Tantalum Capacitor Failure 4

[Tintagel]

Anyone have any information / experience relating to tantalum capacitors failing short? Anything on manufacturing process or circuit transients causing the failure would be greatly appreciated.

 

[Comcokid]

The voltage rating on tantalum capacitors is the voltage maximum. Exceed it, and they're gone. You can temporarly exceed the voltage rating on most other types of capacitors, but not tantalums. When selecting a tantalum for a circuit, be sure to include some extra margin in the voltage rating to cover the unexpected. Tantalums also don't like reverse polarity.

The failure mode tantalum capacitors is to short. Some varieties of tantalum capacitors are made with an internal fuse for circuits that cannot tolerate a short.

 

[Skogsgurra]

If you are using double (or more) power supplies, there is a very real risk that one of them gets reverse polarity during shut-down. As Comcokid says; they do not tolerate that.

The reason for the reverse polarity is often that one of the supplies (usually the negative) gets "overridden" by the positive (which usually is the heftier one) and kills the tantalum capacitors.

It is difficult (impossible) to avoid this in many circuit types - especially if there is some sort of power stage involved. If you think that this is the problem (you should check the voltages with a memory scope at turn-on and turn-off) the easiest solution is to add diodes parallel to the supplies (I don't bother to mention which polarity they shall have ;-) )

 

[sreid]

I've lunched some by exccding the ripple current rating. This is easy to do in switching power supplies. Touch the caps with your finger; it they are hot, that's your problem.

The fix is low ESR caps or paralleling standard ESR caps.

 

[logbook]

The don’t like switch on surge currents either. I have seen manufacturers recommend putting resistance in series with the tant to prevent too high a dV/dt being applied (and hence too much switch-on transient current). If you have a mechanical power switch on a low voltage power rail the rate of rise of the power rail would not be limited; this is a good way of reducing the lifetime of the tant.

Several years ago we were supplying equipment to a Japanese company and they insisted that all the tants be replaced by ordinary electrolytics and solid dielectric capacitors. This was a huge task, but it was a big order so we did it!

 

[walker1]

A real problem with small tantalum capacitors is, that they fail with large currents.

That is, do not for instance use them in power supplies for ripple reduction. At least not if the rush-in/out current at any time can be greater than, say, 500 mA.

You can get tantalum based capacitors, which allow for a larger current, but for the small epoxy pearl things:
Keep the current low.

 

[Tintagel]

In the problem I have in mind, the tant is supplied via a P-channel FET which has a slugging cap across gate-drain with a pull-up resistor gate-source to ensure turn-off. There is no other series resistance.

Measurements seem to indicate that there isn't a huge inrush current, but something is causing it to blow from time to time

 

[logbook]

Does the measured current transient agree with the current in the capacitor calculated from

I=C*dV/dt ?

 

[felixc]

There is a temperature derating on tantalums.
The cap at the fet gate is possibly there to reduce the turn-on time. Good thing.
The maximum current peaks are usually documented in the spec sheets.
What's the voltage rating of your cap? Have you tried one with a higher voltage rating?