Capacitor Troubleshooting

[DigiTan]

I need to troubleshoot this series of old PC card where 1/3rd have been failing after several years of service. The age of the board makes me suspect the passive components are well past their prime. Particularly, there are about 15 tantalum capacitors on the board that are mostly rigged up as bypasses.

I'm thinking of renting one of those capacitor ESR meters that can test the capacitor in-circuit. On the off chance I end up keeping the ESR meter, I'd like to know how the failure modes of tantalum capacitors compare to other capacitor types.

Questions:
* Are these ESR meters pretty good at spotting tantalum capacitor problems?
* Can they detect problems in ceramic or electrolytic capacitors just as well?
* Anyone know of a good ESR meter schematic?

 

[Mikebits]

With a scope just check the DC level of the line being bypassed (I assume Vcc). Signal should be free of AC component, or less then 50 mv or so. If you find a large amount of AC, then the AC will probably be largest near the bad cap, if one exist.

 

[DigiTan]

That's a pretty good idea. There's almost a dozen mixed signal ICs on this board and I'm suspecting a large part of this capacitor population just can't filter the noise anymore. It's something I'll look into, but since my access to a oscilloscope isn't guaranteed, I'll still have to consider the ESR until further notice.

 

[Nigel Goodwin]

In my experience the failure mode of tantalums is almost always short circuit.

However, the ESR meter below (which I've recently bought) displays both ESR and capacitance, and is reasonably priced.


Peak Electronic Design Limited - Atlas ESR - Equivalent Series Resistance Meter and Capacitor Analyser - Model ESR60

 

[mneary]

If you have a large failure rate with similar symptoms you may be looking for one overstressed part (design issue). Tantalum bypass caps seldom fall into this category unless someone forgot to de-rate them. In the case of tantalums you should see at least a few exploded caps.

 

[kchriste]

I'll second Nigel's comment that the most common failure would be shorted caps. Unless you have schematic of the circuit you are trying to test, testing the caps in circuit will be a fruitless effort. Even then, testing in circuit is not accurate and can be misleading.