Industrial Ethernet refers to Ethernet technology used in industrial automation. While technically compatible with commercial Ethernet (i.e., the IEEE 802.3 standard), its product design ensures that the materials used, the product's strength, and its applicability meet the needs of industrial environments, achieving industrial standards for environmental adaptability, reliability, security, and ease of installation.
Ethernet is ubiquitous and cost-effective, utilizing public physical links and offering faster speeds. However, Ethernet communication supporting TCP/IP is typically non-deterministic, with response times often around 100ms. Therefore, Industrial Ethernet requires modifications to meet industry standards. For example, the Industrial Ethernet protocol uses a modified Media Access Control (MAC) layer to achieve very low latency and deterministic responses. Furthermore, Ethernet's ability to provide flexible network topologies and support a larger number of nodes has led to its increasing popularity and wider application in industry in recent years.
However, driven by various industrial equipment manufacturers, a multitude of different industrial Ethernet protocols have emerged, including EtherCAT, PROFINET, EtherNet/IP, Sercos III, and Time-Sensitive Networking (TSN). Below, I will examine the details of these mainstream industrial Ethernet protocols.
EtherCAT: Rapid Recent Development
EtherCAT was originally developed by Beckhoff Automation in Germany. Since 2003, it has been under the framework of the EtherCAT Technology Group (ETG), an industrial fieldbus organization comprised of approximately 7,000 member companies. It is an open but not open-source technology, meaning that while you can freely use the technology, you will need to obtain a license from Beckhoff Automation to develop related devices.
EtherCAT technology overcomes the system limitations of other Ethernet solutions: it eliminates the need to receive Ethernet packets, decode them, and then copy process data to various devices. EtherCAT slave devices read the corresponding addressing data as a packet passes through their node; similarly, input data is inserted into the packet as it passes (see Figure 1). Because this process is handled in hardware, the entire process involves only a few nanoseconds of delay, resulting in extremely short response times.
Figure 1: Process data inserted into the message (Source: ETG)
EtherCAT is a MAC layer protocol that is transparent to any higher-level Ethernet protocol such as TCP/IP, UDP, or web servers.
In terms of topology, EtherCAT supports virtually any topology type, including linear, tree, and star topologies. Bus or linear topologies, derived from fieldbus architectures, can also be used for Ethernet and are not limited by the number of cascaded switches or hubs. This means that EtherCAT can connect up to 65,535 nodes in a system, and the EtherCAT master can be a standard Ethernet controller, simplifying network configuration. With low latency per slave node, EtherCAT provides a flexible, low-cost, and network-compatible industrial Ethernet solution.
Figure 2: Growth in the number of EtherCAT nodes over the past 9 years (Source: ETG)
EtherCAT has developed rapidly in recent years. According to the latest data from ETG, excluding modular I/O devices, ETG estimates the number of EtherCAT nodes worldwide at 59.1 million, with particularly significant recent growth. Since 2014, the number of EtherCAT nodes has grown exponentially, with 18.4 million nodes added in 2022 alone.
EtherNet/IP: Can be implemented using a microprocessor.
The EtherNet/IP industrial Ethernet protocol was originally developed by Rockwell Automation and managed by ODVA. It can be used in program control and other automation applications. Unlike EtherCAT, which is a MAC layer protocol, EtherNet/IP is an application layer protocol over TCP/IP. EtherNet/IP uses the standard Ethernet physical layer, data link layer, network layer, and transport layer. This means it uses commercial Ethernet communication chips, physical media, and a star topology, employing Ethernet switches to achieve point-to-point connections between devices. It can simultaneously support 10Mbps and 100Mbps commercial Ethernet products.
The EtherNet/IP protocol consists of three parts: the IEEE 802.3 physical layer and data link layer standards, the TCP/IP protocol suite, and the Common Industrial Protocol (CIP). CIP provides a set of common messages and services for industrial automation control systems and can be used on various physical media. For example, CIP on a CAN bus is called DeviceNet, CIP on a dedicated network is called ControlNet, and CIP on Ethernet is called EtherNet/IP. EtherNet/IP establishes communication from one application node to another through one TCP connection and multiple CIP connections; multiple CIP connections can be established through a single TCP connection.
Since EtherNet/IP uses the physical layer network of Ethernet and is built on the TCP/IP communication protocol, it can be implemented with software on a microprocessor without the need for special ASICs or FPGAs.
EtherNet/IP uses standard Ethernet and switches, so it has no limit to the number of nodes it can have in a system. This allows a network to be deployed across multiple different endpoints on a factory floor. EtherNet/IP provides complete producer-consumer services and enables highly efficient slave peer-to-peer communication.
EtherNet/IP is compatible with multiple standard Internet and Ethernet protocols, but its real-time and deterministic capabilities are relatively limited. Therefore, it can be used in automated networks where occasional minor non-determinism can be tolerated.
PROFINET: Standard Ethernet compatible, can be networked together.
PROFINET is an open industrial Ethernet communication protocol, primarily proposed by Siemens and the PROFIBUS & PROFINET international association. PROFINET applies TCP/IP and related information technology standards, and is a real-time industrial Ethernet protocol. Since 2003, it has been part of the IEC 61158 and IEC 61784 standards. PROFINET = PROFIbus + etherNET, transplanting the PROFIbus master-slave structure to Ethernet. Therefore, PROFINET has controllers and devices, whose relationship can be simply compared to the PROFIbus master and slave.
It has three different categories: PROFINET Class A can access the PROFIBUS network through a proxy, bridging Ethernet and PROFIBUS with remote procedure calls over TCP/IP. Its cycle time is approximately 100ms, mainly used for parameter data and cyclic I/O, with typical applications including infrastructure and building automation; PROFINET Class B, also known as PROFINET Real-Time (PROFINET RT), introduces a software-based real-time approach and reduces the cycle time to approximately 10ms. Class B is typically used for factory automation and process automation; PROFINET Class C (PROFINET IRT) is isochronous real-time transmission, requiring dedicated hardware to reduce the cycle time to below 1ms, thus providing the performance required for motion control operations in real-time industrial Ethernet.
In addition, since PROFINET is based on Ethernet, it can have Ethernet topologies such as star, tree, and bus, while PROFIbus only has bus topology. Therefore, PROFINET is a product that combines the master-slave structure of PROFIbus with the topology of etherNET.
PROFINET RT can be used for PLC-type applications, while PROFINET IRT is well-suited for motion applications. Branch and star topologies are common PROFINET topologies. Careful topology planning is required to achieve the desired system performance in a PROFINET network.
POWERLINK: A truly open-source industrial Ethernet protocol
POWERLINK is a real-time communication protocol over standard Ethernet. It is an open communication protocol managed by the Ethernet POWERLINK Standardization Group (EPSG) and was developed by the Austrian automation company Bernecker & Rainer Industrie-Elektronik (B&R). The first version was released in November 2001. Ethernet POWERLINK uses IEEE 802.3, thus allowing for free choice of network topology, cross-connects, and hot-plugging.
POWERLINK is an extension of Ethernet that combines polling and time-slicing mechanisms, providing the following:
Time-critical data can be sent in very short isochronic cycles, with a plannable response time.
All nodes on the network can be synchronized in time with an accuracy down to the microsecond level.
Data transfers that are not time-critical are transmitted in a dedicated asynchronous channel.
The current implementation can achieve a loop time of less than 200 µs and a time accuracy (jitter) of less than 1 µs.
POWERLINK masters, or "managed nodes," control time synchronization within tens of nanoseconds through packet jitter. Such systems are suitable for a wide range of automation systems, from PLC-to-PLC communication and visualization to motion and I/O control.
It's worth noting that POWERLINK is a solution that can be implemented over ordinary Ethernet without requiring ASIC chips. Users can implement POWERLINK on various platforms, such as FPGAs, Arm, and x86 CPUs. Wherever there is an Ethernet connection, POWERLINK can be implemented. Furthermore, all its source code is open source, and anyone can download and use it for free (just like Linux). Therefore, the barriers to implementing POWERLINK are minimal.
The POWERLINK source code contains complete code for three layers: the physical layer (standard Ethernet), the data link layer (DLL), and the application layer (CANopen). Users only need to compile and run the POWERLINK program on their existing hardware platform to implement POWERLINK in minutes.
It defines a streamlined, highly real-time data link layer protocol, while defining CANopen as an application layer protocol. This allows users to implement both POWERLINK and CANopen simultaneously. In other words, CANopen is a standard component, enabling customers to easily upgrade their systems from previous fieldbus protocols.
Sercos III: Primarily used for servo controllers
Sercos has been popular in factory automation applications (suitable for mechanical engineering and construction) for over 30 years. Sercos III, its third-generation protocol, was developed in 2003. This efficient and deterministic communication protocol integrates real-time data exchange of the Sercos interface with Ethernet, providing real-time Ethernet and standard TCP/IP communication to create low-latency industrial Ethernet.
Very similar to EtherCAT, Sercos III processes data packets by quickly extracting them and inserting them into Ethernet frames, thus achieving low latency. Sercos III separates input and output data into two frames. Cycle times start at 31.25 microseconds, as fast as EtherCAT and PROFINET IRT.
A single Sercos III master device can control multiple Sercos III slave devices (such as drivers, sensors, and analog and digital I/O devices), as shown in Figure 3. One master device can control up to 511 slave devices (i.e., slave nodes), and it is mainly used for servo driver control.
Figure 3: Exemplary Sercos III network ring topology (Source: TI)
Sercos III supports both ring and linear topologies. A key advantage is its support for ring topologies alongside linear ones. A major benefit of using ring topologies is communication redundancy. If the Ethernet cable fails, the Sercos III network can switch to a linear topology, allowing the master device to continue communicating with all slave devices in the network. Once the Ethernet cable is repaired, the master device can switch the Sercos III network from a linear to a ring topology.
TSN: A technology poised to replace the bus
In 2006, the IEEE 802.1 working group established the AVB Audio/Video Bridging Task Force, primarily to address the issue of real-time synchronous transmission of audio and video data over Ethernet. In 2012, the AVB Task Force expanded the application requirements and scope of Time-Deterministic Ethernet in its charter, and renamed the task force the TSN Working Group.
Time-Sensitive Networking (TSN) is a protocol standard developed by the TSN working group. It aims to achieve deterministic minimum time delay in non-deterministic Ethernet networks. It defines a time-sensitive mechanism for Ethernet data transmission, adding determinism and reliability to standard Ethernet to ensure real-time, deterministic and reliable data transmission.
Figure 4: TSN Standard (Source: ADI)
Beyond its real-time capabilities and determinism, TSN offers another significant technological advantage: network scalability, enabling it to operate at speeds of 10 Mbps, 100 Mbps, 1 Gbps, or even 10 Gbps. However, this requires meticulous (and therefore more complex) network configuration. Transmission rates of 1 Gbps and above represent the logical evolution of today's networks. 1 Gbps paves the way for new (Internet of Things) applications and helps overcome performance bottlenecks in data-intensive applications. However, TSN as a system can only reach its full potential when both the endpoints and Ethernet switches support TSN functionality.
TSN is a LAN-level solution that can work with non-TSN Ethernet, but timeliness is only guaranteed within a TSN LAN. Users can group the TSN standard according to the use cases it addresses: a general time view, guaranteed maximum latency, or coexistence with background or other traffic. Like any popular standard, the TSN standard toolkit is constantly evolving.
Figure 5: ISO/OSI Reference Model (Source: ADI)
In fact, TSN regulates data communication at Layer 2 of the ISO/OS reference model. Strictly speaking, TSN represents Layer 2 in Ethernet that supports real-time performance, not a complete real-time protocol. In other words, TSN will not replace PROFINET, EtherNet/IP, and similar Ethernet protocols. Instead, these industrial Ethernet protocols will support Layer 2 TSN in the long term; therefore, traditional industrial Ethernet protocols will not disappear, but will be built upon TSN in the future. However, fieldbuses may be permanently replaced by Ethernet.
Conclusion
The success of industrial automation relies on highly reliable and efficient communication networks that can connect all parts, no matter how short, for more efficient operation. The increasing prevalence and applicability of Ethernet will continue to drive the upgrading of traditional factories to Industrial Ethernet.
Currently, there is no unified standard for industrial Ethernet protocols, and each protocol has its own advantages and disadvantages. With the development of industrial automation, industrial Ethernet protocols are expected to continue to evolve and converge in the future to adapt to the networking and automation of factory equipment. To simplify the design of industrial Ethernet, many semiconductor manufacturers have launched more integrated chip products.
