https://community.element14.com/challenges-projects/design-challenges/experimenting-with-single-pair-ethernet/Experiment with Single Pair Ethernet to learn its benefits for industrial automation. Ethernet is the most popular communication protocol today for residential and commercial applications. It's also widely used in industrial automation. Traditional Ethen-USTelligent Community 12https://community.element14.com/challenges-projects/design-challenges/experimenting-with-single-pair-ethernet/b/news/posts/winners-announcement-experimenting-with-single-pair-ethernet-challengeFri, 10 Apr 2026 14:54:00 GMT93d5dcb4-84c2-446f-b2cb-99731719e767:1ed0fbd6-ba51-4571-9667-4e32e254636eJoRatcliffeFirst up, a big thanks to everybody who has kept a watch on this challenge, commented, and participated. Let's quickly recap the prizes and the challenge, then reveal the winners... The Prizes Prize Prize Category Grand Prize Winner MULTICOMP PRO MP720025 US Digital Oscilloscope, 4 Analogue, 100 MHz, 1 GSPS, 40 Mpts, 3.5 ns MULTICOMP PRO MP710082 US Bench Power Supply, Adjustable, 1 Output, 500 mV, 36 V, 0 A, 5 A Runner-Up Prize Approximate Value $275 USD * MULTICOMP PRO MP720857 Handheld Oscilloscope, 2+1 Channel, 100 MHz, 500 MSPS Approximate Value $105 USD * TENMA 72-2660 Bench Power Supply, Handheld, Adjustable, 1 Output, 300 mV, 30 V, 0 A, 4 A Finisher Prize** For anyone who completes a final summary blog with the featured product. Approximate Value $48 USD* MULTICOMP PRO MP015203 Repair Tool Kit, Smart Phone, 4mm Driver, Tweezers, Spudger, Pliers, Opening Pick, Screwdriver Bits *Or local equivalent **Grand Prize and Runner Up winners will also earn the finisher prize. The Challenge Four participants received a kit of Single Pair Ethernet components. For the chance to win a prize, participants had to blog and document a project which demonstrated creative, interesting uses of the components, including Molex’s SPE connectors and cables. Now without further ado, here are our Single Pair Ethernet Challenge winners! Grand Prize Winner vmate - Advanced Dashcam and Monitoring System vmate created an always-on monitoring and dashcam system for their car, streaming the camera feed in real-time to a remote server. The project proved its value very quickly when - during the build period - the cameras recorded someone backing into their parked car and driving away. Runner-Up Prize Winner JWx - Line powered SPE IP camera JWx conducted a series of experiments to put the kit through its paces and investigate power delivery possibilities. A huge congratulations to both of you for two fantastic projects! My colleague E14Alice will be in touch to arrange the shipment of the prizes. Well done also to veluv01 and Ben5049 for taking part in this challenge, I will be sharing a roundup post next week highlighting all four projects.twin pair ethernetdesign challengemolexsingle pair ethernetethernethackathonip67rpidcspehttps://community.element14.com/challenges-projects/design-challenges/experimenting-with-single-pair-ethernet/b/projects/posts/yolov3-object-detection-over-10base-t1l-speFri, 10 Apr 2026 09:51:00 GMT93d5dcb4-84c2-446f-b2cb-99731719e767:c7ccfd7a-f41c-4777-a86e-0d2a5a8723a0veluv01Introduction I have been wanting to do vision based projects for a while now, but there was one problem that kept getting in the way. The dev board needs to sit in a remote area, which in my case is quite far from the actual field where the camera needs to be mounted. When I tried using a standard camera link cable to connect them, I quickly ran into its length limit. Camera link cables are just not designed to run long distances, and beyond a certain point the signal starts degrading badly. So the camera could not reach where it needed to be, and that killed the whole idea before it even started. Around the same time I came across the Single Pair Ethernet design challenge, and it felt like the perfect opportunity to give it a go and see how well it actually works. Molex Cables and Cable Assemblies I've used the Molex's Industrial SPE connector and cable assembly lineup, specifically the 220957 Series, rated IP67; especially for the industrial environments for the SPE connections. These are the specific part's I've used in my setup: 1. 220957-0008 2. 220957-0116 3. 220957-0011 4. 220957-0012 Wiring up the Molex's SPE PCB Jack Mounting the PCB jack requires a specialized footprint and it's hard to mount on to the generic perfboard. I began by connecting the wires, ( shorter ones, which are salvaged from a Cat6 cable) to the Molex's SPE PCB jack. I'm not entirely sure if this is the right way to connect wires to this vertical jack . One thing that stood out during soldering was how well the jack's plastic housing held up; despite direct heat from the soldering iron, there was no melting or deformation of the plastic support of the SPE jack. I initially included the shielding wire attached to the connector, but after testing both configurations, the difference in performance was negligible. This was probably due to the minimal EMI noise in my testing environment and the cable length is only 1 meter, which is short when compared to the 10BASE-T1L standard's maximum reach of 1 kilometer. Molex's PCB jack spec's are interesting, supporting higher data rates but the ADIN1100 and ADIN1110 are only 10BASE-T1L compliant and the max data rate of the link is limited to 10Mbits / sec. Attaching the Molex M12 Receptacle Shell Mount So for my actual installation, I ended up mounting the shell mounts onto transparent acrylic sheets as they are provide a better view of everything inside the assembly and also part of the reason being I couldn't find proper enclosures big enough and with the proper IP rating. Now in terms of sealing and protection, the rubber washers protect against dust and moisture coming in from the outside. The inner washer makes sure that the M12 cable assembly is cinching up tight and stays that way despite the vibrations, it provides IP67 rating only in mated condition along with the Molex's Plug Cable. Here's the Front Mount dimensions and how the PCB mount fits into it. Here's the Back Mount dimensions and how the PCB mount fits into it. Here's some close up shots of the SPE mounts in my setup, the first one is the back mount and the second one's the front mount attachment. The back mount cannot be adjusted easily (i.e removed or tightened) without opening the enclosure, a more secure option when used on the outdoor units.These SPE mounts are made from brass (handles mechanical stress well and doesn't corrode easily on its own) and plated with Nickel (extremely resistant to oxidation, moisture, and the kind of atmospheric corrosion in outdoor environments). Connecting using the Molex's T1 SPE Plug Cable Assembly I gotta admit that this is my favorite cable in the kit, there's many reasons apart from it's design. First, just like all the Molex parts provided in the kit, this cable also handles up to 60V AC and 4 amps, which is quite impressive for the SPE's PoDL applications. Second, it has wide operating temperature range enough to use it in freezing winters or hot summers without cracking or degrading. The connector is also rated for 1000 mating cycles which is plenty for a semi permanent industrial installation. For this design challenge specifically, I am starting out with a shorter Molex cable rather than going straight to a long run. The reason for this is straightforward. A shorter cable is easier to work with during the prototyping and testing phase, it is simpler to troubleshoot if something is not behaving as expected, and it still lets me validate the full system end to end. The top part i.e the M12 screw attachment (labelled as 2 in the drawing) is the only part that rotates to screw into the front/back mount There's also an arrow provided on it to properly orient the connector during the connection of SPE cable and the mount. The cable itself is PUR(Polyurethane), jacketed in green, which is the standard color coding for SPE industrial cables, making identification easy in a busy panel or installation. The M12 nut, hex nut and body shell are all brass with nickel plating, which protects against rust and corrosion in outdoor and humid conditions. Another interesting detail worth noting is that the copper contacts inside the cable assembly are gold plated which is a meaningful design choice since it'll have better corrosion resistance over time, and more reliable signal integrity compared to standard tin plated contacts. Setting up the Video Stream The Raspberry Pi 4 acts as the video stream for the YoloV3's input stream. I connected the RPI Camera V1 to the RPI4 and added the CN0575 HAT, which is connected to KR260 via the Molex SPE cable. Because the ADIN1110 and ADIN1100 has a low speed limit of 10 megabits per second (as they support only 10Base-T1L), I knew I had to keep my video feed very small enough to fit within the limited bandwidth. To fix this, I set the camera to a square resolution of 416x416. I chose this size because it matches the exact needs of my YOLOv3 model, meaning no bandwidth goes to waste on extra pixels and I do not have to resize the frames later. raspivid -o - -t 0 -w 416 -h 416 -fps 30 -n | gst-launch-1.0 fdsrc ! h264parse ! rtph264pay config-interval=1 pt=96 ! udpsink host=169.254.158.50 port=5000 To send the stream, I piped the video(30 frames per second and compression in H.264 format). from the camera directly through the GStreamer, as it offers the lowest latency. Also included a setting that sends a signal every second so the KR260 can tune into the stream whenever it starts up. The video stream is streamed straight to the KR260 using a UDP network connection, which is best for live, low-latency video. After the stream is setup, I also checked it on the KR260 if it's able to receive the stream properly And the KR260 is able to receive the stream without any issues. Setting up the KR260 I found the AMD Kria KR260 through DevKit HQ and it was a straightforward pick for running YOLOv3 through the DPU IP over the SPE network. Ubuntu 22.04 on KR260 For this project I am using the Ubuntu 22.04 instead of the 24.04 version which is available as the latest version as I faced some issues due to the Python version mismatch during the startup when I set the default interpreter to python3.10 as the VART 3.5 requires it. For the setup I referred the guide " Booting Kria Starter Kit Linux on KR260 ", which has all the information regarding the basic setup of the Ubuntu 22.04 for KR260.It's also better to choose an SD with a storage capacity of 64 GB max for flashing the Ubuntu 22.04. I used the Raspberry Pi Imager to flash the Ubuntu 22.04 Image onto the SD card. Initialization Before deploying AI workloads or running PYNQ overlays on the KR260, the board must be configured using AMD’s official initialization tools.This process prepares the operating system, sets up FPGA-related services, and installs board support packages to ensure that the runtime and PYNQ framework operate correctly. The first step is to install the xlnx-config tool, which AMD provides through the Snap store. This utility automates KRIA-specific system configuration tasks, such as enabling FPGA management services, installing boot-time utilities, and preparing system directories used by XRT and runtime components. Installing the tool with the --classic option allows it to access the system-level paths needed for hardware configuration. sudo snap install xlnx-config --classic --channel=2.x After installation, the board is initialized using the sysinit routine. xlnx-config.sysinit Setting Up PYNQ for Kria PYNQ(Python Productivity for Zynq) is an open-source framework from AMD/Xilinx that makes it easier to work with Xilinx's reconfigurable hardware, like the FPGA fabric on the Kria KR260. To enable a Python-based workflow on the KR260, AMD provides the Kria PYNQ framework. This repository includes board overlays, configuration scripts, Python drivers, and prevalidated examples.Cloning the repository ensures that the latest PYNQ board files and acceleration libraries are available locally for the setup. git clone https://github.com/Xilinx/Kria-PYNQ.git The final step is to run the installation script, specifying the target board. The install.sh script configures Jupyter, installs required Python packages, deploys PYNQ overlays, and sets up runtime services. Using the "-b KR260" flag ensures that the installation process applies the correct board-specific settings, including kernel modules and device-tree overlays. sudo bash install.sh -b KR260 Setting up the Vitis AI 3.5 Runtime Deploying AI workloads on the AMD/Xilinx KR260 requires the Vitis AI Runtime (VART). This runtime serves as the main execution engine for running quantized deep-learning models on the KR260’s DPU accelerator. The VART stack includes libraries for tensor processing, model graph handling, logging, and device abstraction. All of these need to be installed before running any inference workloads. The Vitis AI runtime for KR260 is offered as a compressed ZIP file from AMD’s public download server. This package contains all the necessary libraries, device-specific binaries, and support tools. wget -O vai3.5_kr260.zip "https://www.xilinx.com/bin/public/openDownload?filename=vai3.5_kr260.zip" Once downloaded, the compressed archive must be extracted. unzip vai3.5_kr260.zip The runtime_deb directory holds Vitis AI's core runtime .deb files. These include important components like libunilog, libxir, libtarget-factory, libvart, and libvitis-ai-library. Together, they make up the VART execution layer. AMD provides a script called setup.sh in this directory to install these packages in the right order and ensure that earlier versions are upgraded properly. pushd vai3.5_kr260/target/runtime_deb/ bash setup.sh This script automatically installs all the needed dependencies and configures the system for Vitis AI inference. The KR260 Ubuntu image might lack certain low-level shared libraries needed by the Vitis AI runtime. AMD provides these in a compressed file named lack_lib.tar.gz. Extracting and copying them into /usr/lib fixes any missing dependency problems that could cause VART applications to fail at runtime due to unresolved symbols cd .. tar -xzf lack_lib.tar.gz sudo cp -r lack_lib/* /usr/lib After installing the runtime and any missing libraries, it’s helpful to return to the original working directory using the popd command. This step navigates back to the main vai3.5_kr260 folder, preparing for the installation of utilities like xbutil2. popd The KR260 needs a functioning xbutil2 utility for system inspection, FPGA health checks, and DPU diagnostics. This tool checks if the accelerator is properly initialized and if the Vitis AI runtime environment is ready for use. The utility is located in the xbutil_tool directory and must be copied to the correct system binary location (/usr/bin/unwrapped/) for the Ubuntu to recognize it. sudo cp ./xbutil_tool/xbutil2 /usr/bin/unwrapped/ After installation, we can verify system health using sudo xbutil2 examine DPU-PYNQ Installation and Setup DPU (Deep Learning Processing Unit) is a specialized soft-core processor built on the FPGA's programmable logic. It is an IP (Intellectual Property) core from AMD/Xilinx, designed to carry out the mathematical operations of a deep neural network, such as convolution, pooling, and matrix multiplication, with high speed and power efficiency. To enable Vitis AI 3.5 acceleration on the Kria KR260 running Ubuntu with PYNQ, we need to install the DPU-PYNQ framework manually. First, clone the official repository and switch to the branch for Vitis AI 3.5. This ensures compatibility with the updated DPU runtime, overlays, and metadata. After cloning the repository, navigate into the new directory to prepare for installation. git clone https://github.com/Xilinx/DPU-PYNQ cd DPU-PYNQ Before installation can start, activate the PYNQ virtual environment (pynq-venv). This environment includes all required Python libraries like pynq, pynqmetadata, and pynqutils. Using the correct virtual environment is crucial since the system Python does not include these libraries. Activating the environment also ensures the installation targets the correct site-packages directory. $ source /etc/profile.d/pynq_venv.sh Once inside the venv, install the DPU-PYNQ package using the venv's Python binary, not the system Python.It's needed to use $ sudo -E to keep the environment variables active, allowing the venv to remain active even with elevated privileges. The flag --no-build-isolation ensures that the build uses the already installed PYNQ dependencies instead of creating temporary build environments. sudo -E /usr/local/share/pynq-venv/bin/python3 -m pip install . --no-build-isolation After installation, download all example notebooks included with the DPU-PYNQ package. These notebooks consist of inference tests, model runners, profiling tools, and Jupyter resources. cd pynq get-notebooks pynq-dpu -p . --force Next, apply an important system level fix. The Vitis AI runtime installs key shared libraries, such as libvart, libunilog, and libxir, in /usr/lib. However, the PYNQ virtual environment doesn’t look in /usr/lib by default. It's needed to extend the LD_LIBRARY_PATH in the venv activation script to make these libraries available during runtime using the below command $ sudo sed -i -e '$aexport LD_LIBRARY_PATH=/usr/lib' /etc/profile.d/pynq_venv.sh The xdputil tool, which inspects DPU configuration, model metadata, and performance statistics, hard codes a reference to /usr/bin/python. This makes it run outside the PYNQ virtual environment and breaks imports. To fix this, remove all absolute python path strings so it uses the currently active python i.e the venv python using the below command $ sudo sed -i "s/\/usr\/bin\///g" /usr/bin/xdputil Some Vitis AI Python bindings are installed in the global system path (/usr/lib/python3.10/site-packages). These modules include components needed by VART. To ensure the venv can import them correctly, extend the PYTHONPATH in the PYNQ environment activation script using the command $ sudo sed -i -e '$aexport PYTHONPATH=/usr/lib/python3.10/site-packages:$PYTHONPATH' /etc/profile.d/pynq_venv.sh Now, that all basic setup for the DPU is complete, I generated the .xmodel file for the YoloV3 using the Vitis AI Docker Image ( the CPU based one, the GPU based one didn't work) by quantizing and pruning the YOLOv3 PyTorch model. I obtained the coco.names file with all the object labels from the official darknet repository. Putting it all together and testing it out My final setup looks like this, After setting up the hardware and powering it on, I've configured the SPE on 169.254.x.x link-local subnet and also tested the connection via ping and it works without any issues. The RPI4 camera stream travels over the 10BASE-T1L link and arrives at the KR260 over the 169.254.x.x link-local subnet that is set up on eth1 interface of the KR260 is connected to the SPE via the ADIN1100 Media converter. Next, it's time to run the YoloV3 Object detection on the KR260 since the SPE network is setup, The FPS was a bit low as since I'm using Python but the Yolov3 Object detection worked, Initially I've used my keyboard and a water bottle to test if the model can identify them and it does without any problem. And I've tested if it can identify both the keyboard and the water bottle within the same frame, and it works!ADIN1100molexsingle pair ethernetKR260dpuraspberry pi 4Molex CablesADIN1110https://community.element14.com/challenges-projects/design-challenges/experimenting-with-single-pair-ethernet/m/managed-videos/151168Sat, 04 Apr 2026 00:38:00 GMT93d5dcb4-84c2-446f-b2cb-99731719e767:49016680-f73a-4b84-a096-a8d6c602557evmate