Is there any way to test capacitors while on the circuit board?

[jharbuz5]

Is there any way to test capacitors while on the circuit board?  I know the voltage the multimeter puts across the capacitor can fry other components in the circuit if you try to test caps on board.  I also know if you test a cap on the circuit board you'll get a false reading since you'll be testing capacitance of the entire circuit.  However I came across this little device (called the CapaCheck 911) on eBay that claims it can accurately test capacitors on the circuit board (http://cgi.ebay.com/Capacheck-Plus-911xl-Capacitor-analyzer-tester-Board-/180633337515?pt=LH_DefaultDomain_0&hash=item2a0e95feab).  Here is what the ebay page for this device says:

"With CAPACheck capacitors are checked without disconnecting the circuit in which they work, even if the circuit is powered (AC mode only). Electronically protected, can be measured in the presence of voltage, up to 630 V DC.  CAPACheck generates a signal of only 25 mV RMS, so small that semiconductors are not activated. The reading is not affected by other electronic components like transistors, diodes, integrated circuits, etc.., Connected in parallel to the electrolytic capacitor under test."

So to you more experienced EE's I ask:

1) Is the claim of this device bogus or true?

2) Is there another way to test capacitors on board without having this device? 

Obviously de-soldering and re-soldering each cap on the board is kind of a pain so if there is a standard way to test capacitors on board I would love know know about it without having to purchase the device above.  First the CapaCheck is kind of expensive for something that only tests capacitors, 2nd it's manual is in Spanish (since it's made in Argentina), and 3rd I don't particularly care for analog meters in most cases (though they are handy to monitor rapidly changing values, like the outputs of audio amplifiers).  I would love it if there is a way to test capacitors on board with one of the many other multimeters I already own, or if there is a way to do it with an LCR Meter (which I've been thinking of purchasing).

Thanks in Advance

[allanw]

ESR meters will work for larger capacitances (1uF+). They only put a low voltage through so it shouldn't damage your other components, but check the specs. Of course, if this cap is in parallel with other ones, or if you really want to be safe, you could just lift one side of the capacitor and measure it that way. This is the same method you'd use to test a resistor in circuit.

Or you could do a quick DIY cap meter yourself with a square wave generator and scope. There's plenty of references online for that.

[jharbuz5]

Great info thanks Allan!

[Simon]

I hope the bogus claims are just down to bad english. You can't measure a cap while the circuit is powered, if anything your throwing in unknowns that the meter can't compenssate for.

[Psi]

You can test the ESR of a capacitor that's in-circuit but you can't test the capacity.

Well.. you could test capacity but you wouldn't get anything useful.

[jimmc]

Easy enough to build your own ESR meter, one of the simplest is http://members.shaw.ca/swstuff/esrmeter.html
If you want a (cheaper) digital meter how about http://clientes.netvisao.pt/greenpal/evb1.htm (but it won't stand 630v!)
Take a look at http://members.ozemail.com.au/~bobpar/esrmeter.htm for background information.

Jim

[jharbuz5]

Thanks all for the responses!  Especially on how to build an ESR meter.

[Zero999]

You can test the ESR of a capacitor that's in-circuit but you can't test the capacity.

No you can't because there will be other impedances in parallel with it so the reading will be meaningless.

Assuming one side of the ESR tested can be grounded without affecting the measurements, you could just disconnect one side of the capacitor and conduct the test whilst it's still half soldered to the PCB.

[ziq8tsi]

You can test the ESR of a capacitor that's in-circuit but you can't test the capacity.

No you can't because there will be other impedances in parallel with it so the reading will be meaningless.

Parallel impedances will only affect the ESR reading significantly if they are of a similar order of magnitude, or lower.  ESR meters that are intended for in-circuit use typically use a <200mV test signal so that they will not switch on semiconductor junctions.  Therefore you can measure the ESR adequately in circuit unless the capacitor literally has a <10? resistor across it, which would seem unusual.

[Zero999]

Unfortunately it's common practise to connect capacitors in parallel, for example on a linear PSU you might have a monster 4700uF capacitor in parallel with a 1uF tantalum and a 100nF ceramic capacitor, the latter both having a very low EAR which will interfere with the result.

[jimmc]

Unfortunately it's common practise to connect capacitors in parallel, for example on a linear PSU you might have a monster 4700uF capacitor in parallel with a 1uF tantalum and a 100nF ceramic capacitor, the latter both having a very low EAR which will interfere with the result.

Whilst I would agree that you cannot measure the individual ESR of similar value capacitors in parallel, if the combined ESR is low enough then it would be reasonable to assume that there is no fault.

For the example you gave, there is no problem in measuring the ESR of the 4700uF capacitor...
The reactance of a 1uF capacitor at 100kHz (typical frequency of ESR meter) is 1.6ohm and that of 100nF 16 ohm.
The expected ESR of the 4700uF is <100mOhm (eg http://parts.digikey.com/1/parts/40938-capacitor-electro-250v-4700uf-st-2222-115-13472.html) so the effect of the other caps is not significant (<10%).
Remember that ESR measurement does not have to be precise, the fail criteria for a capacitor is going to be at least a 50% increase in ESR.

Jim

[jharbuz5]

Great discussions everyone, I really appreciate all of the explanations!

[Psi]

You can test the ESR of a capacitor that's in-circuit but you can't test the capacity.

No you can't because there will be other impedances in parallel with it so the reading will be meaningless.

most ESR meters make the point of saying that they work for in-circuit testing.

Whilst I would agree that you cannot measure the individual ESR of similar value capacitors in parallel, if the combined ESR is low enough then it would be reasonable to assume that there is no fault.

Remember that ESR measurement does not have to be precise, the fail criteria for a capacitor is going to be at least a 50% increase in ESR.

exactly

[allanw]

If you have a failing 1000uF capacitor in parallel with a 0.1uF cap, would the ESR meter report high or low ESR?

[grenert]

You are not looking for an exact measure, just an evaluation of whether a cap is good or bad.  I agree that in any typical circuit, the confounding components will not change the ability to detect a bad cap.

[metalphreak]

Smart ?
The U1272A provides an additional 30 Ohm range for low resistance measurement. This Smart ? function is available
for ranges of 30 Ohm to 300 kOhm. It enhances measurement accuracy with offset compensation by removing residual
DC voltages of up to 1000 mV induced by ground current and thermal EMF. This function also enables ‘live’ resistance
measurement without isolating the measurement circuit. With this, you will be able to obtain leakage current using the
secondary display.

Does that mean you can do in circuit resistance measurement? Obviously it's not meant for capacitors though.

[allanw]

Does that mean you can do in circuit resistance measurement? Obviously it's not meant for capacitors though.

No, ESR measurements have to be done with AC. Multimeters assume the resistance is DC.

[Zero999]

For the example you gave, there is no problem in measuring the ESR of the 4700uF capacitor...
The reactance of a 1uF capacitor at 100kHz (typical frequency of ESR meter) is 1.6ohm and that of 100nF 16 ohm.
The expected ESR of the 4700uF is <100mOhm (eg http://parts.digikey.com/1/parts/40938-capacitor-electro-250v-4700uf-st-2222-115-13472.html) so the effect of the other caps is not significant (<10%).
Remember that ESR measurement does not have to be precise, the fail criteria for a capacitor is going to be at least a 50% increase in ESR.

You're right.

And if you've got two similar capacitors in series you can just disconnect one of them.