Does not boot, does not show in lsusb even in maskrom mode

#1

I have tried connecting to both windows and linux PC:s, with 3 different USB cables (two type A to type C, one type C to type C), and the device does not show up in lsusb on linux (Ubuntu 22.04.1) , or the devtool in windows (up to date updated Windows 11).

I do not have an USB to TTL adapter, but I can order one if it is needed.

I have made sure I am pushing the right button when plugging the device in (the button that is pressed perpendicular to the PCB, of which there are only one on the board), and that my NVME drive is not plugged in, and my SD card slot is free, and all the required software/drivers are installed.

Naturally it will not boot from the NVME, or from the micro SD card either.

What can I try to awaken the beast?

#2

Did you install their device driver? Step 2 here - then reboot.

I was able to do this process using updated Window 11, and have booted both Ubuntu and Debian from SSD. The button you need to push is labeled (inconspicuously - I needed a magnifier) MASKROM - press and hold while you apply power then release. The LED starts blue and stays blue - there’s no indication you’re in MASKROM mode.

I didn’t unpack the dev tool properly which caused some head-scratching as I tried to understand Chinese error messages… You must unpack while preserving their directory structure with

7z x RKDevTool_Release_v2.96-20221121.rar

#3

Yes, all the required packages are installed on Ubuntu, and the drivers installed on windows.

When I keep the button pressed (and yes, now that you said it, I can use a magnifying glass and a periscope to notice it is clearly labeled MASKROM), when inserting the USB cable to any of the computers, the power LED stays green all the time. Never blue; the same thing with stand alone powering it with micro SD card (and not holding down the MASKROM button), always green, never blue.

#4

Perhaps the button is broken? It seems to expose metal, should I short the sides together while booting? I expect that is what the button does, when holding it down.

#5

When I first got mine I plugged it into power, the green light came on, and the machine did nothing as you are describing. Following the getting started instructions on the wiki worked for me. It required creating a SD card with Balena Etcher with the Debian image on it. When I started the machine with the programmed SD card plugged into it, the blue light would come on and flash twice every 1.5 seconds or so to let me know it was running.

If you have an SD card and an adapter for your PC I would suggest trying that method of getting it to boot then if that works then you try to work out the NVMe issues afterwards. I don’t know if the u-boot firmware that comes pre-installed can even support booting from NVMe out of the box. Once you update the machine and get the u-boot update I know that you are supposed to be able to boot from NVMe, but you may have to boot from SD card first to perform the updates.to u-boot firmware.

#6

Yes, when I write it does not boot either via SD card or the NVMe, I mean when those media have bootable, compatible OS installed. I have tried to burn the SD card with both Balena Etcher, and with Rufus, and the image has been the latest Debian. The NVMe was installed from Debian on the board itself.

To start with, the board worked; then I followed the instructions to flash the NVMe boot capable uboot in Debian, on the device itself; that worked as well (although perhaps a bit suspiciously, the writing for both the empty and the real ROM took exactly 180s - perhaps it just broke something and timed out?). The system came back online after “shutdown -r now”, and booted to Debian fine. Perhaps not actually taking power off the device caused a soft reset that bypassed the uboot during this initial restart?

I turned the board off for a couple of days, and when I tried to power it on again, it does not boot, even to MASKROM mode.

#7

hi
had same boot issues, changed the power supply, no booting problems with debian or ubuntu anymore
take care your power supply supports 2.5 -3 A 5VDC

#8

When stand alone, I have tried multiple power sources, including 5 different PD/QC3.0 sources and a 12V/250W ATX power supply, each functional one performing exactly in this same manner (just green light, that is it - however, some PD sources do not seem to power it).

I just used one of them when it worked, but tried many when I saw in the forums there had been issues during bringup with a lot of power sources. I have hard time imagining this as a power supply issue.

#9

Can you try connect some serial cable and check if there is any output? https://wiki.radxa.com/Rock5/dev/serial-console

#10

Not yet I can not, as I do not have a serial to USB converter.

I did order one, and the expected delivery time seems to be on 14th December. I should be able to check the output next week.

#11

There is nothing on the serial console, not when I give it power without pressing the MASKROM button, and not when I connect it to the ubuntu host:

Welcome to minicom 2.8

OPTIONS: I18n
Port /dev/ttyUSB0, 07:52:37
Using character set conversion

Press CTRL-A Z for help on special keys

#12

Ok, so it is dead, and Radxa says they have nothing to do with their boards, the resellers bear the responsibility for warranty, and at least allnet china provides no such thing.

So be warned, do not buy these from allnet china until Radxa has their software mature enough to not brick your devices. And do not update your boot rom while running the device in ubuntu.

I repeat, do not follow the instructions in https://wiki.radxa.com/Rock5/install/spi performing Step 4, Option 1. You will (or at least I did) brick your device, and you will lose your new toy.

You have been warned, I have been burned.

Happy holidays!

2 Likes
#13

i’m definitely no expert but if it’s not booting because of what’s on the SPI it may be possible to buy one of those usb flashers with clips you can attach directly onto the SPI chip and then use flashrom to restore functionality

#14

Just a data point that I did the Step 4, Option 1 method without issue.
When I did it, I was booted into Ubuntu on SD card.
After, Debian on NVME.
Using a 25W PPS Samsung “Super Fast Charging” adapter, with this cable from amazon: Ocetea USB C to USB C Cable(100W/5A), 6.6ft Type C to Type C Cable with LED Display, Nylon Braided PD Fast Charging Cord Compatible with MacBook Pro, iPad Pro, Galaxy S22, Pixel, Switch, PS5, LG
(The cable is cool, get to see watts drawn real time)
It’s been running Debian off the NVME (Samsung 970 Evo 1TB) since the initial flash, plenty of reboots and tinkering around, no problems so far.
I do note that I have 3 different multi-port USB-C charging hubs on my desk, none of those will work to boot up the Rock 5 correctly (they all re-negotiate voltage briefly when a device is plugged in, i suspect this interrupts the boot process)
The Samsung 25W brick doesn’t re-negotiate its voltage and works ok.

#15

@ silverj. Many thanks for the info you shared. I received my marvelous new single board computer (Radxa Rock5b 16g) on the 5th of December. I haven’t taken it out of the box yet because I have no clue on how to put android os on either my sd card or preferably my 1tb nvme. My buddies on YouTube (eta prime, NicoD and explaining computers) did videos on Rock5b but not a tutorial. I’m pretty good at following instructions but it seems to me that the information available at present for installing os I want is confusing. At least to me. My condolences about your board. We spend good money and even much more valuable time energy and effort dealing with these gadgets to have them break due to complications installing software! Thanks again. My board was ordered from Allnet as well. I’ll take your advice to heart.

#16

@wilsonyan if you have one, and want to try, I can send my board to you if you PM me with your address.

I would kindly ask you to reply here if you succeeded or not, but it would be your 16GB board after you receive it, so up to your conscience if you would feel like following up on it or not.

#17

I have an older flash memory programmer, but not sure if I have the right adapter for the IC used to store ROM on Rock 5B. If I have it I could try reading the working image from my rock5 and writing it to your dead one… mine is in a case, so I can’t quite read the writing on the NOR flash chip…

https://wiki.radxa.com/mw/images/b/bb/Rock-5b-hw.png Could you take a pic or read the writing on the IC? the one labeled MXIC…

#18

Looks like it’s this:
https://www.digchip.com/datasheets/parts/datasheet/2/272/MX25U12832F.php

#19

Thanks, Sarah!

My programmer (TL866II Plus) has support for MX25U12835F, it’s not the same exact part # but looking at the spec sheet it’s the same except support for different package types. I also have the proper chip clip to make connection… so I can try doing it this way, taking the working SPI from my Rock 5B and flashing it to silverj’s Rock 5B. Assuming bad SPI is the problem, this should revive it. Note the board needs to at least get power (green light) for this to work, as the chip needs to be powered up to read/write.

Edit: just a note also, @silverj: it is expected for the write speed on the SPI NOR flash to be the same regardless of writing zeroes or actual data, the programming cycle is the same “read, erase, write” for each block on the chip, and flashing a NOR IC over SPI is a very slow operation, so providing that the zeroed-out file and the actual firmware file are roughly the same size, this is correct behavior.

#20

@DagYo let me know your address, and I’ll send it to you.

Yes, the empty and the actual firmware files were the same size. At 180s for 16MB file, the final output speed was 91KB/s, that does not sound too far off what it should have been. I just grew posthumously suspicious when the time matched so exactly with 3.0 minutes - for what I do, those are often synonymous with “something timed out”.