I have put 2 fragments of each board’s console output on this Dropbox folder.
The original files are longer as I rebooted several times (hence the end of the Max fragment) but the excerpts in the folder correspond essentially to one (re)boot.
I have now read the instructions again and I can get to the OoWow boot on the Pro from the SD card but after I ask him to fetch an image (in the present case, Android) from the repository and it loads the compressed image in RAM to etch it into the eMMC, nothing visible happens, including on the serial console (no output). What should I do (or try)? (I will post some logs, when booting into OoWow, on the console, there appears (among other things) a mention that the eMMC is corrupt (bad CRC, I think). I reset the MCU once more to its factory state, but I can imagine that this does not reset or reformat the eMMC.)
I have made no further attempts so far on the new Max.
I’m quite lost. I etched the latest OoWow image just to be sure and followed the instructions for a “sure” boot from SD card into OoWow (3 fast pushes on the Function button). The outcome seems to be anything but sure … Actually the last attempts reflected in the following serial console files didn’t lead to anything and certainly not the OoWow menu. This holds for both my slightly older Edge V Pro and my all new Edge V Max … None installs another image, (typ. Android), successfully.
What can be wrong?
Hello @kialabi
We suggest you to install the image with USB upgrade tool.
You can follow the docs:
- Edge1 Install OS into eMMC via USB Tool [Khadas Docs]
- Download image here: dl.khadas.com - Index of /products/edge1/firmware/android/
And about the serail debug tool, please make sure:
- Whether the connection is correct? Take a picture about the connection
- Whether your serial debug tool support 1500000 baudrate？
- Serial debug tool setup correct? Take a picture about your setup
- Edge1 Setup Serial Debug Console [Khadas Docs]
No improvement … with neither of my 2 Edge-V (Pro and Max). Yes, the connection is checked and works fine (I have already provided console logs. I will put one more in attach, captured by minicom). Also, as demonstrated by the photos, the same setup with my VIM1S and Edge2 ARM PC works as expected but with them, there is a splash screen and then a boot. With the Edge-V, there is a lot going on on the console but nothing shows up on the HDMI display, not even the initial Khadas splash screen but maybe this one only displays wihen booting OoWow and/or from the SD card. After a few minutes, the Edge-V logs show about the same sequence repeating …
Regarding the flashing of the eMMC, as far as I can tell, it also succeeded (see below) but brought no improvement; so, the problem must be elsewhere but seems to be specific to the Edge-V.
– eMMC flash transcript —
$ burn-tool -v rk -i ./edge-android-10-v230524.img
Try to burn Rockchip image…
Burn to eMMC…
Rockchip Android image (or linux image compatible with AndroidTool one image burning) found!
Try to burn Rockchip image…
Using /media/elrond_2tb/devel/thirdparty-use/khadas/utils/rk-flash-tool/tools/Linux_Upgrade_Tool/config.ini
Program Log will save in the /home/alain/upgrade_tool/log/
Loading firmware…
Support Type:RK330C FW Ver:a.0.00 FW Time:2023-05-25 10:11:20
Loader ver:1.26 Loader Time:2023-05-24 20:47:24
Upgrade firmware ok.
Time elapsed: 1 minute(s).
Done!
mer 11 sep 2024 21:00:22 CEST
$ lsusb | grep Rockchip
Bus 001 Device 013: ID 2207:0006 Fuzhou Rockchip Electronics Company rk3399-Android10
– eMMC flash transcript end
Link to the minicom capture of reboot after reflashing : link
Various photos showing the setup of the various boards, some with the HDMI screen or the serial console terminal in the background.
Please supply the power with our official PD adaptor.
Does this log is from the image edge-android-10-v230524.img?
So you can enter the Android serial debug console and can operate well but only no HDMI display?
Can you provide the below information to us under Android serial debug console?
$ su
$ cat /sys/class/drm/card0-HDMI-A-1/modes
$ cat /sys/kernel/debug/dri/0/summary
I’m at the office now, so I cannot try or answer all points. I will have a look tonight and let you know. Thank you for your recent answer.
Regarding the power supply, it’s a 120 W power supply I use with a number of devices (including the other Khadas models and it works well with these, so I don’t see why I should suspect it and try something else (but I note your recommendation as an option).
Yes, the minicom.cap file pertains to this image. My bad; I should indeed have renamed the file for clarity but I was a bit in a hurry 
i cannot say whether display is the only problem, but yes, the machines (both the Pro and the Max) seem to be quite operational and to boot well '(with the display at time totally black - what still means that some signal is received - and sometimes with the mention “No signal” - what is not necessarily abnormal during some phases of a boot sequence -. Display is not flickering; it is either black or “no signal” for significant periods of times, reproducibly. Otherwise, yes, if I interrupt the autoboot sequence, I can log into the device (I will send you the output of the commands you mention. Please note that I have a video switcher between the Khadas and the TFT display (I will also try without), but again exactly the same setup is no problem for the VIM1S and the Mini ARM PC.
After a while (say, 2 minutes, the serial console does not output anything for a while (see the timestamps) and then outputs about the same sequence of log records as just before, and then pauses for a number of seconds … Can you tell whether these log records are normal or denote some error conditions? From experience in Android debugging (smartphones, tablets), I know that Android can be quite verbose, so I’m not particularly anxious at the number of messages. However with these ones, I don’t know. On a related note, there are error messages before (like for example before the wired network becomes really operational, but all error messages don’t sound particularly alarming to me … And they seem to be about the same, upon superficial examination, on the Pro and the Max …
Stay tuned …
Best regards
I experience difficulties in interrupting the autoboot sequence. I have the keyboard of the computer running minicom and I also have a USB keyboard connected to the Edge-V. None of them seems able to interrupt the autoboot sequence: I see the message: “Hit any key to interrupt the autoboot sequence. 2 s” (then “1 s”) but the key does not seem to be taken into account. What should I be doing?
Should I try to install Ubuntu instead of Android, just to check that all the hardware is OK? Or is there another UART through which I could try to get into the device with Android? My target OS remains Android, as I want to do development on the Edge-V but if I need to temporarily install Ubuntu to troubleshoot the issue, I can try.
This should be done on your local PC keyboard not the keyborad to Edge-V.
And you don;t need to stop the autoboot, just setup the serial debug console and boot to android and execute the command in serial debug console with the keyboard to your local PC.
$ su
$ cat /sys/class/drm/card0-HDMI-A-1/modes
$ cat /sys/kernel/debug/dri/0/summary
Thanks, that’s what I assumed. I will try again but it looks like the console is unidirectional (or read-only). I don’t see a prompt appearing, just log records but I will have a closer look.
Some things continue to look very strange: just to name 2:
- The Khadas splash screen appears only very infrequently during reboots
- Out of a sudden the Android screen has appeared once (over many tries) a few minutes ago (and very briefly) when starting the Edge-V Pro (not the Max).
Otherwise:
- I have been able to make the serial console bidirectional (Hardware flow control was still on ON; put it to OFF). Conversely, there are now a few spurious characters on the console that corrupts input command line. I have therefore been able to issue the commands you requested (see files sys_cmds_*.txt in the folder)
- I have also booted from SD card with a generic Ubuntu image. Both machines boot wel but the GUI does not start (maybe it’s normal; I haven’t checked the boot configuration) I haven’t either try to log in, not having searched for the default credentials (ubuntu/ubuntu?)
Do you see anything peculiar in tne sys_cmds_* files?
Best regards
Hello @kialabi
The log seems normal, do you have a chance to check with other screen?
Different board have different power requirement, does your adaptor is PD or QC? Does have details of this adaptor?
To answer your question about the power supply, it’s a travel (multi-standard) adapter with PD.
But I have now solved the mystery (thanks for your help, anyway): the problem seems to have to do with the HDMI source switcher that was between the Edge-V’s and the TFT display. I have tried directly with a portable display and everything works as expected and reproducibly. I don’t think the switcher is faulty, as it works with everything I have used it with so far, including the VIM1S and the Mini ARM PC but maybe the Edge-V is more sensitive to timings or to some signals …
The troubleshooting part of this thread may then be closed. I sincerely appreciated Khadas’ willingness to help (as I had already witnessed with other customers. Great!) Yet, my initial question remains and I too am willing to give a hand if possible. There is not much specific to the Edge-V in Khadas’ documentation. How can we improve it? Never mind if it boils down to writing a page along the lines “The Edge-V is a variant of Edge-1 that differs only in the following respects : (list). Most instructions and procedures described for Edge-1 are also valid for Edge-V.”
What’s your opinion about that? (Without it, it sounds a bit like Edge-V was just an attempt to let Edge-1 evolve but Khadas does not want to support it much more, and it leaves buyers perplexed …)
1 Like
Hello @kialabi
Glad to hear that you have resolved the issues.
And thanks for your suggestion about the documentation missing part of Edge-V, we will improve it.
Cheers.