0
\$\begingroup\$

I found that my heat sinks are not large enough for the surface it is being used on, so I am wondering how to connect multiple to each other. My goal is to connect 4 side by side to each other, and to possibly stack one on top of the other as they are very thin. The question is, what material can I use to do this while still maintaining full conductivity? If I stack two on top of each other, but just use glue to connect them, I fear that the stacked one will not be conducting the full amount heat capable as the glue will interfere. What can I use to properly do this?

\$\endgroup\$
  • \$\begingroup\$ I don't suppose buying a bigger heat sink is an option? That would be the better way. \$\endgroup\$ – Tom Carpenter Sep 12 '16 at 2:18
  • \$\begingroup\$ I would prefer making use of the twenty smaller ones I have rather than buying one bigger one and letting the others go to waste - if this is possible, of course. \$\endgroup\$ – MH0517 Sep 12 '16 at 2:23
  • \$\begingroup\$ A drawing or photo of your proposed setup would be useful. Use mechanical fasteners (screws and nut or spring clips) to fasten the heatsinks together, and use thermal compound (NOT glue!) in every joint. \$\endgroup\$ – Peter Bennett Sep 12 '16 at 2:24
  • \$\begingroup\$ every interface has an optimal Rth value that is in series. The end result depends how well you adopt std. practise and total Rth * PD (watts) for a theoretical junction temp rise. Since devices are in parallel, sharing the same sink, ambient will be increased and results will be higher than individual results. One way is assume all power is from one device. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Sep 12 '16 at 2:43
2
\$\begingroup\$

You could use glue - but only as a bonding agent, not as a connecting agent.

By this I need to draw ASCII diagrams - forgive the crudity!

Surfaces to be bonded:

=== ===
  | |
  | |
=== ===

Use glue ($)? I don't think so!

===$===
  |$|
  |$|
===$===

Use conductive paste (#)! But what holds it together?

===#===
  |#|
  |#|
===#===

Use both:

$$$$$$$
===#===
  |#|
  |#|
===#===
$$$$$$$
\$\endgroup\$
  • \$\begingroup\$ Not sure glue would stick anything in that situation. Tightly assembling both heatsinks with bolts/screws would certainly give better results (both mechanically, and from a thermal conductivity point of view) \$\endgroup\$ – dim Sep 12 '16 at 13:06
1
\$\begingroup\$

If you MUST do this then you should join them with material which has low thermal resistance in the joining section compared to the heatsink itself.

For instance if a heatsink was made of 1.5mm sheet and had dimesnions of around 50mm x 50mm x 20mm, then use of a plate of 3mm or thicker aluminum with as short a length as possible would be desirable.

Most small heatsinks will be made of aluminum. Soldering to aluminum is possible but difficult enough to be inadvisable. Connections can be made using at least 2 screws per bridge to heatsink connection - to ensure mechanical stability. Sink to brdige area should be as large as a heastsink fin and adjacent faces should be abraded clean and then have thermally conductive jointing compound applied and then be firmly fastened together so coated surfaces are in good contact. .

Using a thermometric measuring system - finger, tongue, thermometer, non-contact IR (best of all) ensure that

  • the two heatsinks are at ABOUT the same temperature.

  • There is no significant temperature change across joints or along joining strips.

_______________

Soldering sounds like a good idea, and is something I actually own as oppose to thermal paste. My heatsinks are copper though, will soldering copper work?

Soldering copper is doable with "ordinary" solder provided that you use a powerful enough iron.
BUT heatsinks soak up thermal energy very well - its their job - and you may need a VERY powerful iron.

Thermal compound should be available at modest cost and would be MUCH easier to use.

________________________________________

Thermal compound of a large range of types and sizes here US supplier

You only need a "smear". If you place some on one surface, press the surfaces to be joined thermally together and then slide them slightly across each other, if the second surface takes up a coating of compound across the whole area where it faces the coated surface then you have enougyh. If you have bare spots you may need some more, bearing in mind that if you are bolting or screwing the 2 parts together (as you should be) that they will be pulled together somewhat under screw tension.

"Arctic Silver" brand is generally reputable (if genuine) but also usually not the cheapest. Note that the lowest cost at $1.99 is also in a VERY small quantity and costs more per volume that higher cost per container AS products.

Almost anything that is sold as thermal greasem as long s it's not a scam, will do well enough in most cases. It is possible to get face to face losses with thermal compound that are usuallu much smaller than heatsink to air losses. The exceptions may be with very large specialised heatsinks, lots of heat, fan cooling and proper heatsink design. In most cases as long as it is "real" thermal grease and fills the inter-surface void when clamped together then it will work well enough.

Examples:

AS for 5.99/1.75 gram ~= $3.43/ gram
AS $26 / 12g ~+ $2.17 / gram
Fanner Tech $1.99 / 0.5g ~= $4/gram

Ebay legion

$4 / 20 g !!! free shipping - Hong Kong.
Looks like Yak cream and may be BUT probably OK :-).

30g/$1.18 free shipping ...

Just when you thought ... 30g/$0.99 free shipping :-)

Arctic Silver Yeah, Right $1.43/30g free shipping. If this was the genuine branded product you'd be in business. BUT even if not (as is VERY likely) odds are it may work OK.

____________________________________

Does it work?:

Equipment: Two small metal plates, goo to be tested, soldering iron.

WITGHOUT goo support ONE plate on a fingertip.
CAREFULLY bring hot soldering iron near top side of plate?
Thermal path is now iron - air - plate - fingertip. How hot does it get, how quickly and from how far away?. ie get an idea of how well metal plate conducts soldering iron heat.

Now:

Take two small metal plates.
Apply enough of goo to be tried that when pressed together there is NO metal to metal contact - ie thermal path MUST be via thermal goo.
Support assembly on fingertip.
CAREFULLY bring hot soldering iron near upper plate?
Thermal path is now iron-air-upper plate - thermal goo - lower plate-fingertip. How hot does it get? Result should be VERY similar to system with 1 plate and no goo.

Have you tried this?
ME???!
Are you mad? No. Hmmm. Must do so.

\$\endgroup\$
  • \$\begingroup\$ Soldering sounds like a good idea, and is something I actually own as oppose to thermal paste. My heatsinks are copper though, will soldering copper work? \$\endgroup\$ – MH0517 Sep 18 '16 at 18:44
  • \$\begingroup\$ Soldering copper is doable with a powerful enough iron BUT heatsinks soke up thermal energy well - its their job - and you may need a VERY powerful iron. Thermal compound should be available at modest cost and would be MUCH easier to use. | Where are you located? \$\endgroup\$ – Russell McMahon Sep 18 '16 at 22:49
  • \$\begingroup\$ United States, Georgia specifically, so I shouldn't have any trouble getting it, I just live in a small town. \$\endgroup\$ – MH0517 Sep 20 '16 at 21:48
  • \$\begingroup\$ @Deadfult See additions to answer at bottom \$\endgroup\$ – Russell McMahon Sep 21 '16 at 10:46

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.