Over the past four decades, machine vision has emerged as a transformative technology in manufacturing, offering manufacturers unprecedented efficiency and increased productivity gains. With its ability to outperform human labor in inspection, machine vision technology has significantly increased productivity and enabled faster production times, inspiring a new era of manufacturing excellence.

Comprising both hardware and software, machine vision systems automatically capture video images of objects and apply that data to algorithms to inform decisions. In a standard inspection application, for instance, powerful computers will process, analyze, and measure various characteristics of an acquired image of an object before issuing a pass-fail response for automatic inspection, often in milliseconds. Beyond inspection, machine vision unlocks new possibilities in logistics tracking operations and analysis, remote monitoring and control of industrial processes, and guiding robots in sorting mixed components and precisely aligning parts for assembly. The use cases for machine vision are expanding rapidly across every industrial sector, opening up a world of exciting opportunities. Industrial managed Ethernet switches enhance digital operations and safeguard against cyber threats in these demanding manufacturing environments.

In this article, we explore the role of industrial Ethernet switches in machine vision. We guide you on what to look for when selecting a switch that can reliably handle machine vision’s high bandwidth and data rate requirements, focusing on GigE Vision cameras, today’s most widely used protocol in many other industrial networks and machine vision applications.

Machine Vision Systems Components

The configuration of any machine vision system is dictated by the application’s demands, which include resolution, field of view, frame rate, speed of production, and interface requirements. Industrial switches are crucial for connecting various components and ensuring reliable data transmission in machine vision systems. In general, however, a computer vision system will consist of a camera, lens, cabling, and lighting. Image processing is performed on a connected PC running software and related algorithms to analyze images and extract data. For some protocols, such as CoaXPress or CameraLink, a unique frame grabber must be an intermediary between the camera and the PC. However, that is not the case with GigE Vision.

What is GigE Vision?

GigE Vision is a widely adopted standard for machine vision systems that enables the transmission of high-speed, high-quality video data over Ethernet networks. This protocol allows machine vision cameras to communicate seamlessly with computers and other devices using a standard Ethernet connection, simplifying the integration of machine vision systems into industrial automation environments. GigE Vision is particularly advantageous in applications requiring rapid data transfer, such as manufacturing process inspection, quality control, and robotics. By leveraging the existing Ethernet infrastructure, GigE Vision facilitates the deployment of vision systems in diverse industrial settings, enhancing efficiency and precision.

The GigE Vision standard is a vision-oriented version of the popular Gigabit Ethernet protocol. GigE Vision cameras deliver data rates exceeding 1000 Mbps per port with a maximum cable length of 100 meters, which is longer than most other protocols used in machine vision systems. Because GigE Vision has device discovery capabilities and standards-based, plug-and-play installation, a system is much easier and cost-efficient to configure. Another benefit is that multiple cameras can be controlled and synchronized remotely.

GigE Vision cameras leverage off-the-shelf Ethernet networking equipment and sensors that can be purchased economically. Another cost-saving advantage is that GigE Vision cameras operate without an expensive frame grabber. Instead, they plug directly into an RJ45 port of a PC, adapter, server, or industrial automation switch, where a Network Interface Card (NIC) insulates the camera from general network traffic. 10 GbE Industrial network switches support the high-speed data transfer and robust security features required for GigE Vision systems. NICs are typically built into a PC on the motherboard or as PCI or PCIe cards, ensuring a cost-efficient setup for your machine vision system.

Benefits of GigE Vision

The benefits of GigE Vision make it a preferred choice for industrial machine vision applications. One of the primary advantages is its high-speed data transfer rates, which are crucial for applications that demand real-time processing and analysis. Additionally, GigE Vision supports long cable lengths up to 100 meters, providing more excellent system design and installation flexibility. Using standard Ethernet infrastructure further enhances its appeal, eliminating the need for specialized cables and interfaces and reducing overall costs. Moreover, GigE Vision’s widespread adoption ensures compatibility with various devices and systems, facilitating seamless integration and interoperability in machine vision applications.

The Role of Ethernet Switches in GigE Vision Systems

Ethernet switches are pivotal in GigE Vision systems, enabling the efficient transmission of high-speed video data between devices. Industrial Ethernet switches, in particular, are engineered to meet the rigorous demands of industrial automation environments, offering high-speed data transfer rates, reliability, and robust security features. These industrial switches support Quality of Service (QoS) and traffic prioritization, ensuring that critical machine vision data is transmitted in real-time without delays. Managed switches are especially beneficial in GigE Vision systems, providing advanced features such as network management, security, and troubleshooting capabilities, essential for maintaining optimal performance and reliability in vision systems.

Network Equipment Configuration

The most basic network configuration for a GigE Vision machine vision system is peer-to-peer (P2P), which connects the cameras directly to the host PC, either with a GigE adapter or a multiport GigE adapter. P2P’s upside for machine vision products is its simplicity in design, combined with each camera having full connection bandwidth for transmitting images. The downside for human vision is that P2P dramatically limits the number of main types of cameras connected to a single PC.

Alternatively, you can connect cameras and sensors to a host PC via network switches, giving each camera its IP address. The critical advantage of configuring your system with an industrial Ethernet switch is that it can reduce costs and empower you to connect more cameras to the PC. In addition, if an Ethernet switch features PoE, a separate power cable is eliminated for each camera. Finally, it also permits longer overall cable lengths, enhancing your design’s flexibility. Here’s how: When the first cable between the camera and the industrial switch is added to the second cable from the switch to the PC, you effectively double the 100-meter limits of Ethernet.

Besides basic connectivity, security, and device management tasks, a managed industrial Ethernet switch can serve several purposes in industrial networks, including:

• Trunking to create multiple parallel paths between switches when bandwidth requirements exceed limits
• In a multi-camera configuration where each camera outputs large amounts of image data, an industrial switch can consolidate data from multiple ports and output it to a server or PC using a single port.
• Machine vision systems that use multicasting require an industrial switch to send camera data to multiple host PCs.

Network Management and Security

Effective network management and security are crucial components of GigE Vision systems. Industrial managed Ethernet switches come equipped with network monitoring, configuration, and troubleshooting features, enabling users to manage and maintain their network efficiently. Machine vision software plays a vital role in network management and security, offering functionalities such as data encryption and secure data transfer, ensuring the integrity and confidentiality of the transmitted data within vision systems.

What to Look for in Industrial Ethernet Switches

Choosing the right industrial Ethernet switch is essential for building robust machine vision software systems with GigE Vision cameras. Industrial switches designed for harsh environments offer enhanced digital operations and security, making them ideal for demanding applications. We have identified vital selection criteria below.

Several cameras: An industrial automation switch must have as many ports as cameras, plus an additional port for an upstream connection. This upstream connection should ideally have a higher bandwidth port to enable data aggregation from all cameras and the upstream transmission. A popular method is using a 10 GbE managed switch with smart cameras and a 10Gbps (10GBASE-T) connection for the uplink.

Bandwidth: A GigE Vision camera will output up to a full 1 Gbps. Therefore, an Ethernet switch must have enough bandwidth to process and manage Gigabit traffic. To be non-blocking, the switch fabric must support the aggregate data rate of all ports operating at bandwidths exceeding 1 Gbps.

Packet Buffer Memory: Packet buffer memory is a reserved memory area in network equipment that stores packets awaiting transmission or receiving over a network. The network switch buffer prevents data loss by temporarily storing these incoming packets. For instance, imagine four GigE Vision cameras streaming four separate 1 Gbps streams in a burst to a host PC on a 1 Gbps connection. Since the PC cannot receive 4 Gbps, the packets are stored in the switch until they can be transmitted.

Power supply: A Power over Ethernet (PoE) switch will transmit DC power and data to control a GigE industrial camera, so an external power supply is unnecessary. Pay close attention to the PoE standard specified for a camera and the type supported by the switch.

Jumbo Frame Support: Modern GigE Vision cameras support packets larger than 1500 bytes; the most common size is 9000 bytes, also known as Jumbo Frames. Not all Ethernet switches support frames larger than 1500 bytes or may require complex configuration.

Latency: Machine vision latency is usually constrained by the host PC performing the processing rather than an industrial switch or network infrastructure. That said, machine learning latency will be impacted by two architectures in an Ethernet switch: cut-through switching mode and store-and-forwarding. Cut-through is an operation where a switch receives only a portion of a frame and immediately forwards it to the next node for subsequent processing. This method has a lower network latency than store-and-forward switching, and invalid frames are forwarded instead of dropped.

Durability: When selecting an Ethernet switch for an industrial setting, it’s always critical to consider its durability. Unreliable networking products cost industrial users millions of dollars annually in maintenance expenses, network and product downtime, replacement costs, and loss of service. The more critical the application, the more influential the device’s reliability is.

Hardened safety features in an industrial-grade Ethernet switch include high EFT and ESD protection, resistance to shock and vibration, a wide operating temperature range, redundant power supplies, and fanless cooling.

Troubleshooting and Maintenance

Troubleshooting and maintenance are essential for the smooth operation of GigE Vision systems. Industrial Ethernet switches provide diagnostic tools, error detection, and troubleshooting features, allowing users to identify and resolve issues quickly. Machine vision software also offers error detection and diagnostic tools, facilitating the troubleshooting and maintenance process. Regular maintenance is critical to ensure the system operates optimally, minimizing the risk of downtime and errors. By proactively addressing potential issues and performing routine maintenance, users can enhance the reliability and longevity of their vision systems, ensuring consistent performance in industrial applications.

Conclusion

GigE Vision stands out for its production speed, safety, and ease of system integration. However, as it relies on the often-limited bandwidth of local networks, it can have issues that may critically affect tasks and operations that require high-speed, real-time precision. Industrial switches play a crucial role in ensuring the reliability and efficiency of GigE Vision systems in industrial environments.