ESP32-WROVER Chips from new reel with flash chip failure

[rkager]

We have for more than 3 years a project running and shipped more than 1000 devices. On a new production lot we took chips from a brand-new reel of ESP32-WROVER-E-N8R8. On all of this SOMs we have an issue with the flash. The ESP32 chip communication is fine but the esptool.py flash_id returns all FF and we cannot flash the device as it delivers: Warning: Could not auto-detect Flash size (FlashID=0xffffff, SizeID=0xff), defaulting to 4MB and then fails with a fatal error.

Thanks

[ESP_Sprite]

GPIO12 is a bootstrap pin for the voltage for the internal flash and PSRAM. At the moment, you have it tied high (via RA7B) which only gives the chips 1.8V. (Possibly you used to use different WROVERs before which used 1.8V, making this work.) You could either rework the board, or burn the eFuse to force 3.3V independent of that GPIO (see espefuse.py for more info)

[rkager]

Thanks, I have seen this in the manual, but how do I know which flash type is connected (1,8V or 3,3V) as it is a wrover SOM. Thought the correct voltage is set via efuse from espressif?!

[ESP_Sprite]

The datasheet would tell you that: "The internal pull-up resistor (R9) for MTDI is not populated in the module, as the flash and SRAM in the module only support a power voltage of 3.3 V (output by VDD_SDIO)."

No, we don't burn that fuse by default, sorry.

[rkager]

ok, so it is working to set the voltage to 3.3V via eFuse. That is good, as we have already 400pcs mounted.

So is it correct that all ESP32­WROVER­E has 3.3V and all ESP32­WROVER­ has 1.8V? What if excitedly write the 3.3V eFuse on an old 1.8V ESP32­WROVER­?

Thanks for the support

[ESP_Sprite]

I think there's a version of the ESP32­WROVER­ that has 3.3V flash/psram, but it's indicated by the model number (iirc it has a H somewhere). Generally, a mismatch between flash/psram voltage and what the ESP32 is configured as will lead to unspecified behaviour; generally you can't upload to or execute from flash. Note that you may be able to distinguish the two modules programmatically; the ESP32-Wrover-E always has V3 silicon in it while I think the non-E version has earlier silicon. (You can read out the version using esptool.py or espefuse.py iirc.)

[rkager]

Thanks for your support in this case!

So we have now checked the chips.
We have WROVER-E (ESP32D0WDQ5 rev3) with Flash Manufacturer: 20 which are working with our "wrong" HW bootstrap and 1.8V VDD_SDIO the new reel has the same chip but Flash Manufacturer: 68. There we get the errors during our production flash process.

So we will add the efuses burning for the VDD_SDIO in our production.

That older reels (with manufacturer 20) are working is just by chance and they can get faulty in the field, or do you have any information regarding this configuration/Chip?

Thanks