DFI X6-ORN NVIDIA Jetson Orin NX/Nano Fanless Edge AI Box PCs with PoE Option

Many edge AI projects start with one or two cameras, a PLC network and a cabinet with little spare room or airflow. The DFI X6-ORN series fits that brief: compact, fanless edge AI box PCs built around NVIDIA Jetson Orin NX and Orin Nano modules, with the networking and field I/O a machine builder expects already on the front and rear panels. Rather than a module on a carrier board that still needs an enclosure, connectors and a power stage, each model arrives as a finished system with the Jetson module, cooling and I/O in place.

All eight models share one design. The Jetson module brings a 64-bit Arm Cortex-A78AE CPU and an NVIDIA Ampere GPU with Tensor Cores, with LPDDR5 memory on the module and Ubuntu 22.04 supported. Four RJ45 GbE ports run from separate Intel i210 controllers, and four USB 3.2 Gen 1 ports, HDMI, a DB-9 with RS-232 and CAN bus, a DB-9 with four digital inputs and four outputs, and audio complete the I/O. Three M.2 slots cover NVMe storage, Wi-Fi 6E and 5G or LTE with a nano SIM, and out-of-band management is optional. The M-key storage slot runs on PCIe x4, and a micro USB port on the front is there for flashing a new OS image.

The X6-ORN range splits by module and by PoE. X6-ORN-NX16 and X6-ORN-NX16-PoE carry the Orin NX 16GB, the largest memory in the range, while X6-ORN-NX8 and X6-ORN-NX8-PoE keep Orin NX bandwidth at 8 GB. X6-ORN-NN8 and X6-ORN-NN8-PoE use the Orin Nano 8GB, rated at 40 TOPS, and X6-ORN-NN4 and X6-ORN-NN4-PoE use the Orin Nano 4GB, rated at 20 TOPS. On each PoE model two of the GbE ports supply up to 15 W to cameras or sensors.

Every X6-ORN model runs fanless from -20 to 60 °C on a 9 to 36 V DC supply, measures about 165 x 65 x 150 mm and mounts on a wall or, with the optional kit, a DIN rail. DFI tests the edge AI box PCs to IEC 60068-2-27 for shock and IEC 60068-2-64 for vibration, with CE, FCC Class A and RoHS. Typical installs include machine vision on production lines, smart retail, robotics and AI at the network edge.