SDK Manager with Jetson Nano 4GB production module and custom carrier board

I am interested to know if, generally speaking, flashing a Jetson Nano production module, through SDK Manager, on a custom carrier board is going to automatically detect the 40pin header?

I am developing an AI project on a custom carrier board from Seeed P3449-0000 with P3448-0002 module and I am receiving errors when I try to manually configure the 40pin header to use the SPI interface.

sudo /opt/nvidia/jetson-io/jetson-io.py throws errors.

dmesg | grep spi

outputs

[    0.428824] iommu: Adding device 7000d400.spi to group 7
[    0.429153] iommu: Adding device 7000d600.spi to group 8

I am really confused if the carrier board from Seeed is supposed to work in a similar manner as the carrier board on the Jetson Nano 4g DevKit, regarding the 40pin header.

Our company ordered 10pcs / J101-v2 carrier boards from them and we are about to order more but for the moment our project is stuck due to this situation and we really question ourselves if we made the right choice to choose them. Their support is very poor.

I would really appreciate any advice on this topic.

hello konkhra,

you cannot use SDK Manager to flash a target on a custom carrier board.
please refer to Jetson Nano Adaptation and Bring-Up chapter, you’ll need a customize the pinmux spreadsheet if a board schematic differs from developer kit carrier board.

SDK Manager only support with DevKits,
thanks

Thank you @JerryChang for your reply.

According to this Wiki page Seeed is flashing their custom boards with SDK Manager.

you’ll need a customize the pinmux spreadsheet if a board schematic differs from developer kit carrier board.

I don’t think this procedure should be made by the clients.

We bought a product from a company, in this case Seeed, and we expect this product (J101-v2) to work out of the box, at least similar to the DevKit as they are stating on their product page.
In our case we can’t configure the 40pin header with sudo /opt/nvidia/jetson-io/jetson-io.py and I explaind here what is happening.

After a few emails with Seeed’s customer support they sent me to NVIDIA forums and try my luck here.

My basic questions still remain unanswered:

a) Is the 40pin header(SPI pins/interface) supposed to work on their J101-v2 custom board?

b) If the answer for the first question is YES why Seeed’s customer support is not helping me with that?

hello konkhra,

yes, this pinmux spreadsheet customize is the procedure that should be done by carrier board vendor.

according to that wiki page, did you have exactly same Jetpack release flashed to the target? it should be JetPack 4.6.1.
Jetson-io.py it only works with DevKits, it might also works if their board schematic as same as developer kit carrier board.

back to your questions.
since this is a customer carrier board, please check whether the schematic is identical to DevKits.
otherwise, it must change the pinmux configuration applied by the software.

Their wiki page is not that helpful regarding J101-v2 carrier board.

The only information regarding this board can be found at this page. At the bottom is a datasheet pdf file which doesn’t help in my situation.

hello konkhra,

there’s level shifter, please see-also applicate note for 40-pin expansion header.
however, the major issue here should be the pinmux configuration applied by the software. please contact with your vendor to double check this cfg setting.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.