ESP32-EVB-EA Ethernet problem in some units

[nestornp]

We are using the ESP32-EVB-EA board for a project on my company, we are having problems with some of the units that we have purchased, we need them to have an Ethernet connection, however many of them do not connect to the internet and we get this error :

E (3541) emac: Timed out waiting for PHY register 0x2 to have value 0x0007(mask 0xffff). Current value 0x0000
E (4542) emac: Timed out waiting for PHY register 0x3 to have value 0xc0f0(mask 0xfff0). Current value 0x0000
E (4543) emac: Initialise PHY device Timeout
[E][ETH.cpp:101] begin(): esp_eth_enable error: -1


It only happens in some of the ESP32-EVB-EA in which, by chance, their UID starts with B (e.g BCDDC2F2A5D5), on boards whose UID starts with 2 (e.g 240AC4249310) we don't have this type of problem (same code). I do not know if this should be discussed with the technical service or put it directly here.
Can you help me? Thanks a lot

PD: all non-working units are Rev.H.

[LubOlimex]

all non-working units are Rev.H.

And what about the working units? Are they also revision H? Can you check if both starting and not starting boards have capacitor C31 present?

Can you try to use the on-board reset button RST1 on the failing boards? Does the Ethernet start after this RST1 button reset?

[nestornp]

Rev. I and Rev. F versions work perfectly.

I have a Rev. I version with a C31 capacitor that works perfectly, and I also have a Rev.F without a capacitor C31 that works perfectly, the Rev. H units that are failing don't have a c31 capacitor and the ethernet doesn't work either after pressing RST1...

[LubOlimex]

Can you send me a picture of R24 of the failing boards marked as revision H? Make sure to take it at 45 degrees angle, I want to see if there is one resistor or two (one over the other).

I think this problem can be traced in the hardware revision changes notes here:

https://github.com/OLIMEX/ESP32-EVB/tree/master/HARDWARE

I think it was introduced in the first revision H boards when the design was improved for lower consumption on battery in deep sleep modes. The problem was not discovered until customer feedback, it never happened in our environment here for some reason; a workaround was introduced in H1 (same PCB and marking as H), and definitive solution was implemented in hardware revision I (new PCB).

[nestornp]

It does not appear that there are two resistors one on top of the other.

[LubOlimex]

If the board is revision H marking and there is only one resistor on R24 position with marking 474 (as it seems in your pictures) this can lead to start up problems in certain hardware setup. Please solder 1k resistor (size 0603) over the one already present at R24 (474 marking means it is 470k. You can also remove the original 470k one and solder 1k on its place, if that is easier for you (but soldering over the original one is faster and easier usually). Test again.

[proddy]

I have the same problem (Rev H without the 2nd resistor). Did adding the additional 1k fix your problem?