1
\$\begingroup\$

I am using a tantalum capacitor of 10 uF rated at 20V as a decoupling capacitor. It is placed in parallel with a 5V supply. Now when the circuit is powered, the capacitor burns out.

I replaced the tantalum capacitor assuming that the previous capacitor was defective. But when I powered the circuit the capacitor burned out again.

I replaced the tantalum capacitor with electrolytic capacitor of 10 uF rated at 63V. This time the capacitor did not burn after powering the circuit.

I do not understand why the tantalum capacitor burns up?

\$\endgroup\$
  • 1
    \$\begingroup\$ Could it have been installed backwards? (by the way, you would use a tantalum as a filter capacitor. They make poor decoupling capacitors). \$\endgroup\$ – Mark Jul 12 '16 at 2:47
  • 1
    \$\begingroup\$ How confident are you that the 5V supply is always 5V? Also, have you verified that you are not exceeding the ripple current specs? \$\endgroup\$ – uint128_t Jul 12 '16 at 2:49
  • 1
    \$\begingroup\$ If you energize a cable, then plug it in, the long wire inductance can cause a nasty spike: cds.linear.com/docs/en/application-note/an88f.pdf \$\endgroup\$ – Daniel Jul 12 '16 at 3:34
  • \$\begingroup\$ Does your 5V supply have a minimum load spec that is being met? \$\endgroup\$ – Tut Jul 12 '16 at 10:49
  • \$\begingroup\$ Check the polarity markings ... that band might be for the +ve terminal and not the -ve terminal ... Standardization ? what's that? \$\endgroup\$ – Spoon Jul 12 '16 at 11:30
2
\$\begingroup\$

Tantalum capacitors fail if you install them backwards. No surprise here.

They may also fail due to high surge current, e.g. if the capacitor is initially empty and then connected to a power supply with high peak current capability.

When connected to a power supply a high initial current will be present at the capacitor (can be in the order of 10 Ampere and more). This is due to the very low internal series resistance of the tantalums.

Albeit this is a short event it causes heat and stress within the capacitor and may result in a failure. To find out if this is the culpit you should - as a test - connect the tantalum with an additional series resistor of lets say 1 Ohm and see if this helps. If it does you have to limit the initial inrush current of your device.

Note that you may see such high current spikes even if your power supply is only rated at 500mA or so. One or multiple large and low ESR output capacitors are all what is needed.

\$\endgroup\$
  • \$\begingroup\$ I've had tantalum caps that were installed backwards by a board stuffing house catch on fire with open flame when powering up the boards for first time testing. \$\endgroup\$ – Michael Karas Jul 12 '16 at 4:15
  • \$\begingroup\$ @MichaelKaras I have seen this as well, but it wasn't the board house but me. Will never forget that smell! But in general I avoid tantalums if possible because of the coltan mining situation. \$\endgroup\$ – Nils Pipenbrinck Jul 12 '16 at 4:31
1
\$\begingroup\$

Assuming the part was installed correctly (it's possible for Tantalum caps installed backwards to work for quite a while, especially if the reverse voltage is low compared to the rating), your power supply may be capable of too much current. This is a frequent issue when Ta caps are hot-plugged to batteries and such like with a low source impedance.

Some manufacturers recommend putting series resistor on the capacitor (of course that makes it less useful as a bypass capacitor) as well as using a voltage rating of 3x or more the working voltage.

At 10uF you can successfully use a ceramic capacitor (perhaps it will be a bit bigger) and avoid the ignition issue. You may have to use 15uF or 22uF to actually get 10uF at the working voltage, especially if you choose a relatively small capacitor that is rated for close to 5V.

\$\endgroup\$

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.