SAADC returns persistent erroneous values after short deep sleep on nRF54L (Differential + Oversampling)

[NandanS]

Hi all,

I'm working with the nRF54L and encountering persistent SAADC issues after short deep sleep cycles.

Setup:

• SAADC is configured in differential mode with oversampling enabled. It is configured for very high sampling rate (in the kHz range).
• Before entering deep sleep, I call nrfx_saadc_abort() followed by nrfx_saadc_uninit() to halt conversions and cleanly shut down the peripheral.
• After wake-up, I reinitialize the SAADC and begin sampling.
• If the deep sleep duration is short (e.g., <2s), the SAADC returns consistently erroneous values after wake—these do not self-correct over time.
• If the deep sleep is long (e.g., several seconds or more), the SAADC readings return to normal immediately after wake.

Notes:

• GPIOs and reference voltages are stable.
• No custom calibration routines are in place yet.
• Using internal reference, differential mode, and oversampling via NRFX_SAADC_CONFIG_OVERSAMPLE.
• The issue persists across multiple short sleep cycles and appears duration-dependent.

Questions:

• Is there a known issue with SAADC state retention or reference settling when using differential mode + oversampling on nRF54L?
• Could this be tied to power domain ramp-up, reference voltage stabilization, or peripheral clock synchronization?
• Has anyone implemented a reliable workaround—e.g., dummy reads, forced calibration, or delay injection—after short sleep wake?

Any insights or shared experiences would be greatly appreciated. I can share code snippets or timing diagrams if helpful.

Thanks!

[Hieu]

Hello NandanS,

Do you have a minimal issue reproduction project you can share? It would help me speed up reproducing the issue and investigating it.

Hieu

[NandanS]

Hi,

The root cause was due to improper external peripheral setup and teardown, specific to my setup. So it is unrelated to the SAADC on the NRF54L.

This ticket can be closed.

Thanks