Jetson Nano won’t boot without UART to USB attached to serial console

I’m having the same exact issue as this person:

I have a custom motherboard for the TX2, pin compatible with the other Nano and Xavier. The SoM is not booting properly without UART to USB attatched so the console can print out the boot log. I’m trying to log in remotely via SSH through a usb ethernet adaptor connected to USB 2.0 of the Jetson. The usb ethernet adaptor is only recognized when the UART to USB adaptor is connected.

As soon as I confirm the ethernet adaptor is working and I have the IP address, I disconnect the UART to USB adaptor that connects to putty so I can go back to my desk, and all of a sudden I can’t log in through SSH anymore. I can tell the OS is not booting the same way if I leave UART unplugged because the current meter I have idles at 300mA, whereas when I have it plugged it in idles at 400mA and has a couple of spikes up to 900mA; the patterns aren’t the same.

Is there anything on the OS that would prevent it from booting without being able to display the console? The idea is to have a headless system that boots up and shows up on our LAN so we can remotely develop code on it. The end use-case is also headless.

I don’t want to disable the tty0 because we may need it for troubleshooting.

Does anyone have any suggestions on where I might start to resolve this?

For some reason if I remove the wires from UART GND TX and RX, the system boots as it’s supposed too, even though they aren’t connected to anything. If they’re connected to the USB-UART adaptor and putty is loaded then it works fine. Very odd problem to have but it works now.

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

Sorry, I see this topic closed so you probably cannot reply. I want to add though that during boot, before Linux ever loads, the boot stages offer a chance to drop into the boot console if they see activity on the serial UART. I suspect that your carrier board needs to use something like a 10k resistor on the RX side to prevent RX from settling to an “active” state; I don’t know whether that would be to ground or to 3.3V.