'Spurious watchdog reset at power up'

[FrancisL]

I am working with a ESP32-D0WD device (small package) and I get some time to time some problems at power up. If I run the 'serial monitor', I see in these cases that I have a double reset :
rst:0x1 (POWERON_RESET),boot:0x3b (SPI_FAST_FLASH_BOOT)
ets Jun 8 2016 00:22:57

rst:0x10 (RTCWDT_RTC_RESET),boot:0x2f (UART_BOOT(UART0))
and my application is frozen, waiting for some input at the UART0... Most of times, my application stats properly, but not always.

I am working with a 1.8V FLASH and I burned the efuse bits to force the regulator voltage at 1.8V (checked on the Vcc pin of the FLASH).
This issue looks like the first problem described in the errata list: https://www.espressif.com/sites/default ... p32_en.pdf : spurious reset at power up...
But does this 'old issue' also apply to the new D0WD (october 2017) ? Is the only workaround consists in writing a program for the ULP ?
I am also wondering if the timing at the 'CHIP_PU' is important. I have on my board RC=1k x 100nF. Is it a correct value ?
When I check in the EFUSE summary, I find:
CHIP_VER_REV1 Silicon Revision 1 = 0 R/W (0x0)
CHIP_VERSION Reserved for future chip versions = 0 R/W (0x0)
CHIP_PACKAGE Chip package identifier = 1 R/W (0x1)

Does it mean that my device is still a revision 0 ?
Thanks for your help ?
FrancisL

[ESP_Angus]

Hi Francis,

Yes, your chip is still revision 0 ("CHIP_VER_REV1 Silicon Revision 1 = 0 R/W (0x0)"). Recent versions of esptool will also print the chip revision whenever they connect to the device (to flash it or for some other reason).

Usually the "spurious watchdog reset" only resets the chip, it doesn't change the boot mode. There is an explanation of the boot mode messages here:
https://github.com/espressif/esptool/wi ... de-message

It looks like GPIO0 has gone from high (normal boot) on power on to low (flashing mode) between power on and the WDT reset, and GPIO2 has also gone high which has put into a non-standard boot flashing mode. Suggest checking anything which is connected to a strapping pin (strapping pins are listed in the ESP32 datasheet and also at the wiki page linked above), and making sure they don't pull any of the mentioned pins high/low.

[FrancisL]

Hi ESP_Angus,
You are right: I also use GPIO0 as the ethernet clock input. It is easy on my side to delay the startup of the oscillator, therefore to fix the issue.
Thank you very much for your help.
Francis