Hi，
I am doing some work with multiple TK1 and I need to transfer data between them.
I’ve decided to use UART to transfer data.
The wiring is done by directly connect pin 68(TX) to 65(RX), 65(RX) to 68(TX), ground to ground ( pin 45 )
Then I test the connection with minicom
I’ve type the command “minicom -b 115200 -o -D /dev/ttyTHS1”
but both side don’t works( nothing appearing when typing words ).
Later, I’ve test the ttyTHS0 with changing wiring between pin 41 to pin 44, pin 44 to pin 41
the minicom test success( two side can display what I’ve type ).
For the third try,I’ve connected ttyTHS0 and ttyTHS1
but only one side result ( one side can display, but not the otherside )
Is there any connection problem with my wiring? or there is damage on the circuit?
Thank you very much.
Which connectors are the pins referring to? How long are the TX/RX wires, and are they twisted pair and/or shielded?
Hi linuxdev,
I use dupont line to connect, it just about 10cm long, and single core.
pin 68 65 are refering to ttyTHS1
and pin 41 44 are refering to ttyTHS0
I found another problem, when I connecting another pair of TK1 with ttyTHS0
( pin 41 <-> pin 44, pin 44 <-> pin 41 )
( I have 3 TK1 for testing, so I can change the combination )
The result become one side only…
for 68 <-> 65, 65 <-> 68 it still don’t work…
so from different combination of TK1, I get different result with same wiring…
- I have tried self loop of of ttyTHS1 for my three TK1, which directly connect rx pin to tx pin ( 68 and 65 )
All of them won’t work…
but self loop for ttyTHS0 works…
I have not tried what you are doing, but it seems reasonable. I have a lot of connectors to remember, I’m hoping to verify schematic connector name of the Jetson.
Usually the biggest cause of serial not working on ports where the hardware is available (e.g., no pinmux issue or level shift issue) is just the serial port setting, e.g., normal serial console is set to 115200 8n1, and alternate settings would either not show anything or show as gibberish. Make sure to set both ends to “software flow control” (the CTS/RTS->RTS/CTS could be connected and you could use CTS/RTS flow control, but at short cable lengths I doubt it would matter unless the software depended on this).
If you have an oscilloscope then you can view the RX/TX pairs and see if hitting a key causes output just to verify each side’s TX functions (and thus narrow the error to TX or RX).
10 cm is short and should not be a problem, although I’d still suggest twisted pair as opposed to straight and parallel wires. Shielding would not matter at that length unless you are next to an intense RF noise source.
Hi, I found that after setting hardware flow control off( in minicom ), the connection work well!
I’m still wondering why ttyTHS1 need to turn off hardware flow control…
But anyway,thanks very much for help!!!
Software flow control will always work if the signals are sufficient quality. Hardware flow control will work when signals degrade…but that requires something like the CTS/DTS lines being connected and complicates wiring (not needed for short cables, but a good practice in noisy environments and longer cable runs).
If you cut the wires for hardware flow control (and extra wires are mandatory for non-software-flow-control), then you can no longer use hardware flow control. Not implementing those wires is equivalent to cutting them with wire cutters :)