Identifying potential hardware failure/breakage of ZERO 3E

Yesterday I wanted to try replicating a potential problem with the Ethernet port with our OS images on a somewhat new Radxa ZERO 3E. Sent by a teammate, at least I used it for the very first time.

I flashed our image, which booted up fine, showed serial console output as expected, connected to the network via Ethernet as expected, I connected via SSH, started to install some APT package upgrades. As extracting the packages took somewhat long (maybe even that was first sign of a problem), I moved the board out of the way to work on something else in the meantime. At this moment the SD card was gone completely: SSH output showed it was remounted R/O as ext4 does by default when facing write errors, but when trying to remount it was clear that it could not be read either, all I had was the running bash with its builtin commands, while the rootfs was completely gone.

So I unplugged power and replugged. Serial console showed SD card was not really detected anymore, or at least it could not load everything and boot from it, I did not further check but expected a corrupted filesystem and reflashed it. From that point on, when trying to boot, the serial console did not show anything anymore, green LED steady. Tried another SD card but same.

I thought I might have damaged the SD card slot so that it cannot even find the bootloader on it anymore, checked the mechanical part of the contacts, including the two contacts at the side which touch each other when inserting a card, compared with some high resolution photos of another user, and in the meantime I don’t think anything is wrong there, it all behaves as expected, the two contacts touch, all bottom contacts touch the lanes on the card, it sits tight etc, nothing unusually bent. Also everything else on the board looks good, nothing burned or unusually colored.

First of all, even if no SD card is inserted, there is still supposed to be some ROM-wise output on the serial console (pins 6-8-10 at 1.5 GHz baud rate), isn’t it? That would rule out an issue with the slot already. Then no idea, it just stopped operation mid usage the first 20 minutes I ever used it, first not able to boot from SD card anymore, still with serial console output, next attempt no output at all.

My first ever broken SBC among 35+ SBCs in moreless regular use. If someone has an idea how to verify a hardware failure or what to watch out for, I’d invest an hour more before throwing it to the bin or checking whether a refund/replacement is possible.

Hi @MichaIng

Sorry for the inconvenience.

When you insert the SD card with the operating system, do you see any output on the serial console?

Please note that the Zero 3E has no onboard storage other than the SD card. So if there’s a problem with the SD card slot, there will be no output at all on the serial console. (Of course, there could also be other reasons why the system won’t boot properly, such as a damaged board or CPU issues.)

If you have any related logs, please share them with me.

Have you tried reflashing the system?
Here is the official image from our website:
https://github.com/radxa-build/radxa-zero3/releases/download/b6/radxa-zero3_debian_bullseye_xfce_b6.img.xz
You can give it a try.

If possible, please also provide a real photo of your Zero 3E—especially the SD card slot and the power adapter you’re using to power the board.

Thank you.

2 Likes

No output at all, not when powering with or without SD card (aside of the few gibberish characters due to power spike), neither when inserting it while powered.

Okay that answers one of my question. So there is no output from some board ROM before the SD card is read, but everything is coming from the bootloader on the SD card, or nothing at all. I am aware there is no SPI storage, but was not sure whether there is some initial output before it is even decided whether to load the bootloader from SD card, or SPI storage (if one exits).

Sadly really none, the serial console is/remains blank.

Yes, when the system lost SD card access for the first time, I rebooted once, which lead to early errors in bootloader phase. As I thought it might be just filesystem corruption due to the mid-write I/O access loss, I reflashed the same image, but it did not boot at all with mentioned empty serial console output. I then flashed another SD card with same result.

To rule out any issue with our image, I just downloaded and flashed the one you linked. And surprisingly it did boot. So I flashed one of our images again. At first it did not boot, then I tried another one of ours, and it did boot. Tried the one which did not boot at first again, and it did boot as well. In the end trying a bit back and forth, it seems the SD card slot indeed, and some tiny nuance in how I insert the SD card, which is so small that I cannot really replicate functional and non-functional way of inserting the card.

When I compared my board to the high res photos of another user, it looked like the outer contact on the one side of the slot, which touches an inner one when a card is inserted, is bent slightly more inwards. So I carefully pressed it a bit inwards when being about to give up. Not sure, maybe that fixed the issue partly, though those two contacts did also clearly touch each other before.

Some photos I did before testing your image, sadly not so high quality:

Maybe there is some mechanical detail at the SD card slot which must work when SD card is inserted, so the board can detect and load a bootloader from it?