Capacitor load tester idea

[Th3_uN1Qu3]

While i was testing the vent resistance of some Rulycons (fancy way of saying blowing the shit out of them), an idea came to my mind.

A 100kHz oscillator, a power amplifier, an AC amp meter and a source of DC.

• Apply nominal DC voltage to capacitor.
• Connect 100kHz oscillator to amplifier.
• Connect amplifier to electrolytic under test via high quality poly/film cap (obviously so that the DC voltage doesn't hurt the amplifier).
• Crank up the gain and watch the AC amp meter.
• Load it till it blows. For the more constructive, stick a termocouple on top of the vent and see how much current the cap can take before hitting its max rated temperature.

ESR meters are for kids. This will tell you the REAL impedance of the capacitor under REAL load conditions. PS. If on a normal size low voltage electrolytic you see more than 3A current and the cap hasn't started venting yet, RUN LIKE HELL. Don't ask me how i know that.

[Jack Crow]

Re: Capacitor load tester idea

S,
Sine wave + DC, is not usually how the parts work.
For the test to have meaning the parts should be tested as they are used.

Frequency and wave form are now important.

Then we need to think about physical environment.
If I had to guess about your test set up it's a pair of clips, the part under test and the whole package rests on the bench. A blast shield would not hurt.

Caps in the real world are mounted on boards, and you might want to think about your 'launch pad' options for realistic tests.

If memory is any good there are published tests for caps. Such a percentage of over voltage in a thermal run.

Then there are long term questions. Let's say a part survives your test. How can you trust it for later after it's been abused? So 100% testing is out.

Be well there.
Jack Crow in the less than free world.

[PCBONEZ]

Re: Capacitor load tester idea

Yeah I made up a schematic for that over a year ago only the objective was more as an endurance tester.
If I ever have time to build the thing it will also have a heater and temp control to keep them at 105C.
.

[Th3_uN1Qu3]

Re: Capacitor load tester idea

Frequency and wave form are now important.

The industry standard frequency for testing SMPS caps is 100kHz. That's what i'm going to do as well. Besides, nobody said i'm going to be using a sinewave, a conventional linear amplifier would be overkill for this application.

The "amplifier" will be a forward type SMPS fed from a regulated bench supply. It will run at fixed frequency and duty cycle, 100kHz, 50%. The only difference will be that there will be no diodes in the secondary of the transformer. Just a series capacitor, series inductor, and the capacitor under test. Ripple current will be varied by adjusting voltage going to the primary of the transformer. The series inductor will make the EXACT SAME waveform that we get in a real switching power supply, because it IS a real power supply, just that it doesn't output DC.

A blast shield would not hurt.

Already taken care of.

Caps in the real world are mounted on boards.

I fail to see how does this make any difference except in temperature rise.

Then there are long term questions. Let's say a part survives your test. How can you trust it for later after it's been abused? So 100% testing is out.

Well, we don't trust our caps for long term use at full rated ripple current or temperature, do we? IMO, whatever current the cap takes with a self-heating of less than 20C above ambient (or above whatever sits beside it), should be safe to use in a real power supply.

[Jack Crow]

Re: Capacitor load tester idea

S,
Fair enough.
Much luck on your project. Let us know how it works out.
Mike aka Jack Crow in Virginia USA

[Tube_Dude]

Re: Capacitor load tester idea

Use a 8 Ohms resistor is series with the feeding film capacitor, otherwise it will be a race between the electrolytic load capacitor and the amplifier, for to see who turn first into smoke.

Audio amplifiers are not big fans of output shorts...

[Th3_uN1Qu3]

Re: Capacitor load tester idea

Well the basic schematic is already done as i already had the building blocks. I have a spare 12-0-12 2A transformer which should be enough, so the main power supply will be a linear based on a LM317 with pass transistor for boosted current output, the PWM source will be a 555 wired for 50% duty cycle which will be fed from a 7812, the switcher used to torture the caps will be a two-transistor forward with IRF530s and 1N5819s, the inductor will be 2.2uH or 4.7uH, and will be a store bought part if i can find one with a suitable current rating (around 5A), and the voltage bias source will be another LM317 wired via a 10 ohm resistor (so it doesn't try to regulate the superimposed ripple voltage).

The only variables left are the transformer turns ratio and the circuitry that will be used to get readings on a voltmeter. I'd like to keep it with no opamps, this'll require some fiddling as the sense resistor would have to dissipate a fair bit of power to overcome the voltage drop of the rectifier diodes (remember, voltmeters don't really work at 100kHz so the output would need to be rectified to get an useful reading, and germanium diodes aren't available anymore).

@ Tube_Dude: I said above i won't be using a regular audio amplifier.

Edit: Seems like a sense resistor of 0.1 ohms is adequate to get an output voltage of 100mV per amp after rectification and filtering, which can be read on your average DMM. It'll also dissipate less than 3W at 5A. So, that's settled as well. Now time for a little PCB design. The project box will simply be an old ATX PSU housing.

[PCBONEZ]

Re: Capacitor load tester idea

Use a sig-gen at 100kHz with a saw-tooth[ish] wave to simulate the ripple.

My drawing is set up to abuse several banks of caps such that each bank can be a different brand/series.
Each bank receives the same abuse but a failure in any bank won't affect the other banks.
The objective is to compare reliability of different brands. [See who pops first.]
- A more controlled version of this idea.
https://www.badcaps.net/forum/showthread.php?t=2179
.

[Tube_Dude]

Re: Capacitor load tester idea

@ Tube_Dude: I said above i won't be using a regular audio amplifier.

Surely OK, but using the 8 Ohms humble resistor, anybody can use any "old amp" they have around, without the need to built it from scratch...

[Th3_uN1Qu3]

Re: Capacitor load tester idea

Yes, but not all amps will do 100kHz cleanly. Subwoofer amps, not at all. Besides, you have to agree that a 555 and a couple mosfets are a lot simpler and will take up a smaller footprint than an amplifier of any kind.

Making a sawtooth signal generator would be pointless because again it would require a linear amplifier. Making a high amplitude PWM wave and using an actual inductor to turn it into a sawtooth, just like a real power supply, is a lot easier.

[Tube_Dude]

Re: Capacitor load tester idea

Yes, but not all amps will do 100kHz cleanly. Subwoofer amps, not at all.

Usually a solid state subwoofer amp, have a high intrinsic bandwidth, is the inbuilt crossover filter that limit is range to the low frequencies.

About the need to do 100KHz cleanly;

- for to burn some caps I will not be worried about the linearity of the waveform...

Maybe is good for the self esteem of the caps, when they arrive to Cap Heaven (or Hell ), can say:

- I was shoot! But by a clean waveform!

But, of course you can use, or built, what you want and make you happy.

[PCBONEZ]

Re: Capacitor load tester idea

Yes, but not all amps will do 100kHz cleanly. Subwoofer amps, not at all. Besides, you have to agree that a 555 and a couple mosfets are a lot simpler and will take up a smaller footprint than an amplifier of any kind.

Making a sawtooth signal generator would be pointless because again it would require a linear amplifier. Making a high amplitude PWM wave and using an actual inductor to turn it into a sawtooth, just like a real power supply, is a lot easier.

Using a little baby amp would be pointless when putting the ripple to 60+ caps.
.

[Th3_uN1Qu3]

Re: Capacitor load tester idea

Usually a solid state subwoofer amp, have a high intrinsic bandwidth, is the inbuilt crossover filter that limit is range to the low frequencies.

Well, not really if it's a class D. But, the point is that maybe you don't want to change anything in that amp and just use it for what's been designed for. Besides, i really don't have a spare amp handy.