ADLINK NEON-1000-MDX Smart Camera Deployment and Trigger Configuration

Introduction: Integration Pain Points of Edge AI Vision Systems

In industrial testing, intelligent security, and embedded AI applications, the traditional combination of "camera+industrial computer" not only occupies space, but also needs to deal with complex cable connections and compatibility issues. The ADLINK NEON-1000-MDX series highly integrates the Intel Atom E3930 processor, Movidius Myriad X VPU, Basler image sensor, and FPGA I/O to create an intelligent camera designed specifically for deep learning inference. However, engineers often face specific challenges such as trigger signal configuration, lens selection, power topology, and software environment setup during actual deployment. This article will start with hardware interface parsing, triggering and I/O wiring, image acquisition, and OpenVINO inference examples to systematically outline the engineering implementation steps of NEON-1000-MDX, helping the project team quickly achieve a leap from prototype to mass production.

Product Model and Sensor Selection Guide

NEON-1000-MDX offers four models, with the core difference being the image sensor parameters:

Model Resolution Pixel Size Shutter Type Frame Rate Target Surface

NEON-101B-MDX 1280 × 960 (1.2M) Global Shutter 54 fps 1/3 "

NEON-102B-MDX 1600 × 1200 (1.9M) Global Shutter 60 fps 1/1.8 "

NEON-103B-MDX 1920 × 1080 (2M) Roller Shutter 30 fps 1/3.7 "

NEON-104B-MDX 2592 × 1944 (5M) Roller Shutter 14 fps 1/2.5 "

All models are color sensors based on Basler daA series modules and C-mount lens interfaces. Key selection points:

Moving object detection: Prioritize global shutter speed (101B/102B) to avoid the jelly effect caused by rolling shutter speed.

High resolution details: Choose 104B (5M), but the frame rate is only 14fps, suitable for static detection or low-speed motion.

Universal Balance: 103B provides 1080p 30fps, suitable for most monitoring and defect detection scenarios.

Detailed explanation of hardware interfaces and engineering wiring

The NEON-1000-MDX adopts a vertical I/O design, with all interfaces distributed on three sides, greatly simplifying installation and wiring.

3.1 Bottom interface (core connection)

DC Jack (12V/≥ 30W): Comes with a dedicated adapter (P/N 31-62156-1000-A0). It is recommended to use only official accessories to avoid insufficient power.

USB Type-C: Supports DisplayPort（ 1080p@30fps ）It is compatible with USB 3.0 and can also be used as a backup power input (via USB PD). When powered solely by DC Jack, Type-C can provide up to 3W (5V/0.6A) of power to the outside world for connecting small USB hubs or keyboards and mice. If the power consumption of the peripheral exceeds 3W, a Hub with independent power supply must be used.

Gigabit Ethernet: Standard RJ45, used for image data upload and remote control.

15 pin D-sub I/O: This is the most critical interface in engineering, including UART (Tx/Rx/GND), 4 digital inputs (DI0~DI3, where DI0 can be programmed as an external trigger), and 4 digital outputs (DO0~DO3, where DO0 can be programmed as a flash synchronous output). All DI/DO are grounded (DI/O GND), with a DI high level threshold of 2-5V (0-5V input range), and DO as an open drain output (requiring an external pull-up resistor, maximum current of 100mA, withstand voltage of 3.3-30V).

Typical application wiring:

Hardware trigger: Connect the external photoelectric sensor or encoder signal (NPN/PNP, level converted) to Pin11 (DI0/Trigger), set the camera to external trigger mode, and collect one frame for each falling or rising edge (configurable) received.

Flashlight control: Connect DO0 (Pin3) to the trigger terminal of the LED driver, and output synchronous pulses in conjunction with the camera exposure timing to achieve "capture and fill light".

3.2 Top and Front Interface

Power LED (green): It lights up when powered on.

System status LED (orange): Flashes at startup and stays on after the system is functioning properly, making it easy to quickly determine if the system is stuck.

MicroSD card slot: supports extended storage (ADLINK offers 32GB optional), which can be used to store model files and collect data.

Micro-USB（Type-A）：USB 2.0 OTG， Used for external mouse and keyboard connection or firmware upgrade (requires specific operations).

C-mount lens mount: Transfer Ring is pre installed, only NEON-102B requires a washer to ensure the rear focal length.

Power topology and heat dissipation precautions

NEON-1000-MDX supports two power supply modes:

Independent power supply: DC Jack connected to 12V adapter, Type-C is only used for video output and USB peripherals (power consumption limited).

Combination power supply: Use ADLINK dedicated Type-C Hub (P/N 92-99090-1010) to provide power, video output, and USB expansion simultaneously, reducing cables.

Key thermal management: The overall power consumption is less than 30W, but the built-in VPU (Myriad X) and CPU generate significant heat during high load inference. The manual stipulates that:

When using Type-C Hub for power supply, the upper limit of ambient temperature is 35 ℃;

When using the DC Jack adapter, the upper limit is 40 ℃;

If operation is required at higher temperatures, active air cooling must be provided (recommended 0.6m/s airflow, such as using a 40mm fan for side blowing).

If equipment overheating is found on the construction site (shell temperature>60 ℃), the ventilation conditions should be checked first, and external fans or reduced inference frame rate should be considered.

Software environment and image acquisition examples

NEON-1000-MDX does not come pre installed with an operating system at the factory (images need to be obtained through ADLINK support). The default system is Linux (Yocto or Ubuntu custom version), which integrates the Basler Pylon SDK and Intel OpenVINO toolkit.

5.1 Quickly Verify Images with Pylon Viewer

After powering on, enter the desktop through an HDMI monitor (via Type-C adapter) or connect remotely via SSH.

Open the terminal and execute Pylonviewer (located at/opt/nylon/bin).

Double click the sensor name (such as daA1600-60uc) in the GUI to see the real-time image.

Click on 'Continuous Shot' to start continuous collection, or use 'Single Shot' to capture a single frame.

5.2 Modifying Sensor Parameters

The "Features" panel of the Pylon Viewer allows adjustment of exposure time, gain, white balance, and trigger mode (free run/hardware triggered/software triggered). If you need to save custom configurations, export the. pfs file through Camera ->Save Features..., and later load it through the API.

5.3 Command Line Python Example

ADLINK provides three sets of Python example code (please contact technical support for assistance):

ImageCaptureOneShot.py: Single frame capture, based on the Pylon library.

ImageCaptureContinuus.sy: Continuous acquisition display.

ObjectDetectionYolov3.py: Combining OpenVINO's YOLOv3 model for real-time object detection (while calling the Pylon collection and inference engine).

Operation method:

bash

cd /path/to/samples/imageCaptureInference

python3 ./objectDetectionYolov3.py

The output window will display a real-time screen with a detection box. Press ESC or Ctrl+C to exit.

Trigger Mode Programming and I/O Control Practice

External triggering is the core requirement of industrial vision. The following are the hardware trigger configuration steps:

Hardware connection: Connect the trigger signal (5V logic level, or use optocoupler isolation) to Pin11 (DI0) and share ground with DI/O GND (Pin2/10).

Set Trigger Mode: In the Pylon Viewer, set the sensor's "Trigger Mode" to "On" and "Trigger Source" to "Line1" (corresponding to DI0). At this point, the camera is waiting for an external pulse.

Trigger polarity: You can choose Rising Edge or Falling Edge and adjust it according to the output of external sensors.

Software trigger: If instructions are issued through the upper computer, the ExecuteSoftwareTrigger function in the Pylon API can be used, which is suitable for pipeline synchronization.

Flash synchronization: Connect the LED controller input to DO0 (Pin3) and configure "Strobe Enable" and "Strobe Delay/Width" in the sensor settings to output a high level during exposure (external pull-up resistor to 24V or 12V is required). Note that DO is open drain and must be externally pulled to generate high-level output.

UART communication: There is also UART (Tx/Rx/GND) on the D-sub, with a baud rate of 115200 8N1, which can be used for simple serial communication with external PLCs or sensors, facilitating the transmission of detection results or status information.

Key points for lens selection and installation

Compatibility: Only C-mount lens (rear focal length 17.526mm), does not support CS mount. Incorrect installation may result in inaccurate focal length.

Gasket usage: NEON-102B (1/1.8 "sensor) requires an accompanying gasket to be installed between the lens and the adapter ring to ensure optical back focus. Other models do not require gaskets.

Lens weight: The total weight of the equipment is less than 700g (excluding the lens). It is recommended to use a lightweight lens (less than 200g), as excessive weight may affect the stability of fixation.

Installation screw holes: M3 threaded holes are provided on both the front and rear sides, which can be flexibly installed on the bracket or the end of the robotic arm.

Common troubleshooting and engineering tips

8.1 Equipment unable to recognize or image black screen

Check if the USB Type-C or DC Jack is powered properly and if the power LED is green.

Is the system status LED always orange? If it flashes or goes off, the system may not have fully started and the system image needs to be flashed again.

Has the lens cover been removed? Check if the correct exposure time is set (too short will result in complete darkness).

8.2 Invalid External Trigger

Confirm that the sensor Trigger Mode is set to "On" and the Trigger Source is "Line1".

Use an oscilloscope to measure whether Pin11 has an effective level jump (0-5V) and the polarity setting matches.

Check if DI/O GND is grounded with external devices.

8.3 Insufficient reasoning performance (frame rate lower than expected)

The inference frame rate is constrained by both CPU and VPU, which can reduce input resolution (such as from 1600 × 1200 to 640 × 480) or use lighter models (such as MobileNet SSD).

Monitor CPU temperature, if it is too high (>70 ℃), it will trigger frequency reduction and strengthen heat dissipation.

8.4 Insufficient storage space

32GB eMMC has been used for the system, and it is recommended to store the model and data on a microSD card and mount it in the startup script.

8.5 USB Type-C peripherals cannot be used

When only powered by DC Jack, the Type-C output power is limited to 3W, and high-power hard drives or hubs need to come with their own power supply.

Maintenance and long-term operation suggestions

Regularly clean the lens and sensor: use professional grade air balloons and microfiber cloth to avoid scratching the infrared cut-off filter.

Firmware upgrade: ADLINK may release new firmware to optimize VPU performance or fix bugs, which needs to be burned through the Micro USB port (refer to the "System Flash" section of the manual).

Battery Attention: There is a CR2032 battery inside to maintain RTC, with a lifespan of about 5 years. When replacing, pay attention to the model to avoid the risk of explosion.

Environmental protection: This device is not designed for IP protection and requires an additional protective cover in dusty or humid environments.