??? 05/11/06 03:35 Modified: 05/11/06 03:40 Read: times |
#115971 - Tantals aren't unreliable! Responding to: ???'s previous message |
Bruce said:
I had no idea these smt tantalums were so unreliable! They aren't unreliable, but you have made a mistake, you violated the design rule to provide a source impedance of more than 0.1Ohm/V! The life time of a tantal is normally specified for a source impedance of 3Ohm/V. Life time of todays tantals is very long, especially when compared to cheap aluminium electrolytics. But failure rate increases by a factor of about 10 when decreasing source impedance to 0.1Ohm/V! In most circuits there's enough source impedance to keep the reliability of tantals high. Either you use resistors in series to the tantals (forming RC-lowpass filters, heavily enhancing the power supply decoupling properties) to reduce the inrush currents, or you have just hanging many of the tantals in parallel at the same Vcc to limit the inrush current to a sane value. In the first case even rather small 2R2 resistors at 5V yield a source impedance of 0.44Ohm/V, keeping the reliability of tantals high, not mentioning that the source impedance of supplying circuit can show even much higher source impedances (see below). In the second case, the maximum surge current of voltage regulator, which is assumed to power the tantals, is divided by the number of tantals hanging in parallel. For a LM7805 regulator maximum surge current is less than 3A. Then, ten tantals have to absorb only 0.3A each, which represents a source impedance of roughly 3Ohm/V. You see, in "normal" applications this minimum source impedance is often easily provided. But if you short circuit a power supply showing ultra low source impedance into an unprotected tantal, then it's no surprise, that trouble arises. Rob is right, this ceramic highcaps offer the best decoupling performance avialable today! Nevertheless, in such an application, where a source is used showing ultra low source impedance, I would highly recommend to use some current limiting in any case. Otherwise inrush currents will be huge! Anyway, in your application I would always use a pi-filter at the input of your circuit instead of a simple decoupling cap (even if some caps are paralleled). Because this will heavily enhance the noise filtering performance and provide sufficient inrush current limiting at the same time. Kai |
Topic | Author | Date |
Tantalum on power supply filter - Blowup | 01/01/70 00:00 | |
does it get (luke)warm ? | 01/01/70 00:00 | |
Nope | 01/01/70 00:00 | |
Symbol is marking the "+"-pole! | 01/01/70 00:00 | |
Verified | 01/01/70 00:00 | |
Surge current | 01/01/70 00:00 | |
Some interesting reading | 01/01/70 00:00 | |
Much thanks! | 01/01/70 00:00 | |
Pardon the pun.... | 01/01/70 00:00 | |
Tantals aren't unreliable! | 01/01/70 00:00 | |
not a tantalum pi | 01/01/70 00:00 | |
pi-filters | 01/01/70 00:00 | |
Guilty | 01/01/70 00:00 |