Copper and Aluminum

[CalCoolage]

>How heat transfer to air via convection is not material dependant (except for the
>properties of the air itself...
OK, lets suppose that we have a given volume of flow for both water and air. Is it not the thermal properties of water versus air which determines that water will cool better than air? ( rather than something like viscosity.) Why then is it irrelevant to the heat transfer what the thermal properties of the sink are? Sure the heat is carried away by convection, but the transfer of heat from the sink to the fluid is by conduction. A unit volume of water takes up more heat than air, and that seems to be the relavent property. If aluminum gives up more heat per volume unit then copper, that would seem to be the relavent property.


I did find a design sheet for heat sinks that had some formulas. Unfortunately, the design procedure works in a logical circle and involves using a "nomograph" look up, so I couldn't guess much about what the thermal resistance does to the effectiveness of a heat sink. Except for emissivity, no other property of the material enters into the figuring. Emissivity enters into heat radiation part, but not the heat conduction part. Whether the properties of copper are slipped in via the "fin effectiveness factor" I couldn't guess. But the temperature difference enters into the thermal resistance calculation as the fourth root. No wonder copper sinks perform so closely to aluminum sinks, even though copper is twice the conductor aluminum is. Still, the forth root of 2 is 1.19 so one might expect copper sinks to be 20% better than identical aluminum sinks, which does not seem to be the case.

BTW, the myth is alive and well. At

http://www.amdmb.com/article-display.php?ArticleID=101&PageID=4

They review a silver plated copper sink, and say the silver plating "gives it a higher heat dispersal rate than Copper ."

They also have this copper-aluminum babe. I don't think they were trying to save money or weight on this one:

The Swiftech MC-462A Rev1... the heatsink itself is comprised over a copper bottom and aluminum pins. Weighing in at 760 grams its attachment method alleviates any concerns you might have because of the weight.

 

[*spazzed*]

I'm working on a heatsink that has a copper core with aluminum fins, all held together by silver solder (the jewlers stuff).........copper in the middle, because it absorbs heat better from a heat source, and aluminum fins, since aluminum dissipates heat into the air better than copper. The silver solder is to hold everything together with as little loss of heat transfer as possible
Hope it works

 

[Hoot]

Spaz, I didn't know you could silver solder aluminum, but whatever the process to get there, it sounds like an excellent idea. Can't wait to see it and hear how it performs.

Hoot

 

[Badger]

*spazzed* (Jul 19, 2001 02:47 p.m.):
I'm working on a heatsink that has a copper core with aluminum fins, all held together by silver solder (the jewlers stuff).........copper in the middle, because it absorbs heat better from a heat source, and aluminum fins, since aluminum dissipates heat into the air better than copper. The silver solder is to hold everything together with as little loss of heat transfer as possible
Hope it works

You cant silver solder aluminium, plus if you read this thread you will see the message coming accross that the idea that aluminium conducts heat to air better than copper is a fallacy.
Good luck anyway, but dont wast your time trying to silver solder alu and copper together!

 

[cjtune]

CalCoolage (Jul 19, 2001 07:26 a.m.):
>How heat transfer to air via convection is not material dependant (except for the
>properties of the air itself...
OK, lets suppose that we have a given volume of flow for both water and air. Is it not the thermal properties of water versus air which determines that water will cool better than air? ( rather than something like viscosity.) Why then is it irrelevant to the heat transfer what the thermal properties of the sink are?....

How heat transfers through fluid depends on the property of the fluid -yes, by conduction and advection, the movement of the fluid itself. Only the surface temperatures of the heatsink is needed -the surface of any material.

How heat transfer through a heatsink depends on the material of the heatsink, among others. Only the temperature of the interfaces in needed (did I say it was watercooled or TEC cooled? It doesn't matter, as long as a temperature gradient exists)

But when one says air removes heat BETTER from aluminium, we are looking at the heat transfer process from the view point of the working fluid (be it air, water, or flourinert). But if we say, aluminium transfers heat (conducts) better than copper, we are looking at the latter condition, which requires values of thermal conductivity.

As for specific heat capacity, yes, it is of NO CONCERN because as time progresses (esp. on the hours/days a PC might be on) as once it is saturated, they'll ALL reach a constant temperature. In other words, the heat capacity only delays heat from being transferred form the heatsink to the air (but of course the CPU is being cooled as the waste heat is still being removed form it and is being stored in the heatsink) -for the first few minutes after powering up. In a water cooler setup, if your radiator or bong is inept at throwing heat to the ambient, however large the heat capacity of the water in your rig after adding additives or whatever will not help your CPU except for the first hour or so.

I did find a design sheet for heat sinks that had some formulas...

This is sheet for the design of the geometry of a heatsink to determine its effectiveness in throwing away heat via the combined effects of convection and radiation (how's this, dozier768? Looks like the engineers gave a damn about it). Look for a variable 'k' anywhere in the equations. If you find it, try and substitute different values and see what you come up with. But then it may have been taken as one lump sum when determining the (expermental) correlations between fin performance/effectiveness/thermal resistance with respect to length, thickness, and the convection (ie. airflow) heat transfer coefficient -which also has to be determined expermentally. BTW, the fourth root (ie. deltaT^0.25) is from natural convection, that is, the movement of air without the assistance of fans -it seems either that the engineers where very thorough in including this (which is usually neglegable with forced convection), or this heatsink was meant for passive cooling (no fans). Together, the heat transferred will vary to deltaT to the power of 1.25 (deltaT^0.25*deltaT) for natural convection. Even then, it is NOT always to the power of 0.25 -this depends on the temperature range and fluid type. Between lower temperature ranges, the (additional) deltaT may be to the power of 0.2 or 0.18.

BTW, the myth is alive and well. At
http://www.amdmb.com/article-display.php?ArticleID=101&PageID=4 ....

Did they test the SAME make of heatsink but without the silver coating? Nope, didn't see any 'plain' Akasa there... Anyhow, read my prevous posts above again.

They also have this copper-aluminum babe. I don't think they were trying to save money or weight on this one:

The Swiftech MC-462A Rev1... the heatsink itself is comprised over a copper bottom and aluminum pins....

I think we have discussed copper-based HSFs previously too...

 

[cjtune]

Ergh, just found out there's a 4098 char limit on posts...

 

[cjtune]

*spazzed* (Jul 19, 2001 02:47 p.m.):
I'm working on a heatsink that has a copper core with aluminum fins, all held together by silver solder.....

Hybrid copper-aluminium heatsinks are implemented as a 'cost-effective' measure when the design engineers are at a lost at designing better HSF geometry (very complex, you know). If $$$ permits, you might as well go for an all-copper HSF with the same build. But try and build one anyway -we'd like to see the results and take care to solder the Al-Cu interface properly. BTW, I don't think aluminium can be soldered. AS2 epoxy as alternative, maybe? Mating surfaces obviouslty must be super flat and reasonably smooth.

 

[*spazzed*]

Well the solder did stick........ for a while. When heat was applied again, the bond broke and everything fell apart :-( Gonna try AS epoxy, if I can get my dad to let me borrow his credit card. :'(

 

[cjtune]

*spazzed* (Jul 20, 2001 01:09 p.m.):
Well the solder did stick........ for a while. When heat was applied again, the bond broke and everything fell apart :-( Gonna try AS epoxy, if I can get my dad to let me borrow his credit card. :'(

Nope, looks like aluminium really cannot be soldered upon.

I'm afraid to ask. How many Durons or Tbirds do you have for it to be tested on once you're finished?
Reminder: Apply only a thiiiiiin layer of AS2 and SQUASH to halves real hard till it cures (using a clamp?).

 

[*spazzed*]

cjtune (Jul 21, 2001 11:01 a.m.):

*spazzed* (Jul 20, 2001 01:09 p.m.):
Well the solder did stick........ for a while. When heat was applied again, the bond broke and everything fell apart :-( Gonna try AS epoxy, if I can get my dad to let me borrow his credit card. :'(

Nope, looks like aluminium really cannot be soldered upon.

I'm afraid to ask. How many Durons or Tbirds do you have for it to be tested on once you're finished?
Reminder: Apply only a thiiiiiin layer of AS2 and SQUASH to halves real hard till it cures (using a clamp?).

No durons or t-birds, I had a Plll, but then sold it to get a t-bird. Don;'t have enough cash right now. I have a celeron 533@825(that doesn;t make that much heat). I might try a different type of heat source, to simulate something of a higher heat range. Or just use that crappy celeron