Hi,
When the firmware is run in debug mode (J-Link connected), the system always works correctly. BLE initializes properly, and all sensors function as expected.
When the device is powered from the battery and undergoes a full power cycle (complete power cut or battery disconnect/reconnect) without a debugger attached, the system does not function correctly. BLE and/or sensors fail to initialize properly.
Repeating the same power-cycle process with the debugger attached immediately resolves the issue.
This indicates that the problem occurs only during a cold boot in release mode, and the presence of the debugger masks the issue.
From our analysis, this behavior suggests a power-up, reset, clock, or initialization timing issue, rather than a functional issue in the BLE stack or sensor drivers themselves.
Key observations:
BLE and sensors work correctly when tested individually.
The failure occurs only when both are initialized together after a cold power-on.
Debug mode likely introduces additional delays, stabilizes the power rail, or alters reset/clock startup behavior, allowing the system to initialize correctly.
During a cold boot without the debugger, the application may start executing before:
The power rail is fully stable
HFCLK/LFCLK is ready
External sensors have completed their power-up sequence
We suspect this may be related to:
prj.conf configuration (clock control, power management, DC/DC, logging)
Sensor power sequencing or missing startup delays
Reset or brown-out behavior during battery reconnect
Initialization order or timing differences between debug and release builds
We would appreciate your guidance on:
Recommended prj.conf settings for reliable cold-boot behavior
Best practices for clock and power initialization in battery-powered systems
Any known issues in nRF Connect SDK v3.1.0 related to cold boot vs debug behavior
Please let us know if you need additional information, logs, or a reduced test case.
Best regards,
Kinjal Sarda
