TX1 J21 Breakout Boards: What Would You Like?

This is a bit of a poll or survey for people with TX1’s J21 connector.

EDIT: I’m making this “sticky” for a little while, hoping people can see and answer what they want to make their TX1 development better. Feel free to add a reply and suggest multiple breakout board features or wants…or just explain in plain words what would make development work faster or better for you.

Connector J21 has the serial console wiring…if you were to purchase a cable adapter or breakout board for J21, and if that adapter provided a number of functions, what would you want for the serial console?

Possibility 1: A standard 6-pin header fitting directly to a 6-pin serial UART cable (e.g., similar to http://elinux.org/Jetson_TX1#The_Example_USB_Serial_UART_Assembly). This would be the least expensive, and work directly with a 6-pin standard serial UART connector (3.3V I/O).

Possibility 2: A standard 9-pin D-sub connector. This runs 12V and would require level conversion, plus the more expensive and physically larger connector. This would be significantly more expensive, but allow the old-fashioned NULL modem cable to work directly as serial console.

Which of those choices would you be interested in?

Digital audio I/O also goes to this J21 connector. Would there be interest in an audio breakout board?

How much interest would there be in an optical TOSlink port capable of 8 channels (7.1 surround sound)?

How much interest would there be for an analog stereo line out plus microphone breakout for J21? This would be more expensive than an optical TOSlink.

How much interest would there be for a full 7.1 channel surround sound breakout? This would be expensive compared to a simple optical TOSlink.

NOTE: Any breakout board would pass through remaining unused pins to make the original 40-pin (20x2) connector available and would not block use of those other pins.

Hi linuxdev,

I would like the same interfaces that the TK1 has, namely a 9-pin D connector for a null modem cable, and an analog stereo out and microphone in.


If a PCB will be made, I’d prefer a USB/UART bridge chip e.g.


and a micro or mini USB connector. The components are cheaper than PCB.

I have made few of USB/UART boards for ARM console debugging, it’s not hard to solder CP2103 to PCB.

CP2103 has Linux and Windows drivers.

This is actually being considered, but time to make it available in bulk would go up, along with price. If there is enough interest in breakout boards there is a possibility of more than one version, with the simplest becoming available first.

I use this interface with another non-Jetson Tegra system, and like it. Keep in mind that the complication and expense of the serual USB UART would change from the part plugging into the host to the part plugging into the Jetson…basically that expense would not go away, it would just transfer to which part it applies to.

If you want to do the perf board (PCB), I would suggest the design should include a switch for 4-channel I2C-safe Bi-directional Logic Level Converter - BSS138. I voted for the UART 6 pins for serial interface. That would be really helpful when I try to debug Jetson without USB/Keyboard/HDMI interface

To clarify, the existing J21 serial console port uses 3.3V, and if a simple 6-pin serial connection is added, the level converter would be for using different logic levels on the connector (other than just 3.3V)?

That is correct, it can be for example switch to 1.5V. I think the information I compiled before was for TK1 http://jetsonhacks.com/2015/09/03/level-shifting-uart-and-gps-part-1-nvidia-jetson-tk1/ but again it’s nice to have a switch that can drive into different low voltages output

Any chance of a DisplayPort MST connector to enable multiple monitors (up to 3 with a single DP connector)? I believe the Jetson module has all the signals required, so it should be a simple replacement of the HDMI connector. Cables and splitters are readily available for HDMI and DVI.

I’m sure a DisplayPort adapter would be quite popular…unfortunately stringent requirements for building such an adapter make the job a lot tougher than for lower speed signals. Add a need for an expensive membership in vesa.org for documentation and it just doesn’t seem practical at a reasonable cost.