Build A Compact Fanless Jetson NANO Embedded Systerm

Customers often have such questions: NVIDIA Jetson NANO development kit is so cheap, is it nessecary to use NVIDIA Jetson NANO module for production? The cost will be increased much.

When the mission is on the line, every detail counts. With altitude changes, severe climates, and extreme temperature changes a fact of life for aircraft, UAVs and ground vehicle in the field, every component needs to be able to perform regardless of environmental stresses. This is not what the Jetson nano development kit can bear. Only by using NVIDIA Jetson nano p3448-0002 module to solve such problems.

LCFC announced a compact fanless Jetson NANO embedded systerm EA-B100, which uses NVIDIA NVIDIA Jetson nano p3448-0002 AI computing moudle, providing 472 gflops computing power, and can easily run many neural network applications.

Note:the module operates at - 25 ° C to 80 ° C

EA-B100 adopts dust-proof fanless design, with fully enclosed casing heat dissipation, to meet the industrial dust-proof requirements in extreme environment.

The size is relatively small, similar to the size of the Jetson nano development kit.

EA-B100 has reached IP4X protection level, qualified for mobile devices (vehicles, robots, aircraft). LCFC also specially designed 4 shock cushions.

Interfaces:

Note: EA-B100 is shipped with wireless WIFI

Many developers who are familiar with NVIDIA Jetson product line may have a question, is that the NVIDIA Jetson NANO Module has only 16GB eMMc storage space, can it satisfy the use of application software?

Yeah， after installing Jetpack 4.4 + DeepStream 5.0, there is less than 500MB of free space left, which is not enough to install other application software. This is a very serious problem that affects the practicability of such products.

In order to solve this problem, EA-B100 has a very important design, which is to provide a TF expansion slot, and the expansion slot is very carefully placed under the dust-proof shield (as shown in the figure below) to get complete protection.

This design not only maintains the IPX4 protection level, but also solves the problem of insufficient storage of the Jetson Nano Module(P3448-0002), which is a very important feature of this product at present. Please choose the expanded TF capacity. It is recommended that 32GB or above.

However, after adding the TF card, can the additional application software be installed and executed smoothly? Those who are familiar with Ubuntu should be aware that many applications can be added not just by adding a TF card, because the main storage (16GB EMMC) is less than 500MB, and adding software like Tensorflow/PyTorch will probably burst.

In order to allow EA-B100 to add more applications normally, including installing Tensorflow, PyTorch, YOLO, Jetson-inferece, trt_pose and other Object Detection applications, and also downloading multiple NGC containers to run on EA-B100, We has tested as the following steps:

1. First format the 32G TF card into ext4 format, and mount it to the /media/temp directory

2. Move the contents of /home/lcfc (working directory) to the TF card

3. Modify /etc/fstab, mount the TF card as /home/lcfc at the next startup, and restart the system

4. After confirming that TF is mounted as /home/lcfc, move /usr/local to the /home/lcfc/local directory, and establish a local link to /home/lcfc/local under /usr (using sudo ln -s command)

5. At this time, 16GB EMMC storage can free up about 5GB space

6. Point the working directory of docker to /home/lcfc/docker, restart the docker service, and then you can use the docker image on NGC well

Now the 16GB EMMC has enough space to install more system tools, and there is about 20+GB space left on the TF card to install most of the applications.

Abput LCFC

For every 8 laptops sold in the world, 1 comes from LCFC.

LCFC Technology is the largest PC R&D and manufacturing base of Lenovo Group in the world. It provides customers with solutions including Lenovo and ThinkPad series notebooks, ultra-small computers, servers, intelligent IoT, and edge computing. The products are sold in 126 countries around the world. And region.

With such a production scale and supply chain management, LCFC can control the production cost of its products to the extreme, thereby making the product price extremely competitive.