Copper/Aluminum heatsink - yay or nay?

thylantyr

Member
2001-02-19 10:38 pm
Mars
I can't remember where I seen this concept, was it home computer CPU heatsinks or was it Rockford Fosgate, whatever


Traditional transistor on heatsink

------------------------
aluminum fins
------------------------
aluminum plate 1/4" thick
------------------------
transistor
------------------------


Copper method
------------------------
aluminum fins
------------------------
aluminum plate 1/4" thick
------------------------
copper plate 1/8" thick
------------------------
transistor
------------------------


What is the pro's/con's of using copper sandwich?
 

Nisbeth

Member
2001-01-29 9:58 pm
Denmark
I haven´t got much theoretical knowledge to back this up but I´ll give it a shot anyway:

Copper is better at transporting the heat away from the silicon chip, whereas aluminium is better at radiating it to the surrounding air (besides being cheaper). This means that using a copper plate lowers the thermal resistance between the transistor case and the actual heatsink (the fins) and also ensures that the heat sink will be used as effectively as possible because the copper is better at transporting the heat to the "far corners" of the aluminium heatsink.

My 0,02 etc.

/U.
 
Ok here's the deal. You have two principal characteristics to look for when you're choosing heatsink material; radiation (given by the colour of the sink's surface) and thermal conductivity (given by the metal).
The thing with cooper is that's an EXCELLENT electrical & heat conductor, which means that heat is quickly removed from the power devices, but it's red/orange :) which means it radiates poorly. Most sinks are built in Al, which, while it's not such a good heat conductor as Cu (yet still good enough), it's cheaper and can be readly anodized to any colour you like.

Now, i'm building my sinks using Al extrusions for the surface area and 2mm thick Cu sheet to bolt them to the Al, hoping to get the best of both. Hopefully it'll work!
 

roddyama

Ex-Moderator
2002-01-19 9:25 am
Michigan
Lisandro,

Now, i'm building my sinks using Al extrusions for the surface area and 2mm thick Cu sheet to bolt them to the Al, hoping to get the best of both.
That sounds like a great idea. The copper will rapidly pull the heat away from the devices and than transfer it to the aluminium heat sink over a much larger area. This will make the whole heat sink more efficient.
There must be a downside or manufacturer would have used this method to save cost with smaller heat sinks (less weight). In any case it should make for cooler running FET's or BJT's with a given heat sink.

It would be interesting to hear from anyone who tried this before.
Rodd Yamas***a
 
Lisandro_P said:
Ok here's the deal. You have two principal characteristics to look for when you're choosing heatsink material; radiation (given by the colour of the sink's surface) and thermal conductivity (given by the metal).
The thing with cooper is that's an EXCELLENT electrical & heat conductor, which means that heat is quickly removed from the power devices, but it's red/orange :) which means it radiates poorly. Most sinks are built in Al, which, while it's not such a good heat conductor as Cu (yet still good enough), it's cheaper and can be readly anodized to any colour you like.

Er...kind of. From what I understand, Cu conducts heat better than Al but transmits it less well (air/metal interface transmission). Emissivity (=radiation) also enters into things but I believe this is a small contributer when compared to convection.

When you go to a dual Cu/Al design you have to balance the interface losses. This really is rocket science!
 
roddyama said:
There must be a downside or manufacturer would have used this method to save cost with smaller heat sinks (less weight). In any case it should make for cooler running FET's or BJT's with a given heat sink.

Well, for starters you're introducing an interfase between the Al & Cu, which must make excellent thermal conection; otherwise it's a no-go situation. So everything must be well soldered (could it be done?) or bolted & greased to get it to work. In any case, it wouldn't be cheap anymore... But again, you see the weirdest things, so i dunno... i once ran across *SILVER* heat sinks for PC hardware (video cards). Silver is even better than Cu at conducting heat, but i'm not willing to pay it! :D
On the other hand, it's no unusal to find large sinks using a Cu sheet to help distribute heat quicker, specially in very big ones.

My goal was to get good heatsinking at a moderate price. I already bought and cutted the Al extrusions, and will get the Cu sheet this week. It's a lot of metalwork (a LOT), but i'm not spending over US$50.
 
tiroth said:
Er...kind of. From what I understand, Cu conducts heat better than Al but transmits it less well (air/metal interface transmission). Emissivity (=radiation) also enters into things but I believe this is a small contributer when compared to convection.
When you go to a dual Cu/Al design you have to balance the interface losses. This really is rocket science!

I was convinced that the heat transmision in contact with air was given solely by surface area and not sink material... or am i mistaken? Radiation transmision can build up to 15-20% of the total i readed once, and it does make a notable difference.

And about the rocket science thing... yeah, you're right :) I once tried to get into the physics of this thing and ended up settling with ESP's sink calculator. Didn't have the courage :D
 
I don't think you can extrude copper the way you can aluminium, at least not at reasonable pressures anyway, so aluminium wins on cheapness there. Don't stop at a 2mm sheet of copper as a thermal "spreader", 5mm would be a lot better. For all the devices that have their collectors / drains electrically connected, you could mount them *directly* to the copper and get much improved thermal contact, and put the insulating layer (that now has a much greater cross section and therefore a correspondingly lower thermal resistance) between the copper and the aluminium. This is what I have done with my CDA amplifier, see the pic at thread -> http://diyaudio.com/forums/showthread.php?s=&threadid=3891&perpage=15&pagenumber=2

GP.
 
I'm using 2mm because its the only thing avaliable here at a reasonable price. If i could get thicker sheets i'd use them, but even with 2mm it should have enough thermal inertia.

I like the idea of insulating the copper directly a lot, through i have to see if this is feasible with my h.s. arrangement. Where did you get the insulator sheet?
 
Heatsink etc

My 2 eurocents worth:

- I'm not convinced that you can interface the copper sheet to the heatsink plate. Both should be *very* flat to avoid trapping air or otherwise bad contact which would mean a lot of thermal resistance. If you use screws, use a lot!

- The radiation from the sink material to the ambient air solely depends on the color of the fins, not the material. There should be no difference in this respect between black Al or black Cu. The term used in this respect is *color temperature*.

Jan Didden
 
Lisandro_P, the insulator sheet was just a bit of an unused sample we had at work. It is made by Bergquist I think.

Jan, if you look closely at the pic you can see I used thermal compound on both sides of the insulator, as well as between the fet and the copper. The insulator is normally meant to be used dry, but that wasn't good enough as you rightly point out.

GP.
 

roddyama

Ex-Moderator
2002-01-19 9:25 am
Michigan
Yeti
If you apply the same thermal resistance either with cu or al the copper always will have the heigher wheigt and costs.
This may be true, but it misses the point. If you install the device as Graham did in his link, you get the heat away from the device faster. There are two reason for this, 1.)cu has twice the thermal conductivity of al, and 2.)the device is in direct contact with the cu plate so the thermal resistance is much lower. Once the heat is taken away from the device it is spread (transfered) to the al over a much broader area.
The main result will be that you can run the amp at a lower device temperature or use less heat sink for the same power output.

Rodd Yamas***a
 
heatsink

Well, you're cutting a lot of corners here. The thermal resistance from device to Cu is the same as from device to Al if the same mounting method is used. The heat will indeed spread faster, but there is the additional thermal resistance from Cu to Al. You really can't say that the result is better or worse, unless you do some calculations.

Jan Didden
 

roddyama

Ex-Moderator
2002-01-19 9:25 am
Michigan
Hi Jan,

Yes, the calculations should be done to verify the final results, and yes, you sould be cautious of alternative methods.

I'm only trying to make one point, that is if the device is direct mounted (without an insulator) to the copper plate, the copper will pull the heat away from the device faster than the al heat sink and an insulator. The insulator can than be placed between the cu plate and the heat sink where there's more area to transfer the heat. Each device would have its own cu plate to keep from shorting devices.

The main idea is that the heat is being remove from the device more efficiently.

Rodd Yamas***a
 
I know aircraft people are always watching out for dissimilar metal interfaces due to eventual structual weakening due to corrosion. Over time corrosion will occur between some dissimilar metals fastened to each other. Hopefully none of us has to worry about structural failure of our amps ;) but that corrosion may cause more thermal resistance over time. Not sure if Al and Cu are on that list, but something to perhaps consider.
 

leroy

Member
2001-10-31 10:52 pm
I'm only trying to make one point, that is if the device is direct mounted (without an insulator) to the copper plate, the copper will pull the heat away from the device faster than the al heat sink and an insulator. The insulator can than be placed between the cu plate and the heat sink where there's more area to transfer the heat. Each device would have its own cu plate to keep from shorting devices.

Isn't that already what the manufacturers are doing? Almost all devices have a (yes) copper backplate where the silicon is attached to. Connecting the copper backplate of the device to another copper (back)plate is only adding thermal resistance in my opinion....
 
Aluminum has 60% of the conductivity of Copper. That is a rather big difference. But it has a higher density. Aluminum weighs 48% of the copper at the same conductivity. In other words, the copper's conductivity benefits are usually overcome by the higher price and weight. This is why heatsinks are generally aluminum.

For DIY heatsinks, however, one does not care about weight, only performance. I would be afraid of the copper/aluminum joint that is required. It would have to be very tight to work properly. If I had the means, I would opt for a solid copper heatsink, on aesthetic reasons alone. Well, I hope that your project turns into a reality. Let us know how it turns out.
-Dan