Using DMM to measure coupling cap leakage

[martin manning]

Regarding the measurement of coupling capacitor leakage (in-circuit, with the low-voltage end lifted) RG suggests the following:

If you only have a Digital Multi Meter (DMM), connect a 1M ohm resistor from the free end of the capacitor to signal ground, then measure the voltage across the resistor.

So now I'm curious, what is it about a DMM that causes the problem?

MPM

[Bruce / Mission Amps]

What DMM problem? I'm not following that unless maybe giving the cap a small AC/DC load so the cap doesn't charge up and measure like a battery??
A decent DMM has something like 10Meg to 11Meg impedance, even at high voltages vs a few hundred thousand of an analog meter, so the DMM offers no real drain load and is super sensitive compared to a clunky old analog meter.
That's my guess...

[martin manning]

I think its something like that. I believe a typical VTVM is around 11 meg, and is supposed to work fine for this test without the shunt resistor. I believe a DMM is comparable, but that might be a function of the voltage range in use, and there may be some issue with the auto-ranging feature. As you say, without the 1 meg resistor to ground, the floating end of the cap would initially be at high voltage, and then it would have to drain through the meter.

I was hoping RG might elucidate.

[R.G.]

You rang, sir?

The ideas behind pulling the low voltage end loose and measuring *voltage* between the open end and ground is this:
- a poly or ceramic cap that's leaky may still be a quite large resistor; non-leaky ones are massively more than even a 10M resistor
- multimeters don't do a good job measuring resistances this high, even "low" ones in a "leaky" cap.
- Using the meter to make a voltage divider with the cap and then measuring the divided voltage is much easier.

This technique was current (in the sense of "common at that time") in the analog meter days, when an analog meter at 20K ohms/volt on a 200Vdc scale might have an input impedance through the switchable divider of 1M. A leaky cap with a 10M equivalent leakage resistance and 200V on the high voltage end would show a DC voltage of 20V on the meter when it was divided through the 10M leakage and 1M meter input. The correct result for this test was the meter showing dead zero volts. Othewise you'd replace the cap. This was how it worked in the day when there was a radio/TV repair shop in each town and the guy who ran it had soldering iron scars on his hands.

Old analog meters had their flaws, but sometimes they did things which DMMs can't. In this case it was having a predictable input resistance, which you knew from the ohms/volt rating of the meter. A DMM may have any resistance from 1M on up, and it may be quite large. It may even have a near-infinite input impedance in some conditions. Not a bad thing really, but unpredictable. In most cases in electronics, it's better to have something less perfect but predictable than unpredictably perfect.

You can make a DMM more predictable by putting that 1M to ground. Then the DMM is measuring voltage again, instead of divided leakage current. And that combination will be pretty predictable (in the sense of "consistent") across situations.

In all cases, the right voltage to look for in the series capacitor test is zero. A good non-electrolytic coupling cap should have a voltage reading on this test of zero - which is not really zero, it's just that the cap's parallel leakage resistor is so large that a 1M pulldown, faked by the added resistor, really pulls down the DC voltage, with leakage to zero volts, which is where you want it for a tube.

Did I just muddy it back up?

[martin manning]

No, I think that clears it up- I guess we really want to know the voltage/leakage current with the cap connected to a load impedance which is similar to that in the circuit it is being used in. Thanks!

[mooreamps]

Regarding the measurement of coupling capacitor leakage (in-circuit, with the low-voltage end lifted) RG suggests the following:

If you only have a Digital Multi Meter (DMM), connect a 1M ohm resistor from the free end of the capacitor to signal ground, then measure the voltage across the resistor.


MPM

Maybe not the way I would do it. The series resistor and a voltage meter in parallel limits the leakage current. I've done this type of measurement in my younger days.
You would instead lift the low voltage leg of the cap, put one probe of the multi-meter set at 50 uA current mode, and then the other probe goes to ground. I've first done this on a Simpson 260 multimeter, but the Fluke meters will work just as good.

-g

[J M Fahey]

the multi-meter set at 50 uA current mode

Nice trick.
By using the 50uA current setting you are cloning the needle multimeter 50uA movement.
Analog multimeters are *always* current meters, which for voltage measurements put calibrated resistors in series with it (That's where the 20Kohm per Volt comes from)

[mooreamps]

Nice trick.
By using the 50uA current setting you are cloning the needle multimeter 50uA movement.
Analog multimeters are *always* current meters, which for voltage measurements put calibrated resistors in series with it (That's where the 20Kohm per Volt comes from)

Yes it does !

-g

[Diablo]

Can you measure the voltage on the coupling cap while it's in circuit? If not, how would you guess that the cap is leaking?

[mooreamps]

Can you measure the voltage on the coupling cap while it's in circuit? If not, how would you guess that the cap is leaking?

On a guitar amp, I would take a voltage measurement on the control grid of the next gain stage. Should be close to zero volts. Anything positive voltage, then I would lift one leg of the coupling cap, and then re-measure.

-g

[stokes]

For coupling caps,I've always just measured in circuit and if I see more than 1/4 volt,on the low voltage side,I change it.I know it aint precise,but it hasnt failed me yet.

[Diablo]

Thanks -g & Stokes. That's what I had assumed but I was starting to have my doubts when this thread started going in the direction it did.

[frankeg]

Many years ago, I was troubleshooting a faulty industrial relay with timing circuits using a dvm. I couldn't find the fault. A SR. tech suggested I use an analog meter. Presto - found a leaky cap. It seems the dvm was too fast and gave a constant rms reading. With the analog meter, I could actually see the needle deflect.

[koen]

For coupling caps,I've always just measured in circuit and if I see more than 1/4 volt,on the low voltage side,I change it.I know it aint precise,but it hasnt failed me yet.

I see that on two of my coupling caps - didn't realize that such a low voltage could mean they are leaking. I'll replace them tonight.

[trem]

Pardon me for not getting the crux of this.
Is R.G. saying it is hard for a DMM to measure voltage between Ground and the end of a very large (10M perhaps) resistance...or is the problem something else.?
Thank You