ESP32 stops working for no apparent reason in field

[ESP_Angus]

Hi Kalyan,

Can you please post the output of "espefuse.py -p PORT summary" on some of the failed devices (that still respond over serial, at least).

Angus

[javqui]

Hi,
I'm navigating similar waters. Just started the troubleshooting and 'reasons research'. I have 5 faulty units from 12 preliminary prototypes that are having similar issues as described here. I will post more information as soon I get it.
Got all WROVER-16MB (with a mix of PCB and external antenna configurations) from a reputable vendor (mouser).

There are 2 processor on each product. ESP32 is working together with a STM32. The STM32 is a 'coprocessor' for low level GPIO protocols expansion and the ESP32 is the main actor. STM32 MCUs are working without issues, so probably the external power supply is not the source. (power supply is an isolated DIN rail AC/DC 12V, then an embedded modular power switch move down the 12V to 3.3v.) Will focus on capacitors and other issues.

[ragoth]

Hi, I also had the same failure. I was using it with my DELL laptop to program and it failed to upload with error messages even for a simple sketch like "BLINK". When I touched the ESP32 it was overheating and not able to touch it. When I checked the current, it draws 280mA steadily. I assume it has failed due some manufacturing defect and it has reliability issue.

Espressif should investigate it before it goes to big level.

[ESP_Sprite]

I highly doubt the chip itself has a manufacturing defect like that - with the large amounts we ship to various customers, I'm sure we'd have had lots of complaints by now. What devboard are you using?

[javqui]

Hi,
I'm navigating similar waters. Just started the troubleshooting and 'reasons research'. I have 5 faulty units from 12 preliminary prototypes that are having similar issues as described here. I will post more information as soon I get it.
Got all WROVER-16MB (with a mix of PCB and external antenna configurations) from a reputable vendor (mouser).

There are 2 processor on each product. ESP32 is working together with a STM32. The STM32 is a 'coprocessor' for low level GPIO protocols expansion and the ESP32 is the main actor. STM32 MCUs are working without issues, so probably the external power supply is not the source. (power supply is an isolated DIN rail AC/DC 12V, then an embedded modular power switch move down the 12V to 3.3v.) Will focus on capacitors and other issues.

Just want to add our preliminary findings. Looks like the main reason for this high count faulty units is incorrect technician board manipulation and some exposed pins (not a real problem with the ESP32). These units have been shipped to a location where technicians are doing connections to these boards without precautions and protections. We will improve technician training and put some boards on a side locally for testing to get final results of the source of the issue.

[lbdroidman]

Just want to add our preliminary findings. Looks like the main reason for this high count faulty units is incorrect technician board manipulation and some exposed pins (not a real problem with the ESP32). These units have been shipped to a location where technicians are doing connections to these boards without precautions and protections. We will improve technician training and put some boards on a side locally for testing to get final results of the source of the issue.

Good find.
In addition, it would also be a good idea to put at least a bigger capacitor on the input to the LM2596. Its datasheet calls for 470 uF. Additionally, if there is *ANY* chance AT ALL of input voltage spikes of over 45 volts (its absolute maximum rating), then you definitely should put a TVS diode on there. When LM2596 experiences over-voltage failures, they dump input directly to output. You can see this thread for reference; https://e2e.ti.com/support/power-manage ... tch=LM2596