7
\$\begingroup\$

I'm managing several designs. At one point I identified a few failures that were caused by 0603 caps failing short. After that I switched to all soft-term caps to hopefully eliminate the problem.

Well it's been several years and the problem has not re-occurred that I know of. So I guess it did work, but keeping these soft term caps adds a lot of cost to every board.

So here's what I'm wondering. To those of you who've produced a lot of boards: is this normal? I'm making products that need to last, but they're not safety critical or anything like that where extreme care is required. They're also not used in a high-shock environment (besides being shipped).

If it's really uncommon to have to do this, I'm considering reviewing the designs for board-flex and switching back to regular caps.

\$\endgroup\$
  • 2
    \$\begingroup\$ What type of 0603 caps were you using? You could switch to a type that fails open instead? \$\endgroup\$ – DrFriedParts Sep 20 '18 at 19:27
  • \$\begingroup\$ The big question is what was causing the failure. Soft termination can help with both board flex and cracking from thermal shock. If the cracking and shorting was a result of board flex then the questions is comparing repair cost to production cost. If it is a thermal shock question then you need to look at the soldering process and temps. \$\endgroup\$ – Robert Fay Sep 20 '18 at 19:34
  • 1
    \$\begingroup\$ I'm trying to imagine how much a board has to flex to break a 0603 sized part, and my imagination keeps saying "whoa, scary for a board to bend that far." \$\endgroup\$ – JRE Sep 20 '18 at 19:35
  • \$\begingroup\$ @JRE Actually with ceramic caps (that really, really don't like to flex) it does not take a lot, the bigger ones are worse, but an 0603 near a board edge guillotine cut line will do it. Even worse a part so damaged is a latent failure waiting to happen. \$\endgroup\$ – Dan Mills Sep 20 '18 at 20:04
  • \$\begingroup\$ @DanMills: I know ceramics don't like to flex. Its just 0603 is so small. Used to work with some equipment with a 1206 sized 330nF that liked to break. For some strange reason, they always had a gap underneath (the caps were up off the board.) Any pressure on the caps would cause a crack. Like to have gone crazy looking for the problem before I figured out it was the cap. \$\endgroup\$ – JRE Sep 20 '18 at 20:11
2
\$\begingroup\$

The more you use component the less you have reliability but some times it is not so. Normal MLCC capacitors are vulnerable against tensions due to assembly process and after that especially during lead free process that is much hotter.

soft termination caps are really more reliable but they are not the first choice for mass production even in safety critical applications. In mass production the solution is using two serial normal MLCC capacitors those are assembled perpendicular to each other in the PCB. If one of them became short circuit due to tensions in one direction, the other could do part of the job obviously not as good as a single 0603 capacitor but more reliable than that and also more cheap than soft termination capacitors those are also hard to purchase from the market.

\$\endgroup\$
  • \$\begingroup\$ If you put two caps in series, e.g. 1uF, then you get half the capacitance, (0.5uF). So, if one fails short, then the capacitance INCREASES from 0.5uF to 1 uF. So it will work better if one fails. Unless the short has high DCR, then maybe it will not work well. \$\endgroup\$ – mkeith Oct 2 '18 at 21:45
  • \$\begingroup\$ When we replace a single capacitor with combination of two series capacitor we are increasing the ESL that is not good for frequency response but remember our design must not be perfect it must be good enough. \$\endgroup\$ – BD_CE Oct 3 '18 at 3:47
  • \$\begingroup\$ In your answer you said that "if one of them became short circuit due to tensions in one direction, the other could do part of the job obviously not as good as a single 0603..." My point, I guess, was that once one cap fails, the capacitance increases. So in some ways, the circuit performance may be better AFTER the cap fails. \$\endgroup\$ – mkeith Oct 3 '18 at 3:56
  • 1
    \$\begingroup\$ Yes from capacitance point of view, you are absolutely right but when one of the capacitors became short, we have more capacitance that is good but higher ESL than a single capacitor due to PCB longer traces and the shorted capacitor package's intrinsic ESL that is weaker than single capacitor from frequency response point of view. \$\endgroup\$ – BD_CE Oct 3 '18 at 4:10

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.