AIR-020X: AI Inference System with NVIDIA Jetson Xavier NX
AIR-020XStarting at $1,600.00
- Extreme Compact and performance AI box up to 21 TOPS AI computing
- NVIDIA Jetson Xavier NX built-in
- 12~24V wide power and -10~55 °C wide temp. supported
- Multiple IO ports: Dual LAN, DIO, 2x COM, 2x USB 3.2 and USB type C
- M.2 2280 128GB storage built-in
- Edge AI Suite AI Utility Support for pre-trained models and deep learning
- Linux Ubuntu 18.04 LTS and JetPack 4.5.1 preload
- IP40 Design compliance
- Qualified for Edge AI SRP of WISE-DeviceOn
Frequently Asked Questions
The AIR-020X includes the following ports:
- 2 x Gigabit Ethernet (RJ-45, Intel i210AT controller, 10/100/1000 Mbps)
- 1 x HDMI (max resolution 3840 x 2160 @ 60Hz)
- 2 x USB 3.2 Type A
- 1 x USB 3.2 Type C
- 1 x Micro USB (OTG, for system recovery only)
- 1 x DB9 CANBus port
- 8-bit DI/DO
- 2 x RS-232/RS-422/RS-485 (COM)
- 1 x full-size Mini PCIe slot with Nano SIM slot (USB/PCIe signal)
The AIR-020X accepts 12~24VDC wide-range power input. It supports ATX/AT mode with ATX as the default setting. An optional 65W power adapter (19V) is available.
The AIR-020X operates from -10 to 55 °C with 0.7 m/s air flow (non-throttling). It supports operating humidity up to 95% at 40 °C (non-condensing) and is rated for vibration of 3 Grms at 5-500 Hz (random, 1 hr/axis).
The AIR-020X is CE/FCC Class B, CB, UL, CCC, and BSMI certified. It does not have RED certification. It also carries an IP40 design compliance rating.
The AIR-020X includes 8 GB LPDDR4 memory, 16 GB eMMC onboard flash, and a built-in 128GB M.2 2280 (M Key) storage module.
The AIR-020X ships preloaded with Linux Ubuntu 18.04 LTS and NVIDIA JetPack 4.5.1. It also supports the Edge AI Suite software API and is FaceView compatible.
The AIR-020X measures 138 x 110 x 43.6 mm (W x D x H) and weighs 0.85 kg.
The AIR-020X supports wall mounting. An optional wall mount kit (AMK-W005) is available.
The AIR-020X is built on the NVIDIA Jetson Xavier NX module with an ARM v8.2 six-core Carmel processor (max 1.9 GHz) and a Volta GPU with 384 CUDA cores and 48 Tensor cores. It provides an AI performance reference of 21 TOPS (INT8).
How-to Guides
- Obtain the optional AMK-W005 wall mount kit.
- Refer to the wall-mount mechanical drawing on page 2 of the datasheet for hole placement and orientation.
- Secure the unit to the wall using the wall mount kit per the dimensional layout provided.
- Confirm your power source supplies 12~24VDC, matching the AIR-020X power input rating.
- Use the DC IN connector located on the rear panel I/O layout.
- Optionally, use the 65W (19V) power adapter (part 96PSA-A65W19P2-1) with the appropriate regional power cord (US, EU, UK, AU, Japan, or China).
- Set the power type jumper/mode to ATX (default) or AT as required by your installation.
- Locate the Micro USB (OTG) port on the front panel, labeled for system recovery only.
- Connect the included 40cm Micro USB recovery cable (part 1700028866-01) between the AIR-020X and a host PC.
- Follow the system recovery procedure using this connection (see user manual for full recovery steps).
| Manufacturer | Advantech |
|---|---|
| Short Description | AI Inference System Based on NVIDIA Jetson Xavier NX |
| Processor | ARM v8.2 Six core Carmel processors, Max. 1.9 GHz |
| Graphics | Volta 384 CUDA core and 48 Tensor cores |
| Memory | 8 GB LPDDR4 |
| Video Port | 1 x HDMI (Max. resolution 3840 x 2160 @ 60Hz) |
| DIO | 8 bit |
| M.2 | 1 x M.2 2280 (M Key), 128GB storage built-in |
| Power Adapter | Power adaptor 65W, optional |
| Power Type | ATX/AT mode, ATX default |
| Fan | Fanless |
| Dimension | 138 x 110 x 43.6 mm |
| Weight | 0.85 kg |
| Mounting | Wall mounting |
| Operating Temperature | -10 ~ 55 °C with 0.7 m/s air flow (non-throttling) |
| Humidity | 95% @ 40 °C (non-condensing) |
| Vibration | 3 Grms @ 5 ~ 500 Hz, random, 1 hr/axis |