Jetson Nano B01, no graphical environment

I think that I’ve almost got it sorted out. I tried both my monitors on my Windows 10 machine. The larger of the two HDMI monitors did not work at all. The smaller one initially did not work, but I tried a long-shot which was to replace the USB cable used for power and it worked.

I connected the smaller one to my Jetson Nano using the new power cable as was able to get up the Ubuntu login screen. Unfortunately, I cannot get the cursor to move but I’m sure that I will be able to sort that out myself. I don’t think it’s a Jetson Nano issue but rather something simple with the smaller monitor, so if I can’t get it sorted I’ll contact the monitor manufacturer.

I will let you know the outcome. I would like to use a larger monitor however. Is there a list of compatible monitors available?

Also, if you would like me to send you any log files for your own investigations I’m happy to do that. Thanks for sticking with me on this one.

HI,

Could you share me sdkm log with “export logs” button on sdkmanager? I would like to make sure you really get the board flashed.

If this is true (really flashed the board), I will consider this is hardware problem on this jetson nano. Please dump the sdkm log first. Thanks.

I checked the nvidia download folder for the logs files from my first use of SDK Manager but it was empyy. I’ve tried SDK Manager again in order to get the log files but the SSH connection to the Jetson Nano keeps dropping out. There is a message that I can try a Manual set-up. How do I do this?

The log file we are checking this time is on host. Not jetson nano.

Without using SDK Manager, my smaller HDMI monitor works fine. The larger one only displays, “No Signal”

Using the SDK Manager, the result is the same.

SDK Manager would always lose the SSH connection to my B01, however, I read the terminal window output and it looks like the B01 was successfully flashed. I’ve attached a file with the last part of the terminal window output.

I’ve also attached a list of the files produced as part of the error log export. I did not attach the files themselves because of the size. Even when they are zipped up the zip file is over 3GB. et me know if you would like me to send you some or all of these files.

I may be wrong, but I think it is my larger HDMI monitor that is the problem. My smaller HDMI monitor is too small to use for extended periods. So, all I have to find is a larger HDMI monitor compatible with the B01. May I make a new topic on this forum to solicit suggestions from other users?

If you wish me to produce other log files etc. for your own investigations, I am more that happy to do so.

list_of_error_log_files.docx (7.0 KB)

SDK_Manager_end_of_terminal_window.docx (8.0 KB)

Your comment still has some points that I cannot explain. It feels like something is still wrong.

  1. What do you mean “without using sdkmanager”? Are you talking about using sdcard image? Sdcard image actually has not much difference to sdkmanager. Are you sure this is not an issue coming from the HDMI cable?

  2. SDK Manager would always lose the SSH connection to my B01, however, I read the terminal window output and it looks like the B01 was successfully flashed. I’ve attached a file with the last part of the terminal window output.

My point here is to dump the flash log to me so that I can make sure you really flashed the board correctly.
The method is to click the “export logs” button in below picture of your sdkmanager.

  1. I am not sure how you passed the configuration before. Every new flashed jetson device requires the user to configure their user account/password so that the system can really boot up. If you have some sense about it, you will understand that such password and user account is required for ssh. Which means you cannot ssh to the device until this step is done. However, you monitor fails to run and you didn’t say how you configured the user account in the past time and even now.

There are two ways to configure user account. First one is using keyboard and HDMI monitor to configure. But as you said your monitor has nothing shown, this could not happen on your side. The second method is done by the help with host side ubuntu machine as below page says.

https://docs.nvidia.com/jetson/l4t/index.html#page/Tegra%20Linux%20Driver%20Package%20Development%20Guide/flashing.html#wwpID0E02C0HA
→ Headless Mode Flow in oem-config

As a reminder, I have two Jetson Nano B01s, my host computer is a Lenovo T420 laptop loaded with Ubuntu 18.04. I was using Ubuntu 20.04 but it was not compatible with SDK Manager.

I am working in headless mode, however, I do have two HDMI monitors. I have not been able to get the larger one to work with the B01. The smaller one works fine with the B01.

When you write “configuration”, I understand this to refer to what I have been calling “setup”.

  1. Yes, I have been using an microSD card image.

I am using a file called, “Jetson-nano-jp46-sd-card-image.zip” that I downloaded from a link on the NVIDIA Jetson Nano getting started website:

With my larger HDMI monitor, I have tried an HDMI
cable that I know that works but I still get a message that there is no signal.

  1. I made a mistake saving the flash log file. I did not push the save button. I’ve attached the zip file.

Here’s what I do when using SDK Manager:

  1. I flash a 64GB SanDisk microSD card using Etcher.

  2. I insert it into my B01, connect my B01 to my host using a USB cable and plug in 5v, 4A power using the barrel connector, J25. I have jumpered J48. I then connect to the B01 from my host using: sudo screen /dev/ttyACM0 115200

  3. I then configure the B01 according to the following as per the NVIDIA getting started website:

“First Boot

A green LED next to the Micro-USB connector will light as soon as the developer kit powers on. When you boot the first time, the developer kit will take you through some initial setup, including:

  • Review and accept NVIDIA Jetson software EULA
  • Select system language, keyboard layout, and time zone
  • Create username, password, and computer name
  • Select APP partition size—it is recommended to use the max size suggested”

I do not configure the network connection.

I’ve read though the information on the link that you provided: https://docs.nvidia.com/jetson/l4t/index.html#page/Tegra%20Linux%20Driver%20Package%20Development%20Guide/flashing.html#wwpID0E02C0HA

It is the same except that I am not prompted to create a swap file.

Once configured, I connected to the B01 using the command: ssh james@192.168.55.1

I run SDK Manager. Eventually, SDK Manager tells me that it cannot connect to the B01 and asks if I want to retry (see SDK_1.docx attached). I press the retry button but it still won’t connect (see SDK_2.docx attached).

I export the log files (see attached zip file).

I connect my host to the B01 using: sudo screen /dev/ttyACM0 115200. I configure the B01 again as per the NVIDIA getting started website.

I connect my smaller HDMI monitor to B01 and it works fine. I connect my larger HDMI monitor to the B01 and I still get a message on the monitor that there is no signal.

SDK_1.docx (223.1 KB)

SDK_2.docx (334.7 KB)

SDKM_logs_JetPack_4.6_(rev.2)_Linux_for_Jetson_Nano_modules_2021-10-29_18-38-02.zip (140.4 KB)

  1. There are some steps that do not needed. When sdkmanager is going to flash the board, it will copy the file system to the sdcard. So you don’t need to flash your micro SD with Etcher before running sdkm. Just a empty card would work. Even if it is not empty, sdkm will erase the content and re-install. Thus, your step 1 here is not needed.
    Just in case the board is in recovery mode and host will be able to flash it.
    But above steps do not matter anymore because I saw your zip file have “NV_L4T_FLASH_NANO_WITH_OS_IMAGE_COMP.log” and the result says it flashed the board. So sdkm flashed your board successfully.

  2. And can you clarify what is the exact order of below? Are you sure you are writing these steps in a correct oreder?

Once configured, I connected to the B01 using the command: ssh james@192.168.55.1
I run SDK Manager. Eventually, SDK Manager tells me that it cannot connect to the B01 and asks if I want to retry (see SDK_1.docx attached). I press the retry button but it still won’t connect (see SDK_2.docx attached).

Are you talking about you can ssh to the jetson after the configuration but you still cannot install sdk component with SDK_1.docx log?

  1. Actually, those sdk components in SDK_2.docx are not related to this issue at all. They are not needed to let monitor work. Take Windows system for example, cuda is able to run on Windows too. But you don’t need CUDA to get installed on your Windows host to make your HDMI work. We can just skip them at this moment.

And I just realize that you have two jetson nanos here.

Did you already try to flash both of them with sdkmanager and boot up? Are they both hitting the same HDMI issue?

Yes, I have two Jetson Nano B01s. My host is loaded with Ubuntu 18.04. I also have a laptop with Windows 10. I have tried both my monitors with the Windows 10 laptop, Windows 10 does not detect the larger of my two monitors but it does detect the smaller of the two, but I I only tried this out with one of the B01s. The smaller one works with both my B01s and the larger one works with neither B01.

In answer to your questions:

  1. All I can add is that I did the same with both my B01s.

  2. Yes, these are exactly the steps that I took. I did these steps on both of my B01s with separate 64GB microSD cards.

  3. If I understand correctly, yes, I can ssh to both B01s after configuration even though I get the Installation Failed message in sdkm.

From your subsequent email, Yes, I tried to flash both B01s using sdkm using separate microSD cards with the same result.

fyi. I have found that my larger monitor also accepts VGA input. I will receive the required cable to try this out on Wednesday, pm.

This point is not directly related to this issue. But I have to tell you that sdkmanager installing the sdk through scp and ssh command too. So this does not make sense that you can ssh to the board while sdkm cannot. So below comment does not make sense. Must be something wrong here.

If I understand correctly, yes, I can ssh to both B01s after configuration even though I get the Installation Failed message in sdkm.

Tegra does not support any VGA to HDMI adapter. I guess that is the reason that this monitor does not work.

But I am not sure how you did your test on your larger monitor if you didn’t have any “required cable”. What will that “required cable” look like? If you didn’t have that cable, then how did you test your larger monitor so far?

I’ve only tried an HDMI cable with both monitors. I’ll just cancel the order for the HDMI to VGA cable.

Honestly, I don’t suggest you to try that as we already had lots of users who suffered from problem with HDMI to VGA cable.

If your monitor has a HDMI port, then it should support HDMI. There is no need to buy another HDMI to VGA adapter just to convert VGA signal again to HDMI and try.

My personal suggestion is maybe you can ask some friends to share their HDMI monitor with you and see if changing a HDMI monitor would work or not…

I’ll try to repeat what I wrote about ssh.

Sounds good. Thank for all your help.

There is a local computer store that sells used HDMI monitors. I’ll take my B01 there and I’ll be able to test one out before I buy. In the mean-time I’ll just use my smaller HDMI monitor.

If it is of use to you, I performed 3 tests with SDK Manager. Details of these tests are attached along with zipped folders including screenshots and error logs.

test_details.docx (7.9 KB)

case_1.zip (1.1 MB)

case_2.zip (663.3 KB)

case_3.zip (145.2 KB)

I think you need a correct guidance here.

When installing sdkmanager, you don’t need to pre-install your sdcard with sdcard image.

What you are doing in case 1 is already “partially correct”. I added some comments in below.

Case 1: sdkm in manual mode, no initial ssh connection
a) Used Etcher to flash this file to 64GB microSD card: → This is not needed.
jetson-nano-jp46-sd-card-image.zip
b) Inserted microSD card into Jetson Nano B01, connected USB cable and power and jumpered FC REC with GND → Yes, this is correct way to put board into recovery mode.
c) Initiated SDK Manager and selected Manual Mode
i) SDK Manager said that there was not an ssh connection
ii) Exported error logs:
case_1_SDKM_logs_JetPack_4.6_(rev.2)_Linux_for_Jetson_Nano_modules_2021-11-01_07-08-18.zip
d) Exited SDK Manager and successfully configured Jetson Nano as per: → Ok, there may be something wrong but this is not related to this HDMI issue.
https://developer.nvidia.com/embedded/learn/get-started-jetson-nano-devkit
e) Larger HDMI monitor displays, “No Signal”. Smaller HDMI monitor works.

What we are focusing on is just why the HDMI cannot work, the ssh issue is a separate one that does not matter for the HDMI. If there is any concern about it, better filing a new topic and we can check on the new topic.

But currently we need to see if HDMI monitor is the cause of your issue.

What you should share is the 2 dmseg logs when you use larger monitor and smaller monitor. That might be more helpful for current situation.

Attached are the requested 2 dmseg log files. In both cases, I hot-plugged the monitors.

smaller_monitor_kernel.log (110.1 KB)

larger_monitor_kernel.log (189.1 KB)

Hi,

Need to teach you the correct way to dump log. It is obvious that you are sharing me a log with multiple scenario inside the same log.

I mean your “smaller monitor log” and your “larger monitor log” are actually the same log and you just split them by yourself. This is not a good way and it just creates more confusion to others.

Please just give one complete log for one scenario.

For example, connect your smaller monitor to the board, boot up, hotplug the cable once after boot up, dump the dmesg. → Share as log “smaller_monitor.log”.
And then reboot the machine, repeat above case with larger monitor this time and share the “larger monitor log”.

Do not mix both of case in the same log. Also, please do not wait for 3000 sec to do this. Just do it within few mins after the boot up. Otherwise the log might be missing.

Attached are two kernel.log files produced with the following procedure:

  1. Power off and unplug monitor
  2. Power on
  3. Delete kernel.log file
  4. Reboot
  5. Plug in monitor
  6. Reboot
  7. Generate kernel.log file with: dmesg >> kernel.log
  8. scp kernel.log to host
  9. Rename kernel.log file on host
  10. Repeat with other monitor

small_monitor_kernel.log (75.7 KB)

large_monitor_kernel.log (79.4 KB)