Multimeter capacitance measurement tells you roughly how much charge a component stores for a given voltage, at one low test frequency, at almost no applied voltage. That is a single property. A capacitor that has failed in a switching power supply very often still reads within tolerance on that one property while being completely unfit for the job it was doing. Passing the dial test is not the same as being a good capacitor.
How the meter arrives at its number
Most handheld instruments measure capacitance by charging the part through a known resistance and timing how long it takes to reach a threshold, or by applying a low-frequency alternating stimulus and measuring the resulting current. Either approach gives a figure for the bulk capacitance. Both operate at a very low voltage and, crucially, at a low frequency far removed from where a filter capacitor in a supply actually works. They also assume the component behaves as an ideal capacitance with no significant series resistance, series inductance, or leakage. Real parts violate all three assumptions to some degree, and aged parts violate them badly. The instrument has no way to report a violation it never measured.
The failure mode a capacitance reading misses
Electrolytic capacitors in switching supplies usually fail by drying out. As the electrolyte is lost, the equivalent series resistance climbs, sometimes by an order of magnitude or more, while the measured capacitance falls much more slowly. In service the part now heats itself on every switching cycle, the output ripple rises, downstream regulators misbehave, and the supply may shut down or restart repeatedly. Put that same capacitor on a meter’s capacitance range and it can easily read within its printed tolerance, because the low-frequency bulk capacitance has barely moved. This is the single most common reason a technician replaces the wrong part or declares a board healthy when it is not. The same pattern shows up in inverters, motor drives, and older computer power supplies, where a bank of capacitors ages together and the symptoms appear gradually rather than all at once.
Other properties that stay invisible
Leakage current is the second omission. A capacitor with a partially degraded dielectric conducts a small steady current that will not appear on a capacitance range but will load a timing circuit, shift a bias point, or drain a battery-backed node over weeks. Voltage coefficient is the third: many ceramic types lose a substantial fraction of their capacitance when a working voltage is applied across them, which means the value measured at almost zero volts overstates what the circuit actually gets. Dielectric absorption, self-resonance above a certain frequency, and temperature coefficient all belong to the same category of properties the dial position was never designed to reveal.
What the range is genuinely good for
None of this makes the function useless. It identifies an unmarked part well enough to order a replacement. It catches a capacitor that has shorted or gone open, both of which are unambiguous. It confirms that a new part is roughly the value printed on it, which catches mislabeled stock and picking errors. It is also useful for checking motor run capacitors, where a substantial drop in bulk capacitance is the normal failure signature and the operating frequency is low enough that the meter’s assumptions are not badly violated. Used within those limits, it answers real questions quickly. It is also a reasonable way to verify that a large filter capacitor has been discharged before you go near the board, provided you trust the discharge method as the primary control and treat the reading as confirmation rather than as permission. And on timing circuits built around a resistor and capacitor, comparing the measured value against the design value quickly explains an interval that has drifted long or short.
Measuring in a way that means something
For anything where series resistance matters, use an instrument that measures it. A dedicated meter for equivalent series resistance, or an LCR bridge that lets you choose the test frequency, gives a number that relates to how the part behaves in circuit. Test at a frequency near the part’s working frequency where the instrument allows it. Discharge every capacitor fully before measuring, both for accuracy and because a charged part can damage the meter or injure you. Remove the capacitor from the circuit, or at least lift one leg, unless you are using an instrument designed for in-circuit work, because parallel paths on the board will corrupt the reading.
What to check before you act
- Discharge the capacitor completely before connecting any instrument.
- Lift one leg or remove the part unless the tool handles in-circuit work.
- Measure series resistance separately before declaring a part healthy.
- Compare ceramic readings against the voltage the circuit actually applies.
Common questions
My capacitor reads the correct value but the supply still misbehaves. Why?
Bulk capacitance is only one property. Elevated series resistance is the usual culprit and requires a different instrument to detect.
Can I measure capacitance without removing the part?
Rarely with a general-purpose meter. Anything in parallel on the board adds to or distorts the reading, so a low result in circuit proves nothing.
The practical takeaway
Treat the capacitance range as an identification tool, not a health test. Discharge the part, take it out of circuit, and measure series resistance as well before you decide whether a capacitor is doing its job.
References and further reading
- CPSC: electronics and electrical safety guidance for the home
- Department of Energy: power supplies and electronic equipment efficiency
- CPSC: general electrical safety information
Numerical scenarios are illustrative unless identified otherwise. Follow the exact product instructions; component ratings and local installation requirements can differ.
