copper pins to increase PCB heat sink capability?

[coppercone2]

Do you think that getting copper pins (probobly hardened ones) and sticking them into vias around a transistor would help significantly with power dissipation? I am thinking about it, if there is room for convection around the PCB, maybe it would make sense to put a pin into every other hole and solder them in, so there is a draft going through the VIAS and getting heat off of the pins.

Have you ever seen anything like this?

[OM222O]

the vias will be way too small for any sort of "draft". the pins however can be usefull, but not as much as a proper heat sink and will set you back a lot more if you're not pressing them into a sheet metal which I assume you are not. EVGA has done this and called it pin fins but they are huge and still do very little cooling.
A commercial pin fin heatsink:

Evga pin fins:

[coromonadalix]

Normally any added mass / surface area  will help dissipate more, you can have small, big fins, long short ones  etc...  you have copper mesh in cpu heat sink tubings ...  even saw a small form pc with someking of very thick mesh as a heatsink on the case ??  but i did not saw it come into production ?

Your ides seems good   tests should be done ?  copper needles, copper nails   something like that ? 

I did saw on qfn32 packages somekind of a look a like copper screw thermally glued, it helped dissipate heat.

[SilverSolder]

The idea sounds pretty cool! ( :-) )

Definitely worth a test.

[filssavi]

The problem exotic assembly techniques such as that type of pin placement is that unless the volumes are actually huge (see EVGA) there is little to no chance to find someone willing to do it at a reasonable cost

[T3sl4co1l]

What were those solder-in, screw-mount copper slug heat spreader/sink things?  I can't remember the name of 'em...

You can get vias plated shut (mostly in small sizes), you can also fill them with solder (harder to do with a paste reflow process, mind), both help significantly (~2x or better) with thru-conductivity of the board.

I suppose a lot of large (>40 mil?) vias, closely spaced, would be a moderately effective heatsink, but I'd rather use the drilling time to make a better thermal connection to a more massive heatsink, I think.

Tim

[jonroger]

A copper filled via (eg, solid copper wire) is much better at moving heat from the top of a PCB to the bottom than an air filled via.  Copper "coins" are the same idea (but much larger).

[Psi]

I had a DPAK package with an extended ground tab track that ran in a solid strip underneath and alongside a big ATX style 24pos connector.
It was not thermally connected to the socket pins with copper, it just ran alongside the connector in a massive copper fill.

During thermal testing i was very surprised just how much heat was being sinked away by all the wires going into this connector.
I guess the heat was coupling through the FR4 and onto the connector pins where all that copper wire was just pulled it away and down the cable.

[coppercone2]

Hmm, so you don't think with any reasonable via size (I guess what, up to 1.5mm?) a mixture of smaller vias filled with long pins and air flow would be superior to just bulk copper mass?

What if there was a vent hold UNDER the chassis bottom and negative pressure that sucked in air under the PCB and it naturally wanted to flow through the vias. You could even put some liner around the edge of the PCB to try to direct air through the vias. Is the fluid resistance of those vias so high that it will not matter with reasonable fan options ? Of course it would clog unless you put filters.

The idea makes me think of the reactor on mars from Total Recall.

[coppercone2]

I think a reasonable test would be to make a LFPAK transistor in a 4 layer board populated by vias and run it as a heater. The datasheet tells you what the expected temperature rise should be with some amount of square inches of PCB heat sink with some via density.

Then solder wires in there aimed upwards and see if it changes. I found the datasheet for the LFPAK interesting because it basically tells you, if you increase the area over some limit you won't get any increased thermal performance.

the data sheet has alot more information then the picture I pulled related to surface area etc, but its a good summary

https://assets.nexperia.com/documents/application-note/AN10874.pdf



I also kind of wonder what would happen if you just shoved pins in there spaced closely and put it into a wave solder machine. maybe it would fill it up? so you can get ratty heat sink structures/walls? or even a shield can wall lol. It would likely result in bedlam but I thought also maybe you can bend U-shaped bits of copper wire and stuff them into vias to make a wave solder filled structure like the Sidney opera hall. I have a hunch that it won't increase thermal performance too much, but you can reduce size and space things closer together.

[OM222O]

Vias will have no airflow! Way too high of static pressure with even fans, let alone with passive negative pressure! They will be pointless and dumb. Filling them with copper or solder allows much better heat transfer to either a large plane or a real heat sink.

[coppercone2]

yea I gather that you won't get good flow through the holes, but it will still be interesting to see how much extra performance you get if rods stick out from the PCB rather then just a flush cut fill solder copper bar. Without any air flow holes that are counterproductive towards cooling effort.

How do you calculate the pressure drop on an orifice like that without doing alot of work in solidworks, anyway? everyone seemed to know the figure would be counter productive with no math.. how is it so obvious?

[OM222O]

as I mentioned, EVGA has done it. the effects are minimal at best! you need a ton of fins in order to get any meaningful difference and the airflow through the vias won't be just "not good flow" it will be zero! the static pressure behind fans is simply not high enough, unless you are using a delta fan and blasting air right through those vias which is a waste. this is just a matter of common sense as PC hardware has already proven these facts. even in radiators higher density can increase your temperatures despite the fact that it has more area. here the key is pressure, not area!

[coppercone2]

so pin fin heatsinks (even if its say a solid cast or machined part) is going to be inferior to a blade heat sink?

How do they compare to bladed heat sinks when passively cooling vs forced air? highly inferior in both cases?

A pin fin heatsink is appealing to make from a machined part because you don't need much more then a drill press, braze, and your parts. You can pretty much slap it together with no clamps or anything with minimal work. Like drill holes, dunk it in acid, put some flux and braze in there and heat the whole thing red hot. Then lap it. It's brainless compared to bending a bunch fins.


I mean a dense pin fin like your first picture.

https://www.iarjset.com/upload/2016/november-16/IARJSET%2026.pdf

[OM222O]

a dense array such as the one in the first picture can work well, but that requires a lot of manual work if not automated. and yes pin fins are inferior as opposed to the blade style in both cases. I agree that they are easier to make when you have something to press the fins into a sheet, but that comes with it's own disadvantages.

in water cooling equipment, they use skiving machines to create micro fins with crazy high density. that is much more complex to make but the pay off is much superior cooling compared to air cooling. the more complex a heat sink is, usually the better it performs as well (with exception of peltier cooling as it's useless for many reasons).

[T3sl4co1l]

I'm talking about what's manufacturable.

If you want to sit there skewering, soldering and cutting ten thousand wires a day to produce boards like that, be my guest!  I'll stick to conventional means.

Tim

[coppercone2]

thats a whole different story.

I think I have yet to make two of anything I built in my home. Mass production design is insanely neurotic shit I will not do without salary pay and benefits. 3/15000 PCB a year might go bad during soldering. Deploy ALL bosses   

PPM's used to describe how many screws are put in wrong...........

I try not to use this forum for corporate reasons

[coppercone2]

don't forget to write a TPS with those PCBs

[trophosphere]

What were those solder-in, screw-mount copper slug heat spreader/sink things?  I can't remember the name of 'em...

You can get vias plated shut (mostly in small sizes), you can also fill them with solder (harder to do with a paste reflow process, mind), both help significantly (~2x or better) with thru-conductivity of the board.

I suppose a lot of large (>40 mil?) vias, closely spaced, would be a moderately effective heatsink, but I'd rather use the drilling time to make a better thermal connection to a more massive heatsink, I think.

Tim

Are you referring to PowerPeg?

[coppercone2]

thats extremely interesting. The only part that concerns me is the soldering process, that bit might cool the part quickly and cause unusual strain.

Is it something you should be worried about? I thought about mounting those parts upside down before and using indium nanofoil to solder them to heat sinks without much heat. His method is incredibly convenient.

[David Hess]

If the vias are plated through, then filling them with solder will be easy enough.