Aluminum vs Copper vs Brass

[MikeAusC]

We are just talking about the ability to dissipate a given amount of heat over a given amount of time right? So, wouldn't an ice cold aluminum heatsink (maintained at that temperature) potentially dissipate more heat than a copper heat sink at room temperature?

What sort of light will have it's heatsink maintained at 0 Deg C ?

 

[HarryN]

What sort of light will have it's heatsink maintained at 0 Deg C ?

A light in MN. :crackup:

 

[MikeAusC]

A light in MN. :crackup:

But only after midday in July.

 

[PMM]

Is it your intention to directly solder heatpad of SST90 to copper heatsink ?
or just thermal adhesive / paste

If soldering.. I would go Copper as long as you clamp well & paste the copper to aluminium bond tight the heat spread to the out Alu shell will be more desirable.

However if you using std paste/thermal adhesive I would go fully ALU because your introducing 2 layers of thermal loss through the paste/adhesive so using ALU you limit that hopefully to 1.

Alot of older (Pre-heatpipe) computer heatsinks were mix metals Nickle plated copper base press forged into an Alu Base/fins to gain best of all senerios.

My experiences with CPU cooling shows that thermal paste and applied pressure >50lbs play greatly in optimising heat transfer... something like egraf graphite thermal pads may be better but your want 70-90lbs clamping still.

If you going to use a paste / thermal adhesive "Arctic Alumina" is a better choice over Artic Silver it works better with lower temperatures which you may have if using in a Dive Light.... If temps are still likly to be high esp if overdriving the LED you maybe better off with something like Innovation cooling IC Diamond 7 or 24 but again decent clamp pressure needed.

Thermal paste is not for plugging large gaps but more micro scratches / pits you design so 2 surfaces can be as closely joined as possible with the maximum pressure possible then the TIM((thermal paste) is as thin as possible thus allowing maximum transmission of heat.

So the mating of all joining surfaces need to be as good as possible else your just you introducing barriers to the thermal flow... less barriers the better but if you can't get away from it tight joins/good clamping pressure & good quality thermal paste.

 

[MikeAusC]

. . .
However if you using std paste/thermal adhesive I would go fully ALU because your introducing 2 layers of thermal loss through the paste/adhesive so using ALU you limit that hopefully to 1.
. . . .

I don't understand.

No matter in which part you use it, Copper will ALWAYS deliver less temperature difference (so reduced temperature at LED) than Aluminium.

 

[archer6817j]

I should have said "ice cold" instead of ice cold :) I'm just taking theory. However, in a practical application (say a dive light) having a copper heat sink may not produce performance that is "significantly" better than an aluminum heat sink, given that it can conduct to the water.

 

[Slartibartfast]

I went with C101 for the heatsink for my SSR-90 build. Thin layer of AS5 between the star and the copper and between the heatsink and the body of the m@g. So far so good.

 

[unterhausen]

Copper will ALWAYS deliver less temperature difference (so reduced temperature at LED) than Aluminium.

I botched my previous comment, but this is what I was trying to get at. Let's say your aluminum is in ice water and your copper is buried in insulation. The led attached to the aluminum is going to be cooler than the led attached to the copper. If the copper is not allowed to conduct the heat away to somewhere else, then it's not going to out-perform an aluminum heat sink that is conducting the heat away. Since each interface between materials introduces some inefficiency, it may be the case that a solid aluminum heat sink will be better than a copper heat sink attached to an aluminum heat sink.

 

[PMM]

I don't understand.

No matter in which part you use it, Copper will ALWAYS deliver less temperature difference (so reduced temperature at LED) than Aluminium.

I mean pretty much what 'unterhausen' mentioned.

If your using a Copper slug and mounting to Alu then your using paste twice LED to copper slug & copper slug to Alu sink.

Because the metals are not one the paste is a thermal barrier to the flow of heat and if your not clamping above 50lbs with close mating surfaces you may as well just opt for the one metal Alu.

Copper is good at absorbing and conducting heat but not so good at radiating that heat away into another body.

End of the day you have to still move the heat out of the copper into the Alu if that is not efficient no point having the copper in place it will either equalise out with the Alu temp within reason remember heat only flows from a warmer to colder surface or if the joins between mating surface are naff you may see elevated die temperatures and a need to run at a lower current to avoid thermal stress on the LED.

In a dive light certainly the best design option is getting the water as close to the LED cooling plate as possible and flowing to extract the heat away as close to source as possible.

 

[archer6817j]

In terms of bonding the copper to aluminum, has anyone tried shrink fitting a copper slug into an aluminum shell without thermal paste? So, a copper cylinder with say a .001 interference fit...heat the aluminum sink to expand it and put the copper slug on ice to shrink it...then mate the two together. Assuming the surface finish on both parts is very smooth, this might be the best way to get a lot of contact area instead of say gluing the bottom of the copper to an aluminum sink with thermal epoxy or screwing together with thermal paste.

 

[Mick]

http://www.customthermoelectric.com/TIMs.html

 

[tylernt]

In terms of bonding the copper to aluminum, has anyone tried shrink fitting a copper slug into an aluminum shell without thermal paste? So, a copper cylinder with say a .001 interference fit...

Oh yes the friction force for interference fit is quite high. I imagine that'd make a lovely thermal connection with no paste etc. Case in point, I've seen more than one computer CPU cooler with a copper slug pressed into the middle of aluminum fins.

 

[350xfire]

That is what I usually do. Build the heatsink to fit a mag body pressed in. Then use my hydraulic shop press to press it in!

 

[precisionworks]

has anyone tried shrink fitting a copper slug into an aluminum shell without thermal paste?

Even a tight press fit (Class V) doesn't assure full contact between surfaces. The problem is that most surfaces are far from smooth, with a typical machined finish measuring about 120 Ra. A very fine machined finish will go about 60, rough grinding will yield 30, fine grinding at 15, super fine grinding 8, near mirror at 4, full mirror at 2.

Press together a pair of parts, both having a 2 Ra finish, and there is nearly 100% contact. Press together a pair of "average" machined parts having 120 Ra & 25%-45% of both surfaces might be in contact.

The photo above clearly shows this. The aluminum sleeve (that looks like it has fine grooves) was finished as nicely as possible on my lathe. Slow feed, enough DOC to engage the nose radius, fast rpm, lots of lube, etc. Looked like a 60 to me. Pressing it into the quill (that had just been bored) shows how rough the surface really is. I would guess the contact between those parts, which was a tight fit requiring a 5# sledge for assembly, might be 30% to 40%.

 

[Mick]

The thermal compound will also prevent any galvanic corrosion between the dissimilar metals. Press fit if you like but also apply Arctic Silver 5.

 

[tylernt]

The thermal compound will also prevent any galvanic corrosion between the dissimilar metals.

I thought galvanic corrosion only happened in the presence of water or other electrolyte? There might be a little water vapor between the pressed parts, but not much...

 

[MikeAusC]

If your using a Copper slug and mounting to Alu then your using paste twice LED to copper slug & copper slug to Alu sink..

There is a very small area behind the LED where all the heat has to pass through - you might have a 5 degree drop with copper, but an 8 degree drop with aluminium.

Hopefully the interface from copper to the alumium slug will be large and this may only add 1 degree drop.

 

[MikeAusC]

. . . . Copper is good at absorbing and conducting heat but not so good at radiating that heat away into another body . . . . .

But it's easy to surface-treat or coat any metal to make it as good as the best.

There's thread here somewhere that documented tests on this.

 

[CKOD]

Copper is good at absorbing and conducting heat but not so good at radiating that heat away into another body.

A very small portion of the heat dissipated for our purposes is removed via radiation. Most heat would be removed by conduction and convection, and the difference between aluminum and copper should be minimal also in regards to cooling from radiation.

There are a lot of theories and stuff about aluminum and copper mixed, and why they use it in commercial coolers.

Aluminum is lighter, cheaper, and stronger in some alloys/forms hence why it gets used. In some situations, the differerence could be only academic. (Where thermal conductivity of the metal isnt an issue, and surface of the heatsink is the determining factor of the steady state temperature)

When pure thermal performance isnt the sole design factor, and cost, weight, strength, ease of manufacturing, durability, start coming into play, thats where aluminum takes the lead. You see copper slugs set into aluminum heatsinks because at the source of the heat, where there is a high temperature gradiant, and high "thermal density" (unsure of the actual term... Watts per mm^2) the advantage of the copper is significant. Once the highly conductive copper spreads it out, and you just need bulk surface area to get rid of the head, aluminum is the most cost effective option.

Now, lets see some of the CPF manufacturers start friction welding some copper heatsinks into an aluminum body for the best thermal transfer across the parts ;)

 

[PMM]

There is a very small area behind the LED where all the heat has to pass through - you might have a 5 degree drop with copper, but an 8 degree drop with aluminium.

Hopefully the interface from copper to the alumium slug will be large and this may only add 1 degree drop.

Would you be surprised if I said you could experience 3deg+ alone at surface joins with differing brands of paste even at clamp pressures as high as 70psi which I can tell you that you won't be having with an SST90 led ;) on the bare emitter because the bare emitter will snap been there done that :whistle: so you could experience 6-10 deg additions per surface join area.

Have to admit never tested a sink without paste... good link above to the tims but I am going to be nit picky ;) in reality 2 surfaces will only touch at the 3 highest peaks so contact point is actually alot less there is a great IBM document on surface contact that is a really good read.

I have a testdie simulator at home for computer heatsinks capable of 0.001 deg resolution though I admit 0.01 is more than enough and have done some work recently comparing thermal pastes as test data for supplier of thermal paste... bit of an eye opener.

The delta-t of a computer die can be as high as 15-20 deg over ambient with low clamp pressures and cheap generic paste compared to a top brand paste clamped at say 70psi giving 8deg difference at a 45watt heatload.

Compound failure in notebooks (drying out/pumpout) of TIM has seen delta-t differences as great as 30deg when a quality TIM is reapplied.

That's food for thought.. a Power LED really is not that different from a CPU.