CM5 speaker static at low battery: two workarounds
Below roughly 15 % battery the CM5's speaker picks up boost-converter noise; beams' power_save=0 modprobe option or Rex's audio-disable package quiet it.
What it takes
- Config change
Difficulty 1 of 3 — easy.
Details
The buck/boost converter works harder as the cells sag and the noise ends up in the PWM audio path (Alien's explanation in Rex's thread). beams' workaround disables the sound card's power-save: options snd_usb_audio power_save=0 in /etc/modprobe.d/snd-usb-audio.conf, or options snd_bcm2835 power_save=0 for the built-in audio on Raspberry Pi OS Trixie, then reboot. It helped beams; two replies say it did nothing for them.
Rex's clockworkpi-audio-disable package (sudo apt install clockworkpi-audio-disable, then audio-disable from the System Tools menu) is blunter: it stops the headphone-detect service and routes audio to the jack so the speaker stays quiet until the next reboot. Debian-based images only. Charging the battery is the actual fix.
Links
Open at Forum threadRelated mods
A CM5 needs the RP1 audio remapped to GPIO 12/13 (dtoverlay=audremap,pins_12_13) and the amplifier enable pin driven high; current images already do both.
On a CM4 the speaker amplifier is switched on by the devterm-audio-patch package; if it goes quiet, reinstall the package or drive GPIO 523 high from rc.local.
A USB DAC sidesteps the mainboard's mono AW8110 amplifier and the core module's own codec or PWM output; one owner reports it resampling to 44.1 kHz/16-bit.
Koha's proposed ES9039Q2M DAC card on the expansion slot's I²S: 3.5 mm and 4.4 mm balanced headphone outputs, line-in, OPA1612 amps, about $75 — still a design.
bored99's one-piece replacement back leaves the CM5 heatsink open to the air and covers the cells with a magnet-held plastic lid for quick swaps.
UJW's deeper printed back fits two 10,000 mAh 146074 pouches and a Waveshare CM5 active cooler over the HackerGadgets boards — designed in CAD, not yet built.