testing caps for DC leakage?

[daz]

I assume it's not as easy as simply putting the ground probe to ground then the hot probe to the opposite side from the DC source? Or maybe it is? Or do you have to lift the non DC side of the cap and do what i just described? Also, what id the negative effect of leakages? Tone? Noise?

[Mick Bailey]

I'm assuming tube amps here; Testing a cap for leakage in-circuit is not a good or reliable method due to the effects of the connected circuit. For example, let's say you're looking at an inter-stage coupling cap and getting a small DC voltage on the side connected to the grid of the following stage. You won't know if this is a charge on the grid, or capacitor leakage (or both). If the voltage is high then it probably is DC leakage, though.

Lifting one end of the cap is better, but if the cap is discharged to begin with it has to charge via your DMM's internal resistance (on the voltage scale - usually 10M ohm). Remember, DC leakage is a current measurement, so the leakage you're seeing as a voltage is being measured across the meter's own internal resistance. You have to calculate the current from this. The cap can take a long time to charge using this method. OK with low-value caps, but tedious with larger film or electrolytics.

It's better to use a simple cap leakage tester. I use a cheap Chinese $4 buck/boost converter to generate HT up to 450v and a switch that can rapidly charge the cap via a resistor, then switches to disconnect the + side and connects my DMM between + and ground. I still have to lift one end of the cap, though. And remember to discharge the cap afterwards (it really needs a discharge switch).

DC leakage in tube amp caps is a serious problem in coupling caps - it shifts the bias of a following stage and can cause distortion. Leakage is often accompanied by other issues with the capacitor - noise, change in ESR etc. Leakage in PSU caps causes hum and can stress the transformer.

[daz]

Thanks. So how would i check tone stack and coupling caps? Lift one end, connect the probes to each side, (or black to ground and red to A/C end of cap?) turn on amp and wait till any reading or a zero reading stabilizes?

[Mick Bailey]

Lift one end (the lower voltage end) and connect your probe between the floating cap lead and ground then wait for the voltage to stabilize. If you know the DMMs resistance on the voltage setting, then divide the reading by that to get the actual leakage current. If you don't know the meter's resistance, a more predictable measurement is to clip-lead a 1M resistor between the free end of the cap and ground and measure the voltage across it to calculate the leakage current. My Fluke is 10M ohm, but I'm looking for as close to zero volts as possible and don't usually bother with any calculations. As an example, my '61 Ampeg Reverberocket had leaking Astron caps all over and the coupling caps measured 20-odd volts with just my DMM with one end lifted. I don't need to calculate anything to know they're bad! The replacement film caps had a calculated leakage of just a couple of microamps.

[daz]

I'll try the 1M resistor method. Thanks. I'll post back with any questions and what my results are.

[Tom Phillips]

After you have tested several caps using the method Mick described in post #4 you will become familiar with the trend of the results with your meter and the charging curve for various capacitance values. I find that I don't need to add the 1MΩ resistor to get viable results. You will notice that lower value coupling caps, such as 0.01µF, charge much faster than a 0.1µF. This assumes that the cap was discharged when you started the test. If you run into an amp with old yellow Astron coupling caps then you will find many leaky caps.
Cheers,
Tom

[daz]

Ok, so i lifted the pot end of the bass cap in the tone stack, soldered a 1m to it and the other end of the 1m to ground and attached the probes across the 1m resisitor. After 5 minutes it still fluxuates between around 5 mv constantly moving. 1- what does it mean....is it supposed to stabilize and if yes why won't it? Does it take a lot longer for a .02uf cap to charge? 2-assuming i can assume that it's a round a couple mv even tho it doesn't stabilize but never goes out by more then a few mv, is that my result or do i have to tak say 2mv and calculate it somehow?

[Tom Phillips]

When trying to read such a low DC voltage in the typical bench environment with AC fields radiating from multiple sources, the reading will never stabilize. Since the meter impedance (most likely 1MΩ) is in parallel with the 1MΩ resistor that you added you are actually measuring 5mV across 500kΩ. Therefore, per Ohms law, the leakage current would be 0.005V/500,000Ω = 0.01µA. I'd say that indicates that the cap is good.

[daz]

Thanks. Do you ever come across orange drops or mallory 150's that leak when new?

[Tom Phillips]

Thanks. Do you ever come across orange drops or mallory 150's that leak when new?

Not yet. But no line of parts is100% perfect.

[daz]

Thats good to hear. I read a lot about people getting bad micas from new though. But for larger caps, coupling and tone stack duties i use ODs and mallory and being the ones in my amp now have been used and abused a lot so i'll likely just replace them all to be safe, especially the mallories with this stupid hair thin leads. (take a lesson from sozo, mallory !)

[Helmholtz]

I'm not aware of any aging effects with mallory poly caps.

[daz]

I'm not aware of any aging effects with mallory poly caps.

What i meant by used and abused is they have been soldered in, taken out over and over some for probably 50 times and the leads are all weakened and the bodies are scarred from the iron hitting them, etc etc. They're just not reliable and likely some are leaky. Tomorrow, saturday i will be using it for the first time in a band scenario since i dusted it off and re designed it. So if i feel it sounds great i will replace them all. If not i may just go back to using my modeler or consider getting a fender supersonic 22.

[Mick Bailey]

Since the meter impedance (most likely 1MΩ) is in parallel with the 1MΩ resistor that you added you are actually measuring 5mV across 500kΩ.

1MΩ is low for a DMM on DC volts and would give rise to false measurements especially across resistors due to the parallel resistance of the meter. My Fluke as well as a bunch of cheap Chinese throwaway meters all have an input impedance of 10MΩ across all DC ranges. It's worth checking, though. My Fluke also has a 'LowZ' function to eliminate ghost voltage readings, though that is 10KΩ

[Tom Phillips]

1MΩ is low for a DMM on DC volts...My Fluke as well as a bunch of cheap Chinese throwaway meters all have an input impedance of 10MΩ...

Right you are. I had my thoughts mixed up with the AC input impedance.