Coupling cap leakage how much?

[diydidi]

Recently built a cap leakage tester. My question is, how muck leakage is acceptable?

[Jazz P Bass]

Leakage on a coupling capacitor?

Like on a tube amp?

I would think it really would depend on the following circuit but any leakage would be bad.

After all a coupling caps job IS to prevent Vdc from traveling across.

[g1]

What kind of units of leakage does the meter show?

[R.G.]

The point of a coupling cap is to prevent the plate voltage in a preceding stage from affecting the bias in a following stage.

The question is then how much is too much. For a 12AX7, the entire bias voltage is in the range of 1.5V or less, and you'd probably like to not change that more than 1% with leakage. For a 1M grid resistor, the resistance in the coupling cap then needs to be on the order of 100x that, or 100M ohm minimum. The bigger the better, and a gigaohm would be nice.

If it measures current, the pertinent number is the current that would cause 0.015 volts in a megohm, or about 15 nanoamps. Less is better.

1% is really just a guess. The right numbers are infinite resistance, unmeasurably big, and zero amps leakage.

[mozz]

A few manufacturers data sheets for their capacitors should show acceptable leakage limits.

[DRH1958]

The point of a coupling cap is to prevent the plate voltage in a preceding stage from affecting the bias in a following stage.
The question is then how much is too much. For a 12AX7, the entire bias voltage is in the range of 1.5V or less, and you'd probably like to not change that more than 1% with leakage.
1% is really just a guess. The right numbers are infinite resistance, unmeasurably big, and zero amps leakage.

I've always thought ~10-20mV leakage is fine or will not adversely affect the signal voltage appearing at the input of the next stage and it just so happens that 1% of 1.5 volts = 15mV, so 10-20 mV sets a nice range for using 15mV, RG's "guess" of 1%, as a guide. So if you are just hooking up the cap to a supply voltage typically applied to a 12AX7 gain stage and then measuring leakage volts to ground, then 15mV is a safe number. Applying this to RG's 1.5 volt bias example gives a range of 1.485-1.515V, hardly enough to affect things. Of course this is my personal thought but it seems safe looking at those numbers. Others may have different views and obviously lower numbers are best.

[Justin Thomas]

Depends on what tube is next in line... Chinese 12AX7? Leak away! One of my UOS STR 6L6GC from the 70s? "0.000V" is an acceptable leakage...

I like what Jack Darr said - with the cap disconnected on one end, if you see even 1V, toss it... I would refine that to say, "any measurable voltage that is more than what my probes can pick up out of thin air" is too much. I guess since I try to gig all my amps, I don't accept any compromise. If it leaks one volt today, can I predict it won't leak 10V tomorrow?

Justin

[diydidi]

Yes Tube amps.

Leakage on a coupling capacitor?

Like on a tube amp?

I would think it really would depend on the following circuit but any leakage would be bad.

After all a coupling caps job IS to prevent Vdc from traveling across.

[diydidi]

What kind of units of leakage does the meter show?

I've got it setup to measure uA

[R.G.]

I think that microamps is too coarse a measurement. Grid resistors are often 1M, and a micro-amp through 1M is one volt. One volt of change in bias is a disaster for a 12AX7. It may well cause runaway in an output tube.

Jack Darr's advice about measuring leakage on a volts scale was based on most meters in his day being 0-50uA movements. The moving-coil deflected full scale with 50uA through it. He was looking for currents so low that they would not perceptibly register on a 50uA meter. movement.

[mozz]

Most of the famed Sencore testers measure in microamps also.