Measuring filter caps performance.

[Mark]

It’s not a question I thought I would be asking but here I am.

I use the ESR-70 to determine the condition of the filter caps. I have mentioned this online and been roundly told this tester won’t determine if the cap is leaky and I should simply replace the caps.

I tend to think if the cap was leaky the voltages around the circuit would be somewhat lower. I also don’t like replacing components on a hunch they might be faulty. To replace filter caps on a Marshall 2203 is quite expensive.

What do you guys think is the best way to comprehensively test filter caps?

[trobbins]

Measure the leakage at rated DCV. Measure the ESR. Measure the capacitance. Look for deformities. Fairly simple, but it takes time. It can be done by lifting one leg with care, and can be done using the amp's power supply when you have a variac and use a series resistor for current sensing.

[Mark]

Measure the leakage at rated DCV. Measure the ESR. Measure the capacitance. Look for deformities. Fairly simple, but it takes time. It can be done by lifting one leg with care, and can be done using the amp's power supply when you have a variac and use a series resistor for current sensing.

Right, so you put an ammeter in series with the cap. I assume if the amp is to be used to measure leakage, then the valves need to be removed from the amp.

What is an acceptable level of leakage for an 50uF + 50 uF cap?

[trobbins]

This is delving into hazardous testing, which you may not have done before, in which case imho you should work with a competent person to raise your experience.

The commonly used technique is to insert a 1kΩ resistor in series with the filter cap and clipping a handheld multimeter across the resistor – leakage current should subside to likely below 100uA (100mV) but this can vary with capacitance value, voltage rating, and capacitor age and maker, and experience is your friend to allowing sufficient time for leakage to subside, and to what level to accept as ok.

You have to manage charging a cap to a hazardous voltage, and then discharging it. Taking the capacitor out of the amp to a dedicated test setup removes various risks. If you aim to use the power supply in the amp then that requires a competent assessment of what needs to be disconnected, or not, including assessment of all parts in the signal circuitry chain and coupling caps. Removing valves may well be a part of the process, but also includes risk of over-voltage to circuitry, so an assessment is needed of circuitry and part voltage ratings.

Your ESR meter also has limits on residual voltage that may be on a cap when connecting to it. Many an ESR meter has been damaged by accident, and I've done that too - working to a method and simplifying leads and what is being tested can mitigate the chance of accidents.

[Roe]

This is delving into hazardous testing, which you may not have done before, in which case imho you should work with a competent person to raise your experience.

The commonly used technique is to insert a 1kΩ resistor in series with the filter cap and clipping a handheld multimeter across the resistor – leakage current should subside to likely below 100uA (100mV) but this can vary with capacitance value, voltage rating, and capacitor age and maker, and experience is your friend to allowing sufficient time for leakage to subside, and to what level to accept as ok.

You have to manage charging a cap to a hazardous voltage, and then discharging it. Taking the capacitor out of the amp to a dedicated test setup removes various risks. If you aim to use the power supply in the amp then that requires a competent assessment of what needs to be disconnected, or not, including assessment of all parts in the signal circuitry chain and coupling caps. Removing valves may well be a part of the process, but also includes risk of over-voltage to circuitry, so an assessment is needed of circuitry and part voltage ratings.

Your ESR meter also has limits on residual voltage that may be on a cap when connecting to it. Many an ESR meter has been damaged by accident, and I've done that too - working to a method and simplifying leads and what is being tested can mitigate the chance of accidents.

yes or slow form it with a 100k resistor and measure the voltage drop across the resistor. 10v is absolutely minimum typically. 5v is good, 2v excellent. But again "s can vary with capacitance value, voltage rating, and capacitor age" etc