Hardware

The hardware platform in a custom PCB mainboard with

• Nvidia Jetson AGX system-on-module

• two XTRX SDRs

• USB and HDMI ports

​

It is built on the hardware from the BlueDentist UCSB CE capstone project from 2019-2020.

​

​

​

• Board has been fabricated, and we are currently in the process of flashing the firmware and determining if there are bugs to work through. We flashed the STM32, which is in charge of controlling the power for the board. The next step is to flash the SOM, which controls the Jetson GPU and thus all of our peripherals on the board.

  • Some details regarding the flashing process:​

    • The STM32 flash process is mostly unchanged from last year. Our main steps were taking precautions to make sure all the power rails were safe to power and not shorted by design flaws, as well as powering the board without the SOM attached. This latter step required faking a signal from the SOM, which is part of the power sequence and necessary for the power to turn on. Now that we have found everything on the power end works as expected, we will flash with the SOM attached.​

• Our updates to the board from the original BlueDentist:

  • Main addition is a second mini PCIe component, which allows for the connection of a second XTRX software defined radio.​

    • Adding the second mini PCIe was a significant process, mainly due to the complex constraints regarding the mini PCIe's high speed signals and the dense nature of the board as we received it.

    • We removed the Ethernet component to make space for the second mini PCIe. The Ethernet did not work in the board design we received, and is also not needed for our intended use. We have other methods to send the data we receive to a UI. For example, UART is sufficient since we are not looking for fast real time updates. By removing the Ethernet, we had a perfect location to add the mini PCIe.

    • Although the Ethernet spot was the most ideal for the mini PCIe, we still had to rearrange many polygons and traces for other components to provide enough clearance for the high speed signals and limit the amount of vias per signal to two, as per the specifications.

    • Other necessary changes for the introduction of a second mini PCIe included finding the proper Jetson SOM signals to correspond with a new x2 PCIe, as well as creating a new schematic for the mini PCIe.

  • Another change is the removal of one HDMI port, and one USB connector. The HDMI was removed since only one is needed, and the second did not work. The USB was removed since one of the USB's was needed for the mini PCIe component. The removal of these two components allowed us to move the HDMI port to the top of the board, and the other two USB connectors to the bottom, and remove approximately 20% of the board on the right side.

  • Further changes involved fixing the first mini PCIe from the original design, which was using the incorrect clock port on the Jetson SOM. We rerouted to use the correct clock port to correspond to the data signals we used.