BalenaOS for Jetson Orin Nano without developer kit

Product support
,
1

Hello,

I would like to know if it is possible to use balenaOS out of the box with the Jetson Orin Nano module not relying on the Devkit (so the SOM on our own PCB).

I see only support for jetson-orin-nano-devkit-nvme so I was wondering if this one works with the module only aswell (if of course we have an nvme).

Thanks in advance for your answer.

1 Like
2

Hello @avanmalleghem first of all welcome to the balena community!

We are working with different professional NVIDIA Jetson carrier boards manufacturers to bring up more devices compatible with balenaOS. Currently we have a process to support new devices and make sure the OS will be always compatible with that hardware. You can read more here.

What requirements do you have? Are you planning to use any specific carrier board?

If you are testing a Jetson Orin Nano device you might try that image, and that might work, or at least see the device online. Let us know the results.

3

We are on the creation of a custom PCB using this Jetson Orin Nano module.

And in fact I would like to know if BalenaOS is a viable solution out of the box or if I need to do something special in order to make it work.

Imagine it doesn’t work, do you have a documentation on how to make a custom PCB BalenaOS ready ?

Thank you very much for your answer !

1 Like
4

Hi @avanmalleghem,

It’s possible for balenaOS to work with the custom PCB out of the box, but only if the design uses the exact same kernel and Yocto components that NVIDIA uses and nothing more.

For example, we had a customer hoping to use this board out of the box with its related DevKit device-type, but upon inspection of that device’s BSP, there were a number of differences:

  • uphy configuration changes
  • pinmux and gpio config file changes
  • (these changes all end up in QSPI, and therefore require some balenaOS customization)

We no longer support customers making this sort of balenaOS customization themselves because it requires a lot of specialized knowledge and we don’t have a good way to determine who will have that knowledge until we’re in the midst of doing the work with them.

So in practical terms, I would say if your new hardware works with nothing more than what is provided by NVIDIA for their DevKits, you’ll likely be able to just use that device-type. But to be on the safe side, I would budget time and dollars to work with one of our integration partners to make compatibility adjustments for a custom device-type, in case it’s needed after all.

Here’s a high level look at that process:

  • you’d be sending two boards and all necessary components (power supply, cables, etc.) to balena
  • one of those boards would go to our integration partner and be returned to you after the integration is complete
  • the other board would be added to a balena AutoKit in one of our offices for as long as you use that device-type on balenaCloud
  • you’ll need to budget for related shipping and customs fees
  • you’ll also need to budget ~$6-10k for our Partner’s one-time integration fee
  • your business model will need to support the $2k per month of ongoing maintenance we charge for providing continuous balenaOS updates to your custom hardware
  • from the time you ship your boards, to the time a custom device-type appears in the balenaCloud dashboard, is usually between 1.5-3 months

Hopefully I haven’t overwhelmed you, but I’m a big believer in planning ahead for “worst case scenarios”, so want to share as much as I can with you in advance. We’re here to help project manage through that process, so you won’t be alone in it, but knowing what you’re getting into is a good thing too. :slightly_smiling_face:

2 Likes