I was just playing with a new multimeter (KaiWeets HT118E) when I was curious about the voltage it uses when measuring capacitance.

I hooked up my scope to the DVM in capacitance mode and saw a triangle wav at around 22KHz, this is an AC signal, it crosses the 0volts line.

This means that for brief intervals the capacitor is being subject to the wrong polarity, when I put a 0.1 uF capacitor across the leads the frequency drops right down to some 50Hz or so, but the AC is till there, the capacitor is wrongly "biased" for about 50% of the time, the peak voltage of the triangle wave is just over 1v.

I know the voltage is small, but surely a well designed capacitance meter should be sensitive to the fact that one might be testing an electrolytic...

 

A standard electrolytic can tolerate 1-1.5V in the reverse direction without damage, so maybe that is what they assumed when designing the meter to measure capacitance.

 

My DMM has polarity markings on the capacitance inputs.
I will test with the oscilloscope and report back.

 

This is what I see, a digital unipolar signal, 0 to 3 V, 2.5 Hz.

 

Some DMM's use a charge/discharge profile to determine capacitance. Others use the test capacitor as part of an oscillator circuit and measure the resulting frequency. I don't have a feel for which is more common or how that has probably changed considerably over the years.

 

This is what I see, a digital unipolar signal, 0 to 3 V, 2.5 Hz.

View attachment 368866

That's kind of what I was expecting to see here. The AC triangle/sawtooth was also present on my more expensive bench DMM.

I think @crutschow is on the ball here, I was unaware that ecaps had that ability.

 

in this case this is not an issue since low voltage and short use. it takes time and energy to remove and optionally change polarization.

 

In my experience, depending on the capacitance the DMM will use a different measurement method. Larger capacitances (and thus more probable to be polarized) tend to have a discharge time measurement, while smaller ones will have an AC waveform.

I have seen this in a few DMMs and small component testers, but naturally I don't know how many actually have such different methods of measurement.

What I know is that my "real" LCR meters use zero crossing AC as the exciters for the DUT, regardless of the value.

Like others, I suspect the "small reverse polarity" might not be terribly detrimental to the aluminium capacitors, since I have seen some designs survive decades with slight momentary negative polarity applied to a capacitor's terminals.

 

It occurred to me that a neat feature for a capacitance meter would be one where the measured value is compared to standard values, so if I measure a cap that is actually 945 µ F it displays 1000 µ F perhaps with a tolerance as well, so in this it would display 1000 µ F and -10% in addition to the actual measured value.

One could perhaps do a similar thing with resistors, all we're doing is displaying a bit more info that might be helpful to a user.

Given the density available these days on small display panels this would be easy to accommodate.

 

Seems the Keysight ZA1 can do this.

 

It occurred to me that a neat feature for a capacitance meter would be one where the measured value is compared to standard values, so if I measure a cap that is actually 945 µ F it displays 1000 µ F perhaps with a tolerance as well, so in this it would display 1000 µ F and -10% in addition to the actual measured value.

One could perhaps do a similar thing with resistors, all we're doing is displaying a bit more info that might be helpful to a user.

Given the density available these days on small display panels this would be easy to accommodate.

That’s the classic Wheatstone Bridge measurement technique.

 

It occurred to me that a neat feature for a capacitance meter would be one where the measured value is compared to standard values, so if I measure a cap that is actually 945 µ F it displays 1000 µ F perhaps with a tolerance as well, so in this it would display 1000 µ F and -10% in addition to the actual measured value.

Why would I want it to display -10%?

If the meter is assuming (and that's what it is -- an assumption) that the nominal value is 1000 µF (note that there is no space between the scaling prefix and the unit), then a 945 µF cap is -5.5%.

Or are you saying that this meter is supposed to be able to determine what the tolerance level of the component is?

If so, how is it supposed to do that?

One could perhaps do a similar thing with resistors, all we're doing is displaying a bit more info that might be helpful to a user.

If a resistor measures 3160 Ω, what should be displayed?


"Sorting meters" have been around a long time, but unless that's what you are doing, they are generally more cumbersome than they are worth. You tell them what the bin boundaries are (or select which series you are working with).

 

Why would I want it to display -10%?

If the meter is assuming (and that's what it is -- an assumption) that the nominal value is 1000 µF (note that there is no space between the scaling prefix and the unit), then a 945 µF cap is -5.5%.

I copied the µ character from a website in haste, for whatever reason it rendered a space after the symbol.

Of course it's an assumption, if one doesn't care for it then one would pay no attention to it or disable the feature.

Correct me if I'm wrong, but there is no industry standard tolerance for capacitors of 5.5% it would clearly be > 5% so one would indicate that the measured value is 945 µF and that it's consistent with a nominal 1000 µF with a tolerance of -10%, this info offers an optional interpretation of the measured value.

This is example is real, I measured it earlier and it was a 1000 value with 10% tolerance.

In this case if the device were marked as 5% then we'd know it was suspicious/noteworthy.

Or are you saying that this meter is supposed to be able to determine what the tolerance level of the component is?

If so, how is it supposed to do that?

No I was not proposing that.

If a resistor measures 3160 Ω, what should be displayed?

3160 along with optionally enabled interpretation like it might be a 3.0K or 3.3K with some tolerance, I wasn't proposing a specific implementation or specification, it was just an idea.

"Sorting meters" have been around a long time, but unless that's what you are doing, they are generally more cumbersome than they are worth. You tell them what the bin boundaries are (or select which series you are working with).

I was unaware of them, thank you for enlightening me. I'd imagine students of electronics who are very new to these aspects of the subject, might find such a feature helpful.