ESP32 WROVERS BOOTLOOPING

[StudentSA]

Good Day,

I have a product that uses the ESP32 module : ESP32-WROVER-E-N16R8

We have this on a simple custom PCB powered through LDL1117S33R 3.3V regulator.

Lately we have been received units that are failing to boot. Im trying to understand what could have happened to these devices to cause this issue and what can we do to prevent its reoccurrence???

When checking the boot failures I see the following on serial prompt:

Code: Select all

rst:0x1 (POWERON_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0xee
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:2
load:0x3fff0018,len:4

load:0xff001cff,len:267327
1150 mmu set 00010000, pos 00010000
1150 mmu set 00020000, pos 00020000
1150 mmu set 00030000, pos 00030000
1150 mmu set 00040000, pos 00040000
ho 0 tail 11 room 5
load:0x202b2029,len:942897200
1150 mmu set 00050000, pos 00050000
1150 mmu set 00060000, pos 00060000
1150 mmu set 00070000, pos 00070000
1150 mmu set 00080000, pos 00080000
1150 mmu set 00090000, pos 00090000
1150 mmu set 000a0000, pos

rst:0x10 (RTCWDT_RTC_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0xee
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:2
load:0x3fff0018,len:4
load:0xff001cff,len:267327
1150 mmu set 00010000, pos 00010000
1150 mmu set 00020000, pos 00020000
1150 mmu set 00030000, pos 00030000
1150 mmu set 00040000, pos 00040000
ho 0 tail 11 room 5
load:0x202b2029,len:942897200
1150 mmu set 00050000, pos 00050000
1150 mmu set 00060000, pos 00060000
1150 mmu set 00070000, pos 00070000
1150 mmu set 00080000, pos 00080000
1150 mmu set 00090000, pos 00090000
1150 mmu set 000a0000, pos 000a0000

the loop continues forever with RTCWDT_RTC_RESET

Any help in debug is appreciated.

Thanks,

[StudentSA]

Turns out that there is an issue with some flashes packed with the ESP that sometimes bricks itself on power outage.

https://github.com/espressif/esp-idf/issues/7994

This is really concerning, according to the issue thread, it seems they have a workaround that reduces the chances of occurrence but still it could be a possibility.