Delivering heat-free operation in AI applications up to 60 °C

Edge AI applications often mean that the hardware used tends to be housed in small enclosures, ducts or narrow boxes.

Edge AI controller with flat-top heat sink NRU-156U3-FT Credit: Acceed

While enclosures are unavoidable and are necessary to protect against environmental influences, air circulation and cooling can suffer.

The NRU-156U3-FT AI controller from Acceed provides a new fanless chassis design for installation in confined spaces with reduced airflow and simultaneous operation at high ambient temperatures.

The full-surface, flat heat sink maximises cooling efficiency across the entire top of the housing, enabling efficient heat dissipation even with reduced air circulation. This embedded PC has six USB interfaces (3.2) and is designed for operation at ambient temperatures from -25 to +60 °C.

The NRU-156U3-FT is equipped with the Jetson Orin NX for the targeted edge AI applications. Six USB 3.2 ports are available on the front for connecting cameras, for example. The special feature of the compact controller is its special flat heat sink (flat top). This new development significantly increases the cooling efficiency and makes the controller suitable for installation in closed housings, ducts and narrow installation locations without air circulation.

The controller in a compact all-metal housing uses the Jetson Orin NX, bundled with JetPack 5.1.1, and offers an inference performance of up to 100 TOPS with a power consumption of 25W.

The usual cooling fins have been replaced by a large heat sink mounted directly on the housing. The large heat-conducting surface of the flat heat sink efficiently dissipates the heat generated inside the device to the outside surface, allowing operation in a wide temperature range significantly expanding the possible application scenarios.

Other features include an M.2 socket Gen4 for connecting an NVMe SSD and standard serial interfaces. The DisplayPort video output supports a resolution of 3840 x 2160 pixels.

In addition to the hardware, the Nvidia system comes with a wide range of AI tools and the latest deep learning frameworks.