Capacitor reliability

Hermetic tantalums do not like surge currents, high ripple, and reverse bias. Non-hermetic tantalums do not like moisture (their useful life is shortened). The voltage rating becomes slowly degraded over time. Surface mount capacitors are destroyed and/or damaged if the board is flexed. Otherwise they are relatively immune to: surge currents, high ripple, reverse bias, and moisture.

mark wrote:

Are you using military grade caps, or, have you compared at the reliability numbers for commercial grade caps in 217F? That's one point.

The other point -- when I do reliability analyses, I go to the manufacturer for the latest measured failure rates. It is only as a last resort that I use 217F's empirical failure rates (although they are still very useful). Check with the manufacturer and see if they have published failure rates for their caps.

Tom

Tom thanks for the thoughtfull reply. Yes, I am at the last resort stage. My problem is: should my designs go wholesale to MLCC (if I can) instead of tantalum chip? Is there a failure mechanism in tantalum which tells me I should avoid them? At this time I am not aware of any. What is the Reliability Engineering community seeing?

M Walter

In my experience, which is limited to low volume production (1000s of units), I have seen 2 tantalum chip caps (CWR type) fail catastophically (explode on-board), while I have never seen a multi-layer ceramic fail. Our designs are very conservative as to applied voltage (stress). for one of the failed caps operator error involved (i.e. hooking up a power supply backward, etc.). I didn't read the final failure report as to the cause of failure (wasn't my project), but I heard the parts lab "ruled" it an isolated event, not indicative of a systematic type parts flaw. For the other failed cap, it was on a prototype and the power supply was accidentally connected backwards (this was my project).

Our company doesn't use "dipped" radial tantalums, so I can't speak to that.

I still design with CWR tantalums, and have not been told by the parts-engineering group to quit using them

If you can stand the difference in ESR, size, ESL, and other parameters of the MLCCs compared to the tantalums, to me, that would be the deciding factor -- especitally if you have any doubts about reliability.

Tom

There isn't a simple answer. Manufacturing operations and the hardware's environment induce failure modes. And often, there are "failures" and there are "FAILURES". If your hardware isn't intended to be repairable or live in an explosive atmosphere environment some of the types of failures might be non-issues. If you can stand the low voltages available, the tantalum polymer types are a very good all around choice.

tlbs101 wrote: Tom, Not sure I agree that all things being equal mlcc are the answer, hence why I am asking I am aware of some issues w tantalum caps bursting into flames. Not my major concern (due to other design factors): I am more worried about inherent reliability.

I base some of my answers/concerns from Mil-Hdbk-217F data. The '217F base failure rate for tantalum caps is 10 times better then the base failure rate for ceramics. The penalty for operating at elevated temperatures is higher for ceramics, and if you were derate both to around 50% of rated voltage the stress factors are about the same. Given that Mil-Hdblk-217 says use a tantalum cap where temperature will be high. Also, I get concerned about cracking ceramics in the large case sizes required for say a 4.7 uf device

If I jump onto the mlcc bandwagon to replace tantalums will I be going from the frying pan into the fire?

M Walter

What's the failure mode though ?

Graham

I'm seeing some increased leakage and increased ESR after hi temp exposure. Root cause was identified as cracks in the dielectric film. Also the amnuf at least has said that it is due to reflow soldering process: ind test lab say that cracks are from manf process faults. M Walter