Hi Andrea — that 1.9 mA on VBAT with VDD off is definitely high for an STM32H7 (RTC + LSE should be ~nA/µA range). A few likely causes and quick checks you can do:
First, confirm it’s not hardware: measure VBAT with everything else disconnected (remove the coin cell or the resistor and measure at the MCU pin). Look for any accidental conduction paths (diodes, solder bridges or other circuitry) that could be feeding current into other rails. Also check whether VDD or another rail is being back-fed through IO pins, sensors, or level-shifters — a common cause of unexpected VBAT drain.
Second, suspect the backup domain and LSE: if the LSE oscillator or RTC isn’t properly configured for lowest power, it can draw µA–mA. Temporarily remove or lift the LSE crystal (or unsolder/disable it) and see if the current falls — if it does, the crystal or its loading network or oscillator bias could be the offender. If you can run code before cutting VDD, disable the RTC/LSE or put backup domain into lowest-power mode and then power VDD down and re-measure.
Third, check VBAT domain settings and board design: ensure no external pull-ups, LEDs, or buffers are connected to VBAT domain pins that remain powered. Review option bytes and power configuration — some features or debug pins can keep circuits awake. Also inspect the schematic for any diodes or MOSFETs between VBAT and other rails which could create leakage.
Finally, consult datasheet/errata for your exact STM32H7 part — there are device-specific notes about backup SRAM, tamper pins, and VBAT behaviour. If you still see high current after the above, share the schematic (VBAT area, crystal, and any FETs/diodes) and a photo of the board/measured voltages and I can help pinpoint it.
