I think I must have missed something. The trouble is the on-chip DC-DC does not always appear to be on. Sometimes it's on, some times not. Even the workload on the power supply would make a difference. I'm afraid I haven't been able to find out the rule.
When the on chip DC is not on (3.3V power is always provided), I've tried manually apply a reset to the chip, but nothing would happen. The NVCC_PLL is 1.1V so some of the on-chip power is working. Also, when this DC is not on, the debug interface is off line.
The circuit has been attached. The PWR_CTRL is an output to control the external power supply, when it is high the power keeps on, and the PWR_SW is an input connected to a power push button.
Is there anything can fix this? I would even think to provide an external 1.2V as the bottom line solution.
Many thanks in advance for your inputs.
PS the NRST is actually left unconnected or to the debug interface
PPS the 32kHz OSC pins are floating i.e. not used; a 24MHz (1.1V) clock is provided externally.