conductive. And it just so happens that's what anodizing produces on the aluminum surface.
I think aluminum is more conductive than copper by density; certainly by cost.
Tim |
|
|
| Bill Fitzpatrick |
| quote: | Originally posted by sardonx
so ideally the chips should be mounted to a copper bar (of appropriate dimensions) and the fins made out of black anodized aluminum...? |
I'm pretty sure I've seen that done before. Don't forget the thermal resistance between the copper-alum. interface needs to be minimized. |
|
|
| jneutron |
| quote: | Originally posted by sardonx
so ideally the chips should be mounted to a copper bar (of appropriate dimensions) and the fins made out of black anodized aluminum...? |
Copper has about a factor of two greater thermal conductivity than aluminum.
Copper expands at 16.6 ppm/degree C...aluminum, 25 ppm/C
Maintaining good thermal contact for a bimetal assembly, in any significant size unit, is beyond most DIYers. Most would use some kind of thermal compound at the interface and allow sliding compliance for expansion mismatch, which would be counterproductive.
Best bet is to stick to one metal.
Coloring the heat sink black only affects the radiated energy, in accordance with black-body analysis, and is usually not a significant factor if you keep the sink temperature as low as possible. Also, for a fan cooled assembly, the color won't make a measureable difference, again, with the sink as cool as possible.
I tend to use aluminum for the most part, given the ease of machining and tapping, but when my application is high dissipation density, liquid helium use, or when water cooling is involved, I'll go with copper (with water, it's easier to stay one metal and bypass the DI quality concerns and all those Dielectric isolators).
Hockey puck apps are slightly different, as the contact pressure of 1500 to 2500 psi eliminates contact concerns.
Cheers, John |
|
|
| Pjotr |
Hi,
There is not much difference between copper and aluminium for normal heat sinks. Although thermal conductivity differs it is for both low enough compared to the radiation of the heat to the surrounding air. Only for high duty forced air cooling or water cooling the difference may count, like i.e. for CPU coolers.
What matters most is the radiating surface. Bare metals don’t radiate much heat. This has to do with the free electrons at the surface. Look for a good textbook about the basics behind this. You need a non-conductive surface. A painted heat sink is much more effective than a bare metal one, even if it is painted white. White or black doesn’t matter much although black is slightly more effective. So anodised aluminium is still the optimum for regular heat sinks. But it doesn’t need to be black per se. Red, blue or even blank anodised will do also good.
Cheers ;) |
|
|
| jneutron |
| quote: | Originally posted by Pjotr
Bare metals don’t radiate much heat. This has to do with the free electrons at the surface. Look for a good textbook about the basics behind this. You need a non-conductive surface. A painted heat sink is much more effective than a bare metal one, even if it is painted white. White or black doesn’t matter much although black is slightly more effective. So anodised aluminium is still the optimum for regular heat sinks. But it doesn’t need to be black per se. Red, blue or even blank anodised will do also good.
Cheers ;) |
Here is a link showing the equation for calculating the radiation of a surface:
http://hyperphysics.phy-astr.gsu.edu/hbase/hframe.html
Here is a link to the emissivities of many materials..
http://www.electro-optical.com/bb_r...atlemisivty.htm
Here's another, with temp included..
http://www.x20.org/library/thermal/emissivity.htm
As pjotr stated, bare metals are in the mud, radiation wise.
If your app needs that added little radiation factor, painting or anodizing is the way to go.
Cheers, John |
|
|
| rif |
I think you're missing the with blackbody radiation:
the color of a blackbody is determined by the temperature, not the other way around. So hot materials "glow", where the color of the glow is related to temperature. Blackbodies are named that b/c a perfect blackbody absorbs all energy incident on it and then re-radiates it according to very specific formulae -- devrived using statistcal thermodynamics.
If you have a black metal, or red, or green color, that affects its absorption. Don't see how it affect radtion of EM. Painting or dying a metal red doesn't make it red hot.
Black body does not mean black color, just the opposite. Black is the absence...actually defining black is not trivial -- anyone want to try? |
|
|
| jneutron |
What color you paint it affects both the absorbtion and the emission.. one of those links showed actual emissivity values for different color paints...I had looked it up after pjotr stated that metal and electron thing..I didn't understand it. It would appear his statement was correct, from what I found with a simple google.
I was surprised that white paint was more emissive than all the metals..I hadn't expected that.
I make the assumption you are talking to me, btw...
I've not confused the wavelengths being reflected off a body with the emission spectra of an extremely hot one..
Cheers, John |
|
|
| rif |
Huh -- I'll have to check that out. Sorry if my tone was misunderstood -- I was just trying to add to an interesting discussion.
Anyway it doesn't sound quite right to me, but it's been a LONG time since I've done any science. My first guess is that if it's paint we're talking about, maybe it has to do with the paint itself, not the color -- the paint has thickness and is acting like a blanket and keeping heat/radiation in. Different color paints have different chem. compaositions -- maybe it's also reflecting back to the metal before it gets to the outside paint layer.
The best way to tell woiuld be to take lots of samples and take intensity readings through the spectrum at various temperatures for each sample. Plot intensity vs. freq. for a given temp and smaple and you get standard blackbody curve (as much as it acts like one). Plot peak of the intensity curve vs. freq.(color) of samples to see if it depends on sample color.
Is that right? |
|
|
| Sch3mat1c |
Paint is going to disappear at the power levels we're talking about. Maybe 1/2W/in^2 at best? You need some thick, soft paint to do anything to that.
Now, if you have forced air, or worse yet, liquid cooling, a thin layer of hard paint, or just anodizing if possible, would be best.
For convection-cooled heatsinks, a black finish will affect it a bit due to the low rate at which convection cools. With forced cooling, radiation may well have essentially nowhere to go, and certainly, be a negligible part of the power being dissipated.
Tim |
|
|
| uvodee |
| quote: | Originally posted by AMT-freak
After copper, the list continues with aluminium.
Cheap, easy to work with, non-magnetic, ... There is a reason why everybody keeps using it ;) |
however, my worries start when I see people mount copper on aluminum or vice versa...
for some reason I do not trust the contact, there are always forces present (erosion ) and when eroding copper and eroding aluminum get together, I have a strange feeling about that. I hope I'm wrong but, having lived on a boat for a few years and having seen all kinds of metals erode (not corode) I try to keep different metals not in contact with others.
J-P |
|
|
| JonPike |
OK, yes, black anodized Aluminum does have a higher emissivity than bare, polished Al.
But go work it out on the Hyperphysics site with their calculator.. unless we're talking large surface areas, or large heat differences.. it's not that big of a factor.
1 square foot (and that's a lot of heat sink) at a 45C temperature, will radiate 13W (e of .98) compared to the bare, polished Al radiating .5W (e of .038) Sounds like something large, but compared with the conduction cooling of airflow, shrinks into insignificance. You'ld probably use a square foot of heatsink if you were dissapating some hundreds of watts.. then an extra 13 isn't a big deal.
We did tests on some smaller heat sinks, to specifically test this effect, it wasn't measurable (on our 3x5" heatsinks) if there was forced airflow, barely noticable on unforced convection.
It all depends on your delta of temperature, and area of heat sink.. on the Aavid site a paper states with natural convection (no forced air) radiation can contribute up to 25% of the cooling. Would seem the radiation into the air helps move more of it for more convective flow.
On the other hand, I agree with the comments that paint and such will probably put a low condudtivity coating around your nice highly conducting Al, thereby insulating it from the air more than any emissive effect could help. |
|
|
| rif |
Ah -- I see the other disconnect now! When speaking about blackbodies and radiative effects, I assume no convection at all -- forced or otherwise. ie it's in a vacuum.
No one wants to take a crack at defining black? |
|
|
| Magura |
| quote: | Originally posted by uvodee
however, my worries start when I see people mount copper on aluminum or vice versa...
|
And with good reason. You will have to machine or grind the mating surfaces to a surface finish better than Ra 0.15 to avoid problems....and even then its still a matter of choosing the right alloys, else its still the same problem.
Magura:) |
|
|
| CraigBuckingham |
Black is a measure of how much an object or surface absorbs light. Using the human eye as the detector then it would be the visual spectrum we would be talking about. So a theoretical pure black object has 100% absorption of light in the visual spectrum. Given that the eye does not have equal sensitivity at all wavelengths in the visual range then an object could appear to be perfectly black but still could be reflecting a small percentage of light iin the near IR and near UV ends of the visual range.
A balck hole is has some similarities. Nothing including light in the visual spectrum escapes a black hole except some forms of non-visual radiation. Its presence can be seen by the effect it has on radiation around it and observing/detecting the other types of radition emitted from it. |
|
|
| Peter Daniel |
| quote: | Originally posted by uvodee
however, my worries start when I see people mount copper on aluminum or vice versa...
for some reason I do not trust the contact, there are always forces present (erosion ) and when eroding copper and eroding aluminum get together, I have a strange feeling about that. I hope I'm wrong but, having lived on a boat for a few years and having seen all kinds of metals erode (not corode) I try to keep different metals not in contact with others.
J-P |
They've been in contact (in my amp) for almost a year, and nothing happened. Copper is non treated, aluminum was alodined. |
|
|
| Damon Hill |
There's the galvanic reaction of dissimilar metals to consider,
but only if it's exposed to high humidity, I'd think. Some
types of heat sinks do combine aluminum and copper, pressed
together >really< firmly--the copper's usually a heat spreader
and the aluminum is the radiator.
Heck, I'm having a ton of trouble finding the heatsink I want,
especially at a reasonable price. |
|
|
| officeboy |
Computer heatsinks have been using a combo of copper contact plate/patch and aluminum fins for years now and I've yet to hear of a problem. Most are smashed together under a lot of force.
:smash: :smash: :smash:
Here is a link to my fav manufacturer for heatsinks that says a little about how they attach the 2 metals.
http://www.micforg.co.jp/en/alcu_embed.html |
|
|
| mcp |
| Material used is Aluminum Alloy is 6063. Some info can be found at: R-Theta |
|
|
| S.C |
| One of my opinion of the reason why aluminum was used so fequenly is that aluminum is cheaper than copper by almost half the cost. It is like why the pennies are made of almost 9x% of zine and the rest is the copper (0.001mm /0.01mm I forget which one) shell. I tried copper bar from onlinemetal.com, and it cost more than a double of the cost of aluminum. Since Aluminum have well performance in cooling and the lower cost, Industry tend to use it more. However, is aluminum a better cooling materials than copper? I'm not sure.;) |
|
|
| Sch3mat1c |
Considering you can use a lot more without breaking the bank, I would say YES :D Not to mention the ease with which it can be anodized black.
Tim |
|
|
| Magura |
The only benefits copper offers over aluminium is that it requires less space for transfering a given amount of heat, and that it can transfer more heat for the same cross section.
Magura:) |
|
|
| Magura |
The idea of using an electrically live bar is good, though to take full advantage of that approach the bar should be copper.
Magura:) |
|
|
| Peter Daniel |
| quote: | Originally posted by Magura
The only benefits copper offers over aluminium is that it requires less space for transfering a given amount of heat, and that it can transfer more heat for the same cross section.
Magura:) |
You forget about one other reason, copper is much heavier. In some cases heat transfer might not be priority for using copper. |
|
|
| marsupialx |
| Usually, when I use diamond heat sinks, I don't paint them black. |
|
|
| mandat |
| I do not use black paint when heatsinks are maid of black diamond, only. |
|
|
| Sch3mat1c |
Which produces a better sound, clear or black diamonds?
I bet industrial quality diamonds give it a harsh, abrasive sound.
Tim |
|
|
| Christer |
I haven't quite followed the thread, but doesn't the thermal
impedance play a role too? My CPU heasink has a cupper
kernel, which I understand to be for the reason that it can
"store" temporary heat peaks until the fins can dissipate it.
Yes, yes, it was experssed sloppily, but I am tired and I think
you understand what I mean. |
|
|
| CraigBuckingham |
looked at post 70 on the other link.
Insulators can have significant thermal resistance for high power devices if you use common types.
If a little research is done it is possible to find insulating washers with thermal impedances of 0.07 ºC/W for a TO3 type case.
Which leaves the biggest contributor by a huge margin to the thermal resitance chain, the junction to case thermal connection. |
|
|
|
