I am designing a portable (handheld, battery-operated) consumer electronic. The limited nature of the power source makes high efficiency critical at all loads, but will a lack of an OOA (out of audio) mode (preventing the converter switching frequency from entering the human audible frequency range) assumably result in audible annoyances to people near the device?

If it is relevant, my typical converter operating state is:

V.in  = 05.0 [V.usb ] or
V.in  = 12.6 [V.batt]

V.out = 02.8 [V]      or
V.out = 03.9 [V]

I.out = 00.7 [A.max]


With respect to dc/dc buck converters, TI offers several named modes for improving light-load efficiency. Among them is PSM (power save mode), PFM (pulse frequency modulation), ECO, and OOA (out of audio).

Generally, in these modes switching frequency is optimized in relation to load current. At low enough load currents, however, the switching frequency enters the human ear's most perceptible range of frequencies. In OOA mode, the switching frequency is clamped to a minimum before it reaches the most audible frequencies, at the tradeoff of operating at a frequency which is non-optimal for efficiency, depicted below from TI's SLUA946 : Understanding OOA™ Operation.

enter image description here

  • \$\begingroup\$ Depends on how susceptible your inductor is to coil whine (Lenz force) and capacitor to piezoelectric effects. Have you tried it with yours? \$\endgroup\$
    – winny
    Commented Dec 9, 2023 at 19:37
  • \$\begingroup\$ Still in concept — trying to determine if I missed something and need to veer completely away from the best efficiency converters I had originally downselected and instead start the downselect from scratch, instead looking solely at OOA converters. OOA is unfortunately not a filterable feature in TI's component searches, so I wondered if was all that common to begin with. \$\endgroup\$
    – kando
    Commented Dec 9, 2023 at 19:45
  • \$\begingroup\$ @winny Magnetostriction is the stronger effect, especially at assumed lower currents of a portable device; how much, depends on core material, with some ferrites being worse, which may make iron/composite type inductors more attractive. \$\endgroup\$ Commented Dec 9, 2023 at 19:46
  • \$\begingroup\$ @TimWilliams I didn’t know that. I have to read up on the subject. Thanks! \$\endgroup\$
    – winny
    Commented Dec 9, 2023 at 21:29

1 Answer 1


PFM mode doesn't necessarily create enough audible noise to be noticeable. The typical culprit is Class II dielectric ceramic capacitors, which exhibit piezoelectric behavior. When mounted to a board, the board can act as a loudspeaker and transmit the vibrations through the air.

Note that load transients in the audio band can do the same thing.

There are ceramic caps available that minimize the problem, and another trick is to split the capacitance in two and mount one cap on the top of the board and the other in exactly the same location on the bottom, so the mechanical forces on the PCB cancel out.

You could use polymer tantalum caps as well, which don't "sing" and have the added benefit of stable capacitance with DC bias.

Sometimes the inductor can create audible noise as well, but a good, shielded inductor usually is quiet.

Building and testing is usually the best way to be sure. If you pick a part with selectable OOO mode you can always enable OOO if you get into trouble.

  • \$\begingroup\$ "The typical culprit is Class II dielectric ceramic capacitors" Would the effect be reduced with an X7R (Class II) in comparison with a Z5U (Class II), or would the only safeguard be to jump to a C0G (Class I)? \$\endgroup\$
    – kando
    Commented Dec 9, 2023 at 20:02
  • \$\begingroup\$ There might be a small improvement for X7R over Z5U, but I've certainly seen audible noise problems with X7R or X5R caps. C0G of course will be fine, but it's often impractical for DC-DC bulk capacitance. \$\endgroup\$
    – John D
    Commented Dec 9, 2023 at 20:06
  • \$\begingroup\$ Type 2 are piezoelectric, so the effect is there regardless, but the amount can be reduced with design tricks as above, or using leaded types perhaps, and also simply keeping the ripple voltage small. C0G would be adequate, but large values are... expensive. \$\endgroup\$ Commented Dec 9, 2023 at 22:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.