How to test for leaky capacitors?

[wattsup]

How do I check for a leaky capacitor? I'm especially wanting to know how to with the power supply filter caps and coupling caps too. I know this may be a real basic question but I'm trying to learn.

[Firestorm]

Two different things. Coupling caps you disconnect the low voltage side and measure DC to ground. It will try to fool you, though, because initially the cap will show a bit of DC (caps do this) and then settle down. You need a good meter. Something with Fluke in the name, probably. Acceptable leakage depends on how badly it affects the next stage.

Filters though are different. You don't want too much DC current going through the cap to ground. Ideally there will be none. The simplest test is age. If it's old, replace it.

[sluckey]

An ESR meter is good for electrolytics.

In the days of analog meters there was a crude but fairly reliable method to check for leaky electrolytics. Connect an ohm meter across the cap. The needle will quickly swing toward zero (may even peg the needle momentarily) Then the needle will slowly drift back toward infinity as the cap charges. A good cap will stop at some high resistance reading and remain constant. A leaky cap will stop at some resistance (may or may not be high) and then will begin to creep back toward zero as current increases. This 'creep' indicates a leaky cap. A steady low resistance reading also indicates a leaky cap. This is one of those checks that you develop a feel for by testing known good caps as well as bad caps. Different meters will give different resistance readings. The actual reading is not important. You're looking for the 'creep' or a relatively low resistance reading.

Another indicator of a leaky electrolytic is temperature. A cap with lots of leakage will likely be warm to the touch. The temperature of a good cap should be the same as the ambient air temp. Of course, if the cap is located near a radiant heat source such as a tube or PT the touch/temp check may not be valid. Use caution when touching a cap with a metal shell. The shell may be connected to a high voltage! These are usually insulated but not always. If unsure, poke it with a voltmeter probe first.

[xtian]

Another low tech test: use your eyes. If it's leaking, it's leaky.

[wattsup]

Two different things. Coupling caps you disconnect the low voltage side and measure DC to ground. It will try to fool you, though, because initially the cap will show a bit of DC (caps do this) and then settle down. You need a good meter. Something with Fluke in the name, probably. Acceptable leakage depends on how badly it affects the next stage.

Filters though are different. You don't want too much DC current going through the cap to ground. Ideally there will be none. The simplest test is age. If it's old, replace it.

OK, for the coupling caps that's easy enough.

For filter caps, do I unsolder them from the board and hook them up to a low voltage source such as a 6V lantern battery, then measure from neg side to ground?

[tubeswell]

For leaky coupling cap tests, see Page 57 in the attachment

[wattsup]

Thanks tubeswell

[dorrisant]

Here is a good project:

http://members.shaw.ca/swstuff/esrmeter.html

PCB layout included. I built one of these about five years ago. Works great, use it almost every day.

[eniam rognab]

Thanks Tony!