Based on the S32K guideline, when there is 1M feedback external resistor in the crystal oscillator peripheral , the high gain should be set. However, we set it with low gain which still works very well with a lot of MCU units and only very few handful of samples sometimes can't be waken up.
May we know the root causes and what is the best solution?
The defected Unit(PCB printing) worked at the beginning. MCU is triggered by a ignition off-on line for wake up and sleep.
after several hours, the unit didn't work somehow(we observed the crystal outputted waveform after the failure happened, it seemed
the MCU is stuck not to wake up due to internal parts).
The capacitors matching with MCU was confirmed with oscillator suppliers without issues.
It seemed it is not PCB printing problem rather than NXP itself internal issues for the defected unit.
if it is from a PCB print issue, how do we explain it worked at the beginning?
Changing the HW may trigger some issues on the HW so that we didn't do the SWAP testing on MCU or other peripheral devices(such as the feedback resistors).
Have you ever met some cases from NXP that MCU didn't work due to loss of clocks.
Any additional thoughts are appreciated.
You meant changing the 1M resistor to 10M resistor with low gain setting, right?
if so ,may we know the reasons?
In addition, How to handle other samples already delivered to customers testing without issues?
if testing for a long times, no issues, does that mean it is fine forever?
what is the root cause for the different behaviors in different MCU units with the same SW?
Shedding Any lights will be really appreciated.