Testing polarized CAPS with DSO (and signal gen?)

[mcinque]

Hello,

I guess that most of you will agree that for in-circuit electrolitic cap testing and low experience, an ESR value is an "idiot proof" method, but at the moment I don't own an (expensive) ESR meter.

I've read that ESR measurement involves injecting a low voltage (like 10mVpp) square wave at a low frequency (1/10KHz) on a lead of the cap to test, read the output in the other lead and finally do some math.

Hmmm... so, basically, with a function gen and a dso I should be able to collect some (raw) values as a waveforms!

As I guessed over the Internet there are people that reports simple methods to test in-circuit polarized caps, all of them involves frequency injection and probing (almost the same physical principle than the commercial ESR meters uses): inject a few KHz and low mV square waveform (some suggest to use a resistor first) in one lead of the cap to test, and then probe the other lead with a dso. The waveform should change in some way relatively to ESR value of the cap tested. More ESR, more changes. It is not an "monkey proof" numerical value but should do the trick.

So simple?: just take some good caps, put them one at time in between the fgen and dso and then learn their characteristic "reference" waveform. Then test a suspect cap and see any difference in the dso readings. Fantastic! Like a curve tracer!

But it doesn't work.  I've tried various caps of different values using different voltages and frequencies, injecting the signal with a shielded RG58 coax cable and a grabber connector at one lead and a 1X/10X probe with a hook on the other, adjusting properly time scale and sensitivity to have only one duty cycle on the screen, but the waveforms I get are identical, both with the cap in between the signal and dso and without it (with a metal jumper). No ringing, no under/overshoot, just the same squarewave.

There is something I'm doing wrong, that's for sure.

I'm done and I must buy an ESR meter?  Help me to save money

[Paul Price]

There is really no need for an ESR meter, at least in troubleshooting equipment.
I have been repairing electronic equipment for a zillion years without one.

If I suspect a capacitor has developed a high ESR or might be open I  look with an oscilloscope at the AC ripple or noise appearing across the capacitor and if it is high, I bridge a known good capacitor across the suspect capacitor and  see on the scope if the capacitor immediately if the cap under test is the problem.

The reactance of a capacitor at any given frequency is 1/(2 * pi *F *C).  If you put a capacitor in series with a load resistor of a ohm or so, abd then apply a sine wave from a function generator, any voltage drop in excess of the reactance is due to ESR. Again, bridging or comparing the "high ESR" cap with a known good cap of a close cap. value will verify any ESR problem and show the difference.

So, maybe what you really need is a function generator and an oscilloscope or else a small stock of various cap values.

Look at it this way, if you have found a capacitor with an high ESR, then you need to replace it, so just keep a small stock of spares to test caps and to replace.

[dfmischler]

I liked this approach, and tried it, and it works, but is not effective for me in circuit because of the need to control the voltage to be below one diode drop.  BTW - you don't need to use pulses like the author does, a square wave will work fine.  I also have a blue ESR meter, and that does work pretty well in circuit, but I have found that I rarely need it for power supply troubleshooting because my main tool for that is an old analog oscilloscope which can be very easily floated (I have to tell you not to try this yourself).

[mcinque]

Thank you dfmischler for the advice, I've already read this some time ago.

Paul, your method is interesting: I'd like to investigate it.