Testing vintage caps for leakage

[Pdavis68]

As per Enzo's recommendation, I am trying to test the caps I got from a '59 Conn Caprice organ for leakage.

The caps are all 200V and 400V, so I've set up a power supply that provides 190V and 390V, roughly.

I've got several RC stages and the ripple is down pretty low.

I've tested about 10 caps so far and none of them are showing much DC leakage and so I wanted to verify I'm doing this right.

I'm taking my 390V and that goes into one end of the cap. On the other end of the cap I have my DMM and oscope probes clipped to it. And then the DMM and oscope have their grounds connected to the common ground.

If I set my DMM for AC, I can see the ripple coming through and of course, I see it on my scope as well. But what I'm not seeing, is any appreciable DC coming across. I have my DMM set for the 200mV level and so far the caps I've done are showing about 5mV tops so far. Most are under 3mV.

Am I doing this right? I don't have any known leaky caps to test, but I did test with a 4.7M resistor and it would have made for a seriously leaky cap, so I'm guessing this setup is good.

[52 Bill]

If I was working on an amp and wanted to test a coupling cap for leakage, I would disconnect the end away from the plate power supply and read the dc voltage with my meter, which is exactly what you have done with your test setup.

[Pdavis68]

Awesome. Just wanted to make sure. After my wife goes to bed I'll probably sit around and start trying to test the rest, but so far it's looking like I've got an excellent batch of vintage caps.

[Pdavis68]

What's an acceptable level of leakage? So far (I've done about 40 caps) they're all under 5mV. Most are 2-3mV.

[Tom Phillips]

I interpret your test results to mean that all the caps you tested are OK. If you run across a leaky cap it will probably stand out by showing significantly more leakage voltage. You can always run some new modern caps through your test to establish a baseline.

I'm not sure exactly what you meant by "...did test with a 4.7M resistor and it would have made for a seriously leaky cap." If you just substituted a 4.7M resistor for the cap then Yes, I'd expect a significant dc voltage reading on the metered end of the resistor. In that setup you just have a resistive voltage divider with the test equipment acting as the lower leg of the divider.

Cheers,
Tom

[Pdavis68]

Tom, I was kind of joking about the resistor. I mean, I did actually test it just to make sure I'd be seeing a voltage on the far side because i was just surprised so far, none of these 50+ year old caps were showing up as leaking.

I will test with a modern cap, but I figured I'd be seeing something over 10mV by now. But everything so far has come in at 5mV or less. But I wanted to know what a reasonable cutoff is. Is 10mV? What about 50mV? 100mV? 500mV? I just don't know at what point you declare it leaky. I don't have any idea what the scale is. I assume a volt is too much, but other than that, I don't know.

[Enzo]

I suppose in a test situation like that I wouldn't want very much leakage, but in circuit, we are usually dealing with old amps, and so we are going to find either working caps or seriously leaky ones. In other words, say I have 200v on a triode plate and a cap to a following grid. If I find +12v on that grid, my cap is leaky. If a few millivolts were leaking through, they might just disappear into the circuit. I don't usually measure the leakage, I usually measure the performance in the circuit.


Establish your own baseline. You are testing a lot of caps from the same source, so they all probably have the same age and have endured the same conditions. SO I'd look for ones that differ substantially from the rest, rather than ones that don't meet an arbitrary standard.

[Pdavis68]

Sounds reasonable. My main purpose in testing them is for selling them on e-bay. I can get more if I guarantee they're not leaky. I will and I'll offer refund or replacement on request. That should get me a decent price.

[Tom Phillips]

...I wanted to know what a reasonable cutoff is. Is 10mV? What about 50mV? 100mV? 500mV? I just don't know at what point you declare it leaky...

My experience is that most caps in the 0.001 to 0.1uF range will settle below 5mV within a few seconds. I accept that as good. The leaky ones stay well above 10mV and I consider them leaky.

[Pdavis68]

Awesome. Thanks, Tom!

[vmazz]

I dont have a lot of experience with leaky cap testing . But I have recently had some trouble with symptoms in an amp which sound like dc leakage on coupling caps. When I test with my multimeter I get around 5 volts of DC on the grid side which should be 0 I think. I have replaced 4 different caps in this same spot and they all end up(after a day or 2) to show dc voltage. I assume this is leakage from the plate? But very wierd that the same spot would all result in leaky caps. SOme are brand new mallory others nos polystyrene. When tested with Eico 950 they dont show much leakage up to 400v (rated at 400v caps). Is there another way that dc voltage would be getting to this node. ITs the v1b to v2a coupling cap in a Marshall 2203 style circuit. Any suggestions appreciated. Thanks.

[Pdavis68]

Have you tried replacing the tube?

[Enzo]

That was my first thought too.

[vmazz]

Pulled tube and tested good for leakage. Also right now there is no dc leakage at that point since I last changed the cap this morning. It has come and gone as I change the cap it goes away but then comes back....maybe I had a string of bad leaky caps? Time will tell , if this cap last a week Ill figure that was the problem.Thanks for your input.

[vmazz]

anyone who wants to help with my situation can follow the other thread I started here........http://music-electronics-forum.com/t40635/