The Truth About TSN and Industrial Ethernet Technology
We recently received feedback from ETG members that some articles circulating online comparing various industrial Ethernet communication technologies contain errors in their descriptions of EtherCAT. Members request that ETG correct these errors to prevent readers from being misled. Furthermore, we would like to take this opportunity to provide our interpretation of the article's claim that TSN technology is suitable for field-level applications.
A comparison of multiple industrial Ethernet fieldbuses named after a particular fieldbus organization only provides a technical comparison from the perspective of one type of bus. This comparison is neither representative nor authoritative for most users. Furthermore, due to the limited understanding of EtherCAT technology, the article contains many technical errors, causing considerable confusion for industrial Ethernet users.
We have listed the following technical errors and made corresponding corrections to help users correctly understand Industrial Ethernet and EtherCAT technology.
According to the EtherCAT parameters shown in the diagram: more than 70 nodes, 100 bytes of data, and a cycle time of more than 1700 microseconds.
The performance of EtherCAT, as tested in practical applications using hardware, is as follows:
• 1000 switching signals are distributed across 100 nodes, with a cycle time of 30 microseconds.
• 100 servo axes, each with 8 bytes of input/output data, and a cycle time of 100 microseconds.
Performance comparison of EtherCAT with other Gigabit Ethernet and TSN
As shown in the diagram above, compared to other gigabit Ethernet-based bus technologies like TSN, the actual performance of EtherCAT should be as represented by the bottom plane (blue plane). Clearly, EtherCAT outperforms other technologies.
For a typical EtherCAT system containing three types of devices: input devices, output devices, and mixed input/output devices, the minimum cycle time formula for an EtherCAT system containing these three types of devices should be:
Note: The EtherCAT protocol defines a data frame structure that can divide a data packet into multiple sub-packets, each of which can correspond to one or more slave stations. Generally, sub-packets correspond to a series of devices of the same type; for example, all input modules correspond to one sub-packet, all output modules to one sub-packet, and all input/output modules to one sub-packet. Therefore, if the system includes these three types of devices, plus a broader sub-packet for monitoring the device state machine, the formula should be 4 × (x + 12). The text mentions that each slave station corresponds to a sub-packet, which is not mandatory, and this extreme approach is rarely used.
Error 3: The following content is mentioned after the formula description:
"All the equations presented here assume a simple case where the input and output data volumes are equal and the topology is a perfect bus. However, in practical applications, this comparison depends on many other parameters:"
• Ratio of input data to output data
• Percentage of devices with direct cross-communication
• Utilize different cycle periods
• Topology (bus, star, ring), and the number of hops between devices
• Availability of modular I/O with its own backplane bus
EtherCAT's performance is highly deterministic. For applications with a fixed number of nodes and load, the EtherCAT system performance is deterministic, without needing to consider other unnecessary factors.
• EtherCAT is independent of the ratio of input to output data. EtherCAT system performance is related to transmission time and total data volume, not the ratio of input to output data.
• EtherCAT performance is independent of the percentage of devices with direct cross-connection. EtherCAT slave physical layers use cross-indexed adaptive PHYs, so it is independent of the percentage of devices with direct cross-connection, and no full-duplex switches or half-duplex hubs are required in the system.
• EtherCAT performance is affected by different cycle times, but this approach means that the master station can optimize messages, and different types of devices send data according to different task cycles, thereby freeing up the performance of the master station.
EtherCAT supports various topologies (bus, star, ring, tree, and linear), and its performance is independent of the topology. Furthermore, it does not involve the number of hops between devices, so there is no impact from this.
• EtherCAT is a single-line network with no backplane bus, so there is no impact from the backplane bus.
Regarding TSN and EtherCAT:
TSN is useful for real-time data exchange in heterogeneous networks, but it will not replace EtherCAT in fieldbus layer applications.
TSN stands for Time Sensitive Networking, a project of the IEEE Bridging ("Switching Technology") working group. Because traditional "best effort" approaches for ordinary Ethernet applications cannot meet the widespread real-time requirements (e.g., in audio/video and extensive IT communications with EtherCAT systems), this working group aims to improve the real-time performance of Ethernet through deterministic research. It has the following characteristics:
- In the IEEE 802.1 standard, data frames are forwarded as quickly as possible to avoid congestion.
- A portion of the bandwidth is reserved for "Stream" (high-speed data channel) to transmit communications with higher time requirements.
- The remaining bandwidth is used for normal data transmission ("delay channel").
As shown in the diagram above, real-time performance is achieved by establishing high-speed data channels within the TSN. However, for field-level data communication, which typically involves a large number of nodes, requires rapid response, and involves periodic, cyclical communication, directly connecting to a system with n nodes via the TSN would necessitate establishing n high-speed data channels. This would obviously impact other types of data communication and is impractical in terms of network configuration.
The true purpose of TSN lies in building heterogeneous networks. In future industrial networks or broader industrial applications, a single network needs to integrate different types of devices and different types of communication. When these communication devices need to interact in real time, this is where TSN comes in. TSN connects different devices or network segments through the configuration of TSN switches, enabling real-time data exchange. This real-time capability is achieved through the "high-speed data channel" configured for each network segment or device by TSN. Therefore, embedding TSN in IEEE 802 technology helps coordinate communication; it eliminates some unnecessary conflicts but does not change the fundamental rules. TSN's two main performance limitations are unchangeable: its efficiency in processing small Ethernet frames and its complex and time-consuming forwarding process.
In machine control, EtherCAT outperforms TSN by nearly ten times in typical I/O segments. As the fastest industrial Ethernet fieldbus available today, EtherCAT can be perfectly integrated with TSN technology.
As shown in the figure above, in heterogeneous networks, machine control (multiple nodes, high-speed response, hard real-time requirements) is involved, and EtherCAT network segments are used. Each EtherCAT network segment (containing multiple EtherCAT node devices) is connected to the master station through a high-speed data channel established by a TSN switch.
This usage is highly efficient. Based on the shared frame method, EtherCAT allows data from multiple slave devices in the network to be transmitted within the same data frame. Connecting EtherCAT to a TSN network makes TSN configuration easier and more efficient. The entire EtherCAT segment utilizes only one "high-speed channel" of the TSN, ensuring real-time communication between all devices in the EtherCAT segment and other types of devices, as well as the EtherCAT master station, in heterogeneous networks.
Furthermore, the development of TSN specifications is still ongoing. As shown in the diagram, some specifications are still in the draft or voting stages, and only a portion has been released to the public. It is worth noting that configuration tools for TSN networks have not yet been released, and effective TSN networks cannot yet be configured.
To ensure the compatibility of EtherCAT technology with TSN, the EtherCAT Technology Association (ETG) released an industry standard for EtherCAT devices regarding TSN at the end of 2017. This standard specifies the use of EtherCAT and TSN, defining the integration of the published TSN specification with EtherCAT technology. This document can be downloaded from the ETG official website. As the TSN specification is improved, this document will be updated accordingly; the current version is ETG.1700 S (D) V0.9.0.
The real advantage of TSN lies in enhancing the performance of local networks, including those with many machines. Its future development will inevitably reduce the complexity of current machine-to-machine network devices. This will require a common protocol infrastructure and network control system, both of which need devices capable of efficient processing at the machine layer.
In future automation systems, how communication is organized will be more important than the communication features themselves. This is why EtherCAT retains existing components and provides separate adaptations for TSN functionality. Therefore, we can support more powerful features and protect your investment in EtherCAT.
About the EtherCAT Technology Group (ETG):
The EtherCAT Technology Association is an organization supported, promoted, and refined by key users and leading automation suppliers across various industrial sectors. With over 4,600 member companies from 65 countries, the EtherCAT Technology Association is currently the world's largest fieldbus organization. Founded in 2003, it is also the fastest-growing fieldbus organization.
About EtherCAT®:
EtherCAT is an industrial Ethernet technology characterized by high performance, low cost, ease of use, and flexible topologies. Introduced to the market in 2003, it became an IEC international standard and a SEMI standard in 2007. EtherCAT is an open technology: anyone can implement or use it.
For more information, please visit: www.ethercat.org.cn
