2
\$\begingroup\$

I have done heat sinking before, but now I'm trying to do it on SMT chips which is a bit more difficult. Anyway, I'm wondering whether the resistance between junction and ambient, Rja, is in parallel with the heatsink resistance. So that the actual resistance would be Rja||(Rjc+Rca)?

Also, I'm considering putting the chip on the bottom due to space constraints. However, I'm worried that the SMT heatsink, which seems to only be connected with solder will simply fall off or stress the board too much if it's connected on the bottom. Would that be a problem with such a heatsink.

\$\endgroup\$
2
\$\begingroup\$

AFAIK Rja is a 'summary' value, the result of the Rjc and the Rca, the latter in a particular situation, for SMDs often a recommended copper pattern. If you follow that recommendation, you can use the Rja figure.

If you don't follow the recommendation, you must calculate the Rca' that you have created, add it to Rjc, and you have the Rja' that applies in your situation.

In no circumstance should you calculate Rja||(Rjc+Rca), because you would in effect 'use' the heat path components twice.

There is of course some parallelism, in that the Rca' = Rca || R2ca, where Rca is the case-ambitient resisance path caused by the 'bare' housing and the prescribed copper pattern, and R2ca cause by any additional heatsinking you apply. But be careful with this approach, any heatsinking you add can block part of the path that the manufacturer calculated into the Rca mentioned in the datasheet.

In situations where the total Rth is a few C/W you must also take the Rch (case to heatsink) into approach, but this is typically in the order of 0.1 .. 1 C/W, so it is not very relevant when the total Rth is 10's of C/W.

Are you worried about the heatsink falling off during production or during use? AFAIK the solution for production is to use a glue. If you are worried about the heatsink falling off in normal use I think the temperature has become so high you have other more pressing worries.

|improve this answer|||||
\$\endgroup\$
  • \$\begingroup\$ I'm worried that in normal use conditions (not too hot) the solder is not strong enough to hold the heatsink upside down. Not that it will melt off. How strong is solder? \$\endgroup\$ – Matt Jan 13 '13 at 19:11
  • \$\begingroup\$ IME solder is pretty strong, at leasst up to 100C or so. \$\endgroup\$ – Wouter van Ooijen Jan 14 '13 at 7:34
0
\$\begingroup\$

As mentioned above, Rja is usually an inclusive measurement. Junction to ambient.

What's not clear from your post that hopefully your datasheet can answer, is how Rja is calculated. There are usually two parallel paths that make up Rja

Rjc (junction to case) + Rcs (case to sink) + Rsa (sink to ambient)

in parallel with

Rjb (junction to board) + Rba (board to ambient)

Don't neglect the board because it has the potential to dissipate a lot of heat. Typically it is given for a standard 4"x4" PCB.

Your chosen heatsink should be fine assuming you're doing hand soldering. Placing components on the bottom of the board doesn't work well for mass production wave soldering.

|improve this answer|||||
\$\endgroup\$
  • \$\begingroup\$ In most datasheets I saw they list Rja and Rjc, e.g. 35 (for D2Pak 1" copper) and 5, respectively. I thought that after adding a heatsink, the heatsink is in paralell to the 1" copper so that the effective R is 35||(2.5+Rheatsink). But from the answers I gather this is wrong. \$\endgroup\$ – Matt Jan 13 '13 at 19:44
  • \$\begingroup\$ I added some clarification about parallel Rth paths. \$\endgroup\$ – Wouter van Ooijen Jan 14 '13 at 7:33

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.