Grafting Heatsinks together

G'day Guys,

If I have several identical heatsinks, could I stack them on top of each other and graft them together with a spreader plate of say 5mm aluminium?
With the purpose of creating a larger heatsink of lower C/W.

20240507_213405.jpg
 

Attachments

  • 20240507_213412.jpg
    20240507_213412.jpg
    451.5 KB · Views: 59
Something like this would be the plan:
20240508_171308.jpg 20240508_171327.jpg

They are 30cm long so not practical to increase in length.

The next question is what is the best way to affix to and transfer the heat from the spreader plate to the heatsinks?
Bolts and thermal goo?
Heat conductive glue?
 
Member
Joined 2019
Paid Member
I'd think bolts and paste. Structurally more secure and the thermal conductivity of the paste would be better from what I've read paste can be up to 2x the conductivity of glue. The more bolts the merrier I'd think because they'll also get heat directly into the heatsink and ensure contact between plate and sink is as good and even. Might be a good idea to use aluminium bolts too so they expand and contract at more or less the same rate as the material they're holding together. Or were you planning on drilling through the sink and using locking washers?

Also worth considering the different coefficients of expansion of differing alu alloys?

Will the output transistors be arranged 50/50 on each side of the board? So each sink could get the same heat input? If so, would it be worth just bracketing the sinks together to keep them mechanically stable? Half the heat goes to one sink, half to the other?
 
If you are using a plate to mate them, try to get 5mm copper, not alminum. It will spread the heat much better than aluminum. You need not cover the whole surface with it, depending on the semi's you fix you can leave some space on top and bottom.
Be sure to leave a tiny gap like .3 mm between the heatsinks, otherwise the construction may bend when heated.
Use good, clean (!) thermal paste in a thin layer, not too much. You should use large washers under the screws pressing the parts together and not over tighten them.
Better to let them set over night and then repeat next day as the paste will still move.
 
Also worth considering the different coefficients of expansion of differing alu alloys?
No really, its only a few ppm/C, more important for very different materials like steel/aluminium, though even copper and aluminium are pretty well matched for this kind of use. If only diamond was a practical material for heatsinks, almost no thermal expansion and extremely good thermal conductivity...
 
Short answer is yes, but there will be some thermal resistance between the interfaces (where metal parts contact) so surfaces need to be as flat as possible with conducting paste - check out metal based pastes for computer CPUs.

You will end up with slightly more heatsinks than the equivalent single piece heatsink.
 
Short answer is yes, but there will be some thermal resistance between the interfaces (where metal parts contact) so surfaces need to be as flat as possible with conducting paste - check out metal based pastes for computer CPUs.

You will end up with slightly more heatsinks than the equivalent single piece heatsink
That's roughly what I thought.

Not as efficient as a proper large heatsink but I happen to have a bunch of these onkyo heatsinks and proper heatsinks are not available in NZ. They require bringing in for overseas.

My evil plan is to graft 6 of them together to make 2 monster heatsinks as per the photos. I'd expect them to weigh at least 3.5kg as the individual heatsinks are 0.9kg. This should be more than enough for my wolverine monoblocks build when I eventually get there.