Can a capacitor test fine, but fail under load?

[hberg32]

Has anyone seen a failure mode for capacitors where they look fine and test fine on a multimeter, but short under vcc (in this case 3.3v)?

I'm working on repairing a damaged circuit board and accidentally powered it up with a bad solder shorting vcc/ground for a few seconds (a bit of smoke coming from the ferrite bead alerted me to the problem). Since then I've very carefully done a continuity check between vcc/ground before powering up to make sure there is no short but it keeps happening anyway at power up. If I disconnect a portion of the circuit the problem goes away so I think I can rule out the voltage regulator and narrow the fault down to a collection of tiny SMD caps that are probably just a few pF each and one larger 68 uF cap. I've removed the larger cap and it tests ok for capacitance and resistance on the multimeter. I tested the others in-circuit for resistance and didn't see anything odd. Alternately, does it seem possible that the voltage regulator is damaged in such a way that it will work fine with reduced load but fail short with more load? The regulator in question does not appear to have short circuit protection (LT1129-3.3)

Thanks,
Henry

[adafruit_support_mike]

A shorted capacitor should have a measurable (and low) DC resistance for the kind of failure you described.

There's an easy test though: connect the cap's negative lead to GND and its positive lead to a 10k resistor, then connect VCC to the other end of the resistor. If the voltage across the capacitor measures the same as the supply voltage, you don't have a short.

As an aside, capacitors get stronger as they get smaller. SMD caps have values up to 10uF down to 0402 and farther.

Testing capacitors in-circuit takes an oscilloscope and an AC signal of about +/-150mV. That's low enough to avoid turning on any semiconductors, leaving you with simple(r) RLC circuit behavior. Putting resistors between the circuit and the test signal and/or GND can give you more information, especially the capacitor values.

[hberg32]

Thanks for the reply, I've harvested the cap that is my chief suspect and will run the test as you suggested.