Erratic behavior of VDD_SDIO when configured to 1.8V

[petnnw]

Hi there!

I'm experiencing strange behaviors on the VDD_SDIO voltage line: it is rather unstable, and among all the samples I have (and operating conditions) it varies from 1.7V up to 2.04V.

Setup:
Custom ESP32D0WDQ5 (revision 1) board with 4 MB Flash and 8MB PSRAM @ 1.8V. MTDI is pulled high. VDD_SDIO-related eFuses are not programmed (read as 0), so MTDI determines the voltage (1.8V).
VDD_SDIO is only connected to the flash and RAM chips (of course with the decoupling caps). The 2kOhm and 4.7 uF capacitors are connected to VDD_SDIO, and all the other passives closely match the values on the ESP32 Wrover module schematics.
The ESP32 is powered by a 3.3V DCDC converter, which is able to provide up to 1A. The voltage, measured with a scope is stable, with less than 20mVpp ripple, at the switching frequency (> 1MHz).

Detailed description:
When the device is in bootloader mode, the voltage is typically 1.70 - 1.75V in all samples. Once I upload a sketch (even a blank one) using Arduino IDE, the voltage increases. On some samples it stays within 1.9V, but on many other sample it reaches a peak of 2.04V (according the flash and RAM datasheet the maximum VDD should be 1.98V...) . This voltage is also very unstable. I measured it with a scope, and has a 100mVpp sawtooth type oscillation: it jumps to about 2V and it decays almost linearly to about 1.9, with a frequency between 1.4 and 2 Hz. This instability is present also in bootloader mode, but the average voltage is about 1.75V. The exact same behavior occurs even on boards without soldered RAM (i.e. even when just the flash is present).
Despite all of this, all the sketches run fine.

Question:
Has anyone experienced this? Is this instability intrisinc to the regulator? (the requirement of the external 2kOhm resistor is rather suspicious...).

Cheers!

Nicola

[ESP_Sprite]

The 1.7V early on is to be expected: the issue is that the LDO in practice outputs a voltage that is a bit lower than expected and we compensate that as soon as the app boots. (This is configurable in menuconfig if you use esp-idf, by the way.) The sawtooth oscillation is also expected: it's a side effect of the LDOs design.

[petnnw]

Thank you very much!

Is there aregister I can write to, to quickly disable via firmware this increased voltage when running the app? While 1.7V is within specs of RAM and Flash, 2.04V is not (the specs of my chips are 1.8 +/- 10%).

Cheers!

[ESP_Sprite]

I don't know off the top of my head, but the boost to 1.9V is enabled here, perhaps you can find how to get it back to 1.8V from there. Alternatively, in esp-idf this is configurable in menuconfig and you can add Arduino as a component to esp-idf, so you can also go that route.