I'm having a problem in a 1 gram tracker that uses an stm32wle5 powered by a 2.4V LTO battery and 4mW solar cell (through bq25504 energy harvesting chip). If the battery discharges below 1.7V ("should never happen"...) the solar cell cannot recharge it above ~1.75V because around that point the stm32's SMPS (Vlxsmps) starts oscillating at about 4kHz dumping charge into the attached inductor and cap. The solar cell is not powerful enough to overcome this.
If I just power the stm32wle5 directly from an adjustable power supply I can see the same phenomenon, which stops when Vdd rises to about 1.78V, at which point Vlxsmps becomes stable (I have the power configured to use LDO). Is this a known "feature" and is there a workaround?
I realize this is below the voltage at which operation is specified but I believe the chip is intended to allow slow power ramps gracefully. During the whole process, reset is low, and I have the BOR level set to 1, i.e. 2.0-2.1V. As far as I can tell, this is not an overall chip initialization problem but a localized issue around the SMPS.
I know that the proper fix is to produce a V-good signal from the bq25504 and then drive a high-side switch, but I have neither the weight nor space budget for all this, especially not for a state that "should never happen" (which I know means that it will happen when most inconvenient...).
This scope shot shows the issue when I power the stm32wle5 from a bench power supply. In purple Vdd, in cyan Vlxsmps.
