1
\$\begingroup\$

I am troubleshooting a design that uses a 47 µF 20 V tantalum SMD cap across a 15 Vdc rail.

While the derating guidelines I follow dictate a 50% voltage derating for tantalum caps, the design I'm looking at slipped through the cracks, and 15 Vdc is applied continuously.

Is it possible that these tantalum caps could short from over-voltage and then self-heal, causing the overall circuit to malfunction for a moment, but upon later inspection appear fine?

These caps are across the rails of a relay driver IC. Supposedly relay contact closures were detected, but no commands from the micro that would result in relays energizing were recorded.

If self-healing tantalum caps are a possible culprit, what is the timescale for the self-healing process? I have not been able to find any data with regard to healing time.

\$\endgroup\$
  • \$\begingroup\$ Self-healing from what? What makes you think it would be hurt at 75% of it's rating? Why do you think that would cause a relay to fire? \$\endgroup\$ – Trevor_G Feb 21 '18 at 21:55
  • 9
    \$\begingroup\$ I thought Tantalum caps tend to explode violently, not self-heal \$\endgroup\$ – user253751 Feb 21 '18 at 22:12
  • 1
    \$\begingroup\$ I have blown enough and I have never seen one heal. But then I normally replace them after the top has disappeared. I have never tried to see if they heal. \$\endgroup\$ – Oldfart Feb 21 '18 at 22:26
  • \$\begingroup\$ @immibis. They also fail by oxidation, resulting in a 99% drop in capacitance. Usually a latent failure. I have never seen one heal itself. If it was bad it was replaced. \$\endgroup\$ – Sparky256 Feb 21 '18 at 23:59
  • 1
    \$\begingroup\$ Dry tantalums only self-heal if they are on a high impedance source. (The short from the tantalum slug to the MnO2 causes a local oxidisation and converts the MnO2 to MnO3 which is an insulator). On a low impedance source, a short circuit within the device will usually have what is quaintly termed an ignition event. \$\endgroup\$ – Peter Smith Feb 22 '18 at 12:51
3
\$\begingroup\$

I guess I'm confused by your question. If I'm reading your question right, you've seen the capacitors fail, and you're wondering why they failed and how long it will take them to self-heal.

To answer the "why" of this situation: If they failed, it's not because of an overvoltage. Yes, they're not derated enough, but they should still withstand 15 VDC without failing immediately. It's most likely because your 15 VDC isn't current-limited, and the current surge triggered a short. This is a common and dangerous failure mode of Tantalums.

As far as self-healing, in general, self-healing should happen rather fast. But to be clear it will only happen during operation since it's based on the localized heating due to current passing through the dielectric fault in the capacitor. If you're just leaving a tant unpowered then it will never heal.

Also, you can only rely on self-heating for fixing mild shorts/dielectric defects at reasonable currents. If you put an non-current-limited 15 VDC across a Tant with no slow-start circuit then you will get a shorting failure mode from the current surge and something will explode long before the cap self-heals. Self-healing should be relied on to fix manufacturing defects in the dielectric, not full-on shorts due to a current surge.

\$\endgroup\$
  • \$\begingroup\$ That all makes perfect sense... except the caps are 20V caps.. \$\endgroup\$ – Trevor_G Feb 22 '18 at 14:04
  • \$\begingroup\$ @Trevor_G You might have misread my answer, that isn't in disagreement with any of my points. \$\endgroup\$ – jalalipop Feb 22 '18 at 14:26
  • \$\begingroup\$ Ya it's the line "they should still withstand 15 VDC without failing immediately" implies they will eventually fail at 15V... \$\endgroup\$ – Trevor_G Feb 22 '18 at 14:44
  • 1
    \$\begingroup\$ @Trevor_G Well, they will eventually fail. Tantalums have terrible lifespans, especially at high temperatures. There's a reason they're derated aggressively; a 25% derate would not make it past any design review. However, that's not the failure mode I discuss in my answer. The failure mode I'm referring to is an immediate one due to initial surge current and is irrespective of the rail voltage. I suggest you research surge current in tants for a better explanation than I can give. \$\endgroup\$ – jalalipop Feb 22 '18 at 14:47
  • 2
    \$\begingroup\$ Tantalums that have been through reflow will not withstand their rated voltage from a low impedance source 100% of the time if they have been through a standard reflow profile. Indeed, they will not withstand 50% of their rated voltage. This applies to case sizes D and larger. \$\endgroup\$ – Peter Smith Feb 22 '18 at 18:32
2
\$\begingroup\$

Tants have a self-healing mechanism but it’s less than perfect. Derating them by 2:1 is little enough and running them at 75% is really not a good idea. In applications where there is a serious turn-on current surge 3:1 derating is better.

\$\endgroup\$
  • \$\begingroup\$ IPC-9592B specifies 60% for "normal" equipment with 5 year operating life and 40% for "extended life". \$\endgroup\$ – Barleyman Feb 22 '18 at 13:24

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