ADLINK NEON-1020/1040 Smart Camera

ADLINK NEON-1020/1040 x86 Smart Camera Hardware Architecture and Industrial Vision Deployment Guide

In the field of high-speed industrial visual inspection and intelligent sensing, traditional intelligent cameras are often limited by the computing power of ARM architecture or FPGA with fixed functions, making it difficult to balance high frame rate image acquisition and complex algorithm processing. NEON-1020 launched by ADLINK（ 2MP@120fps ）Compared to NEON-1040（ 4MP@60fps ）The new generation x86 intelligent camera is based on Intel Atom quad core processor E3845 (1.91GHz), combined with FPGA co processor and GPU acceleration, IP67 protection level and M12 industrial connector, providing a high-performance and high reliability solution for harsh scenarios such as robot guidance, 3D vision, medical imaging, and machine tool processing. This article is based on the official data manual and provides a deep analysis of its hardware architecture, I/O configuration, software ecosystem, and deployment points, helping visual engineers comprehensively evaluate and quickly implement platform migration.

Product positioning and core differences

The NEON series consists of two models, with the main difference being sensor resolution and frame rate:

Model Sensor Frame Rate (Global Shutter) Applicable Scenarios

NEON-1020 2MP (1920 × 1080 approximately) 120 fps high-speed motion object detection, fast sorting, barcode tracing

NEON-1040 4MP (approximately 2592 × 1944) 60 fps high-precision size measurement, surface defect detection OCR

Both use global shutter CMOS sensors (monochrome) to ensure no jelly effect during high-speed motion, and the hardware platform is completely identical, only the sensors are different. This design allows developers to flexibly choose a balance between resolution and speed within the same software framework.

Core computing hardware architecture

1. Processor and Acceleration Unit

Intel Atom E3845: quad core, 1.91GHz, 64 bit, supports Intel VT-x virtualization and AES-NI encryption instructions, with a TDP of only 10W, suitable for fanless enclosed chassis.

FPGA coprocessor: responsible for image preprocessing (such as bad pixel correction, gain/offset adjustment, ROI cropping, pixel merging), freeing up CPU burden.

Integrated GPU (Intel HD Graphics): Supports OpenCL 1.2, which can accelerate parallel algorithms such as filtering, edge detection, and morphological operations.

Memory and Storage: Onboard DDR3L (capacity not specified, but usually ≥ 4GB), up to 32GB SSD storage (available through optional configurations such as NEON-1020/SSD16G), used for program, image caching, and log archiving.

2. Image sensor interface

The sensor is connected to the FPGA through a high-speed serial interface and supports external triggering and flash synchronization. The global shutter design ensures that all pixels are exposed simultaneously, which is crucial for moving workpieces or vibrating environments.

Industrial grade I/O and scalability

The NEON series provides a wide range of external interfaces, all of which are waterproof and dustproof achieved through M12 circular connectors, meeting the IP67 rating:

4-channel isolated digital input: can be connected to photoelectric sensors, proximity switches, encoder Z-phase pulses, etc., supports 5-24V levels, and has surge protection.

4-channel isolated digital output: can directly drive small solenoid valves, indicator lights, or solid-state relays, with a typical output current of 100mA and a withstand voltage of 30V.

1 additional GigE Vision slave camera interface: As the main camera, an external slave camera (Slave) can be connected to achieve binocular stereo vision or multi view detection without affecting the frame rate of the main camera (via an independent Gigabit Ethernet MAC). This significantly reduces the total cost of ownership (TCO) of multi camera systems, without the need for additional image acquisition cards or industrial computers.

USB 2.0 port: can be used to connect Wi Fi adapters, USB storage, or debug keyboards and mice.

RS-232 serial port: used for communication with PLC, robot controller, or barcode scanner.

VGA display output: Supports up to 2560 × 1600 @ 60Hz, making it easy to connect to a monitor for real-time image preview and parameter adjustment on site, without the need for an additional graphics card.

In addition, the built-in PWM lighting control output can directly drive LED light sources, supporting strobe or constant light modes, and the frequency and duty cycle can be adjusted by software without the need for an external controller.

Robustness design and environmental adaptability

IP67 casing: completely dust-proof and can withstand short-term immersion (30 minutes at a depth of 1 meter), suitable for humid and dusty environments such as food processing, automotive welding, outdoor inspections, etc.

M12 connector: All external interfaces (power, network, I/O, trigger) use M12 circular connectors with locking threads, anti vibration and anti pull, which are superior to traditional RJ45 or DB9.

Wide temperature operation: Supports temperatures ranging from -25 ℃ to+60 ℃ (depending on configuration), meeting the needs of most industrial sites.

Resistance to impact and vibration: Complies with IEC 60068-2-27 and IEC 60068-2-6 standards, suitable for mobile robotic arms or vibration platforms.

Software Ecology and Development Migration Convenience

The most prominent advantage of the NEON series is its x86 architecture, which is fully compatible with standard PCs and can run Windows 7/Windows Embedded Standard 7 (as well as potentially supported Linux or Win10 IoT in the future). This means:

Developers can directly use commonly used tools such as Visual Studio, C++/C #, Python, etc. without the need to learn embedded cross compilation environments.

Existing x86 platform visual code (such as using OpenCV, HALCON, Common Vision Blox, etc.) can be seamlessly migrated without the need to rewrite drivers or underlying APIs.

Supports GenICam standard and GenTL acquisition interface, compatible with various third-party image acquisition software.

Provide support for Open CV and Open CL, fully utilizing GPU and CPU parallel computing.

For users migrating from old x86 smart cameras or industrial control computers+standalone camera solutions, the software compatibility of the NEON series significantly shortens the development cycle and reduces maintenance costs.

Typical application case analysis

1. High speed robot grasping (NEON-1020)

On food or drug packaging lines, the conveyor belt speed can reach 5m/s, requiring cameras to capture moving objects in a very short exposure time. The 120fps global shutter of NEON-1020 combined with PWM strobe lighting can freeze motion, and with the fast ROI extraction of FPGA, real-time positioning and attitude calculation can be achieved, triggering robot grasping through digital output.

2. High precision 3D measurement (NEON-1040+from camera)

Utilizing dual camera stereo vision, with NEON-1040 as the primary camera and another GigE camera as the secondary camera, synchronously capturing different perspectives of the workpiece. Obtain dual images through GenTL, run stereo matching algorithms (such as SGB) on the CPU, generate point cloud data for size verification or reverse engineering. Integrated I/O can trigger the rotating table to achieve multi angle scanning.

3. OCR and barcode reading in harsh environments

In the foundry workshop or outdoor logistics station, IP67 protection ensures that the camera is not affected by dust and rain. The built-in x86 CPU can directly run OCR neural networks (such as Tesseract) or barcode decoding libraries without the need for an additional industrial computer, and send the results to the PLC via RS-232.

Notes on System Integration and Deployment

1. Power supply requirements

Requires 12-24V DC power supply (with a power consumption of approximately 15W), it is recommended to use industrial grade power supply and add surge suppressors. The power supply is connected through the M12 interface, pay attention to polarity.

2. Network configuration

The main camera has two network interfaces - one for main Ethernet (including PoE function)? But the document is not clear, the actual NEON series usually does not support PoE and requires separate power supply), and another one is used to connect from the camera. Suggest configuring an independent IP segment from the camera to avoid broadcast conflicts.

3. Trigger and synchronization

The external trigger signal is connected to the isolated input channel and can be set as a rising edge or falling edge trigger. Simultaneously, it can be triggered through software or PWM (internal timing). If multiple cameras are used for linkage, it is necessary to ensure that the synchronization delay of the trigger signal is less than 1ms.

4. Heat dissipation

Although the TDP is low, it is recommended to add auxiliary heat sinks or small fans to avoid frequency reduction when operating at full load for a long time in a closed chassis (such as continuous 120fps acquisition+deep learning inference).

5. Software installation

Applications can be deployed to SSDs via USB or network during development. It is recommended to enable the write filter function (EWF) of Windows to protect the lifespan of flash memory and prevent accidental shutdown from damaging the file system.

Common problems and troubleshooting suggestions

Possible causes and solutions for the problem phenomenon

The camera cannot recognize insufficient power supply or USB driver issues. Check the power supply voltage; Replace the USB cable; View Device Manager

Reduce exposure time and increase illumination brightness if the image is blurry or the drag exposure time is too long or the motion speed is too fast; Enable global shutter mode; Using external triggering for synchronization

Set the camera IP address separately if there is no image IP address conflict or GenTL configuration error from the camera; check the network cable and switch (if necessary); Confirm that the main camera software has started collecting data from the slave device

The digital output has no response, and the load current is too high or the output is not enabled to measure the load impedance, ensuring that it is less than the rated value; Check if the output channel in the software is set to high

Overheating, automatic shutdown. If the ambient temperature exceeds the specifications or the heat dissipation is poor, improve the ventilation of the chassis and add thermal pads or heat sinks

Comparative advantage with competitors

Compared to traditional ARM+FPGA smart cameras (such as some models from Cognex and Keyence), the advantages of the NEON series are:

Computing power: The quad core 1.91GHz is much higher than ARM Cortex-A9 and can run deep learning inference.

Software flexibility: Standard Windows ecosystem, supports any third-party visual library, not limited to manufacturer closed IDEs.

Scalability: It can be externally connected to cameras, VGA displays, and USB peripherals, far exceeding the limited I/O capabilities of competitors.

Upgrade Path: With processor iterations, x86 platforms can easily upgrade software without the need to replace hardware.

Of course, its cost is slightly higher than low-end ARM solutions, but for complex projects that require high performance and rapid deployment, its cost-effectiveness is outstanding.