1
\$\begingroup\$

I am designing a drive and need to electrically and thermally connect the tab of a series of FETs to an aluminum internal heat sink that is also used as an electrical bus.

It seems the best way to go would be to solder the FET tab to an intermediary copper layer and the copper layer to the aluminum.

Is there some sort of special flux and/or solder paste that would be needed to perform this? Methods to deal with oxidations? Would a reflow oven of some sort be required to ensure the copper/aluminum are soldered properly?

Are there processes out there to do this sort of thing?

I have struggled to find much information on this.

Thank you for your time.

Related question that came up from comments/answers: Does bolting to aluminum for electrical contact have any oxidation concerns?

Improve this question
\$\endgroup\$
4
  • \$\begingroup\$ Have you done the calculations if lapping and bolting it on isn't sufficient? \$\endgroup\$ – PlasmaHH Nov 8 '16 at 16:25
  • \$\begingroup\$ We may be able to get by bolting the tabs to the heatsink, but do have concerns of thermal/electrical conduction. I honestly haven't had much experience in this area myself and will have to investigate the lapping technique. Thank you for your feedback. \$\endgroup\$ – radix07 Nov 8 '16 at 16:31
  • \$\begingroup\$ Can you use lapping and bolting to ensure a good electrical connection? Would be a little leery of connecting a FET to a bus this way. Any information on this? Not seeing much for using it outside of thermal/physical connections... \$\endgroup\$ – radix07 Nov 8 '16 at 16:42
  • \$\begingroup\$ I have connected big fat kA carrying copper busbars just by bolting them together after a thorough cleaning, without lapping, and everything is still fine today. Contact resistance of proper connection is not nearly as high as many people fear \$\endgroup\$ – PlasmaHH Nov 8 '16 at 18:45
2
\$\begingroup\$

Do NOT solder your FET packages to aluminium. The temperature required to activate the flux is higher than the FET packages will be rated for.

It may be permissible to solder a copper intermediate to the alli, then reflow the FETs onto the copper. However, observe the 'time at temperature' figures for the FET packages, and know whether you can heat and cool the mass of the heatsink without subjecting the FETs to 'too long at too hot'. There are lower temperature solders available, which may allow you to dwell longer at soldering temperature.

Bolting is a perfectly sound way to connect electrically and thermally. Try it experimentally. It's so much simpler to do first time, and to rework, that you have to try it before dismissing it.

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ Reflowing shouldn't be any worse than what a board house does, in fact the Aluminum should cool faster than a PCB. But bolting does seem to be the simplest way to go. Will have to test out the various methods and see what works best for what we are doing. Thanks! \$\endgroup\$ – radix07 Nov 8 '16 at 17:28
  • \$\begingroup\$ Going to put this here as well... Does bolting to aluminum for electrical contact have any oxidation concerns? \$\endgroup\$ – radix07 Nov 8 '16 at 17:34
1
\$\begingroup\$

You can solder aluminum with the right flux. Be sure to get it all cleaned off afterward, it may not play well with electronics (I would bet on it). Copper would be easier.

Any kind of heat sink of substantial size is going to be rather difficult to solder to, it's in their nature.

Bolted connections, possibly with some thermal compound, may in fact be a better solution. It's what is used in most such situations from the 120W CPU in your PC to the manly hockey puck thyristors used in industrial power applications.

Improve this answer
\$\endgroup\$
9
  • \$\begingroup\$ I would be leery of electrically connecting straight to the Aluminum... Probably end up bonding some Cu to the Al separately somehow. Then reflow the heatsink and FETs together, separate from the board then attach to the board as needed. Not sure I can trust a thermal paste to be a main electrical contact, but wouldn't mind seeing more information on that. \$\endgroup\$ – radix07 Nov 8 '16 at 17:10
  • \$\begingroup\$ Thermal paste is a thermal contact, but if there is a bolt connecting the heatsink to the other part with a toothed washer biting into the aluminum it might be okay- I'd want to look into it a bit deeper though, especially if it has to pass salt spray or similar testing. \$\endgroup\$ – Spehro Pefhany Nov 8 '16 at 17:15
  • \$\begingroup\$ Will the Aluminum oxidize around/across a bolted connection? Never had to use Aluminum as a primary electrical contact.. \$\endgroup\$ – radix07 Nov 8 '16 at 17:25
  • \$\begingroup\$ @radix07 I'd want to look into it more. There may be ways to make it gas-tight and reliable. Copper clad aluminum is used sometimes. \$\endgroup\$ – Spehro Pefhany Nov 8 '16 at 17:32
  • 1
    \$\begingroup\$ Consider PEM fasteners. They crimp permanently into aluminum. I frequently specify this type. \$\endgroup\$ – Spehro Pefhany Nov 8 '16 at 19:40
0
\$\begingroup\$

I'm not sure it is "good practice" but you can solder to aluminum with the right flux. (This stuff works.) with all the solders I tried. So you could tin the Al heat sink and then solder on the FET's. I have never done this!

Improve this answer
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.