Battery with esp32 s3 supermini

[lkos98]

It's just to let people know of a design flaw of esp32 s3 supermini with battery pins.
Symptoms: When a battery and USB are both connected and code is uploaded, S3 Supermini stops working on battery only. 3.3V pin shows 2.3-2.4 volts.
Reason: There is a Schottky diode connected between 5V and onboard 3.3V regulator, to prevent a short connection between 5V USB supply and battery. When code is uploaded this diode burns out, so the battery disconnects from onboard 3.3V regulator and no longer supplies the chip. With USB power everything else works fine - even charging of the battery. I don't know why this diode burns, since it does not have any connection with usb data, probably some spikes or overloading (it's a 200mA Schottky).
Solution: Replace the diode with a higher current or solder a new one between 5V pin and Battery+ pin. Disconnect battery from batt Pins (jumper or switch) when uploading. Replacing the diode is extremely difficult since it's very close to USB connector and 5V/GND/3.3V pins and is SMD0806.

[Sprite]

Seems archive.org has a capture of a schematic here. If that is correct, the board is a bit janky... it means that there's only that diode between the battery and Vbus. This means that if you plug something unpowered into the USB-C connector, the battery will try to backfeed that device via the diode. If whatever you plug in for some reason connects Vbus to ground, then that diode will definitely get smoked, leading to the issues you saw.

[lkos98]

Great, thanks for the schematic. I could not find one. You are right about load on Vcc, but why would uploading a code short the VCC to GND? And I can tell for sure it happens only when uploading - I have burnt 3 of those the same way. Pity - it's a very nice board, cheap and small and battery powered with charging, without additional modules.

[Sprite]

Hard to say - could be because you're using USB-C and that needs some negotiation before it outputs 5V, which happens using the 5K1 resistors on the CC pins. Could be that before that, Vbus is grounded, or Vbus is disconnected but there is a fairly large capacitance on it. That would lead to an inrush current that potentially is large enough to kill the diode.

Fix would be to put a second diode in the Vbus line, so there can't be any current flowing back up the USB port, or generally better power muxing using a P-chan mosfet. Not sure why they didn't do that in the first place.