I wanted to tweak some u-boot parameters on a brand new Nano device (shipped last week, received Monday).
I followed the directions provided by Jetsonhacks for connecting to the serial console on J44. (https://www.jetsonhacks.com/2019/04/19/jetson-nano-serial-console/) This resulted in non-printable binary characters (shown in minicom as question marks) rendering to the terminal with roughly the cadence of a boot sequence.
Others have reported that this kind of behavior may be due to bad cabling (https://devtalk.nvidia.com/default/topic/1032498/serial-port-uart0-spews-garbage-how-to-use-it-correctly-/) so I tested again – same process, same cable – on some older devices (all 115200 8N1):
- TX2 -- Pass
- Raspberry Pi 3B+ -- Pass
Please advise, thanks.
If the serial UART is set up correctly with 115200 8N1, then possibly the character set could cause such a result. On both Jetson and host, what do you see for “echo $LANG”? Do the two differ?
One of the firmware guys from upstairs happened to stop by just now. I showed him the issue and he agreed that something was amiss. He came back with a proper six port FTDI TTL-232R-3V3 cable. That solved the problem, no further changes needed.
I have also faced similar problem…Using a prolific203 3.3V USB adapter, working fine for TX1 serial console, but failing to work on nano with J44 connector as shown by jetsonhacks…Checked the /proc/cmdline mentioned /dev/ttyS0 framing 115200 8N1.
I’ll try a FTDI adapter tomorrow…
May you describe further your six port FTDI cable ? Does it enable HW flow control ?
Thanks. It seems it indeed has cabling for HW flow control (RTS/CTS).
Not sure if it’s required, though, but it may make sense to give it try.