Nigel Goodwin

Capacitance doesn't always go down when it fails, that's the whole point - you also can't usually measure capacitance in-circuit, where as you can ESR.

Having said that, my new meter does both (or at least trys to), but while it always reads ESR in-circuit, sometimes it just displays the capacitance as 'in circuit' rather than a value.

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

HS3

From what I understand each one of us is talking about something different. The OP asked about capacitor characteristics in general, no specifically about trying to diagnose possible faulty caps in-circuit.

It seems some are trying to argue against a straw man here because at the beginning of the thread I said
Let us stop moving the goal posts around because we're all saying essentially the same thing.

HS3

I would qualify that by saying that a very common way, the most common way, for caps to fail is because due to heat or too much ripple current or cheap cap or whatever the ESR goes up a bit, which causes the cap to heat up more which in turn causes electrolite evaporation which causes higher ESR and the cap enters a spiral of growing ESR and growing temperature which ultimately leads to its destruction. It would be rare that you catch a capacitor in the middle of this process. By the time you get the electronic circuit the cap has usually failed catastrophically or has not started the degradation process. Once I get a power supply with a bad cap I look at the rest and if they are of the same cheap brand I replace them all even if they test good. There is no point in saving a few pennies only to have another cap fail later on.
Yes that is the main advantage of testing ESR: that it can be done in-circuit. Unsoldering components is a PITA.
"Always" leaves no alternative possibility. I would say "Almost always" because I think that any ESR meter would have a problem measuring ESR of a capacitor which has a very low resistor in parallel. This is, admitedly, a very rare situation in real circuits but it does happen.

So, yes, we agree, the *first* test I do on any faulty SMPS is test all caps with the ESR meter and most of the time it detects bad caps . Maybe replacing the caps is all it takes but very often the failure of the caps has lead to the destruction of the switching elements and other disasters.