Explainer

Ceramic capacitors can make a circuit board sing

Solder joints can transmit the ceramic’s movement into the board, where a bending mode can make an otherwise quiet capacitor audible.
A macro photograph of multilayer ceramic capacitors soldered to a printed circuit board.

A converter running in burst mode may switch well above the audible range during each burst, yet still produce an audible tone at light load. The output capacitor charges during the burst and discharges between bursts, so the voltage across it varies at the slower burst rate as well as at the switching frequency. If that slower variation drives the board near a mechanical resonance, the board can radiate sound.

High-permittivity ceramics such as the barium titanate formulations commonly used in X5R and X7R multilayer ceramic capacitors change dimensions slightly as the electric field varies. The solder joints transmit that small movement to the laminate. A board can then bend over a much larger area than the footprint of the part, which is why the loudest point may be some distance from the capacitor. Change the board thickness, the mounting points or the enclosure and the same capacitor can sound different.

Moving the capacitor to a stiffer part of the board may reduce the sound, but moving it away from the current loop it bypasses can raise the loop inductance. An interposer or compliant metal terminals reduce the transfer of movement without necessarily moving the part. A polymer capacitor changes the electrical design too: its impedance and effective capacitance need to suit the converter’s control loop.

Murata tested a laptop battery rail feeding several DC-DC converters. Capacitors before the rail branches see much the same voltage variation, and Murata measured a progressive fall in sound pressure as it substituted quieter parts across successive groups. Both the standard and the low acoustic noise parts in the test were nominally 10 µF. Swapping one group can leave the other capacitors driving the same board, which can explain why replacing a single part makes less difference than expected.

At the bench, compare the peak in the microphone spectrum with the burst rate on the oscilloscope as the load changes. If they move together, the converter is probably driving the tone, though an inductor excited at the same frequency is still a suspect. If the tone changes with the enclosure fitted while the burst rate stays put, the mounting or enclosure is affecting the sound. Murata says the sound alone does not indicate that the capacitor is degrading.

Sources: Murata’s capacitor acoustic-noise FAQ; Texas Instruments’ application article on acoustic noise in power circuits; Murata’s laptop battery rail test.

Continue reading