High Leakage From VBAT to VBUS #3
Description
This issue is spawned from the Sparkfun forum post here.
I've noticed the Thing Plus USB-C has an issue with high leakage from VBATT to VBUS when only the battery is connected. From testing I've done on my own board, the issue appears to be that D4 has extremely poor reverse leakage (along the lines of the Infineon BAT60A) which allows a non-negligible amount of current to flow through R7 when on battery.
Schematic
On my board, this results in VBUS being pulled to 2.5V when on battery power when it really should be ~0V. Given the 10K pulldown on this rail, it seems about 250uA is reverse leaking through the diode. The ESP32's deep sleep current is specced at ~10uA, so this leakage represents an obstacle to battery-powered projects.
This issue also prevents effectively recognizing when power is coming from USB vs when it's coming from battery b/c the rail never drops low enough to be registered as a digital LOW by the ESP32 (.25v). In my application this prevents me from setting the ESP to wake from deep sleep when USB power is supplied. I've mitigated by adding a 500ohm resistor to pull the rail low harder but I really wish this wasn't a problem in the first place.
Whatever diode is in the BOM really ought to be replaced with one that has a better reverse leakage spec. It looks like there's probably enough space on the PCB to even include one with a larger package which should allow the existing spec to be met without the reverse leakage problem.