“The topic of Time Sensitive Networking (TSN) is something anyone working in industrial communications has faced these days. TSN is bound to come; it’s just a matter of when and how. However, even today, its advantages in industrial communications are not well understood.
The topic of Time Sensitive Networking (TSN) is something anyone working in industrial communications has faced these days. TSN is bound to come; it’s just a matter of when and how. However, even today, its advantages in industrial communications are not well understood.
Ethernet was introduced to the office space in the early 1980s and quickly became popular due to its astonishingly high throughput of 10Mbps (relative to the time). However, this Ethernet is not practical for real-time applications because it uses a common medium called a “community wire”. Conflicts are prone to occur in situations of high utilization, leading to problems with office setups.
Later, with the development of Ethernet, the conflict problem was solved by introducing switching network. In addition, Ethernet datagram priority is introduced through Quality of Service (QoS).
For industrial applications, it is especially important to guarantee low latency. Despite QoS, standard Ethernet used in office environments can only guarantee latency down to a certain point, especially with high network utilization.
This stems from a number of reasons, the main ones being the store-and-forward strategy common to commercial multiport switches and the fact that it is impossible to reserve bandwidth.
Store-and-forward means that the switch needs to receive a complete data packet before forwarding. This has advantages in switch processing, but also introduces potential problems that can negatively impact latency and reliability:
▲When the data packet passes through the switch, a certain delay will be generated according to its length. If multiple switches are cascaded, the latency effect is amplified.
▲ Since the storage capacity of the switch is not unlimited, if the network is overused (too much traffic), it may reject data packets; this means that data packets (even those with higher priority) may be lost.
▲Long data packets will block the port for a long time.
Switch cascading has created challenges for industrial environment applications from the very beginning. In addition to star topologies used in IT, line, ring and tree topologies are often used in automation. These adjusted topologies significantly reduce the cabling requirements and cost of Ethernet installations. Therefore, in industrial applications, two-port switches with cut-through strategies are integrated into field devices. Straight-through means that data packets are forwarded before being fully received.
Figure 1. Ethernet frame: Data fields related to TSN data stream identification are shown in green.
Figure 2. Topology.
It must be guaranteed that sufficient bandwidth (and buffer space) is always available for high-priority datagrams. Standard Ethernet is currently unable to do this.
One-size-fits-all technology: Industrial Ethernet is born
Since traditional Ethernet does not have sufficient bandwidth reservation capabilities, experts in the automation field have been developing their own versions of Ethernet since 2000. However, the development paths they take vary. Each of the following methods has its own characteristics:
▲Using Ethernet as a protocol for fieldbus transmission medium. These protocols require themselves to fully control the Ethernet medium. Traditional TCP/IP communication can only be piggybacked via fieldbus (EtherCAT® and POWERLINK®) or fieldbus (Sercos) assigned channels. Fieldbuses have a firm grip on bandwidth control.
▲A protocol to ensure bandwidth reservation through a time slicing program on Ethernet. Here to mention PROFINET? IRT. IRT supports deterministic hard real-time data transfer over the same cable running soft real-time or background traffic. Time slice planning requires an accurate transmission path timing model.
▲ Protocol based on shared Ethernet cable. These protocols use QoS and are well suited for factory automation and process automation applications.
PROFINET RT and EtherNet/IP are interesting examples. These protocols are limited to soft real-time (cycle time ≥ 1 ms) range.
Figure 3. Timing model: PHYs, cables, and switches contribute to data transfer latency. This must be taken into account when using slotted methods (PROFINET IRT and TSN Time Aware Shaper (TAS)).
These standards require special hardware to support and therefore special ASICs. Since PROFINET RT and EtherNet/IP™ are also based on embedded dual-port cut-through switches, they are no exception in this regard. Flexible, hardware-based multiprotocol solutions such as ADI’s fido5000 solve this problem in a streamlined manner.
Step into TSN
With TSN, the industry has successfully developed an extended version of the standard Ethernet that conforms to IEEE 802.1, successfully breaking away from the limitations of the past. Therefore, the standardized Layer 2 in the ISO seven-layer model is now upwardly compatible with previous Ethernet and hard real-time functions. With 802.1AS-rev, TSN also defines an interoperable unified method for synchronizing distributed clocks in the network. Shared cables are possible in hard real-time applications as well as in all other applications (web servers, SSH, etc.) since “best effort” communication is always possible over TSN. In this respect, TSN is no different from PROFINET IRT, which also offers comparable performance.
An added requirement for adopting TSN is the need for a wider network configuration. Centralized or distributed configurations are possible. A configuration that takes into account both types is currently being discussed and implemented. A future development goal is to achieve interoperability between the two configuration mechanisms.
All this could not have been better, what are the actual advantages of TSN?
The most common answer is that it provides a lower cost network interface on the market and is suitable for a wider market. After all, TSN can also be used in the building automation and automotive industries of the future. In fact, the market size for embedded TSN solutions is expected to be significantly larger than the current market for all industrial Ethernet solutions combined.
Compared with the previous industrial Ethernet method, the biggest technical advantage of TSN is its scalability. Unlike current industrial networks, TSN is not defined for a specific transmission rate. TSN can be used for 100 Mbps, as well as 1 Gbps, 10 Mbps or 5 Gbps.
It also better optimizes the topology, as the data rate can now be selected for each different segment. Whether it is 1 Gbps, 100 Mbps or 10 Mbps, a unified layer 2 – IEEE802.1/TSN is used.
A unified network infrastructure also helps those involved in the task of building and maintaining the network, because with TSN, solutions can now be used in areas other than automation, such as building, process and factory automation, and energy distribution.
This brings us to the next advantage, the training factor. TSN has become the subject of many universities, most of which are still in the research stage. However, technical and vocational colleges have shown great interest in this topic. We can fairly responsibly say that TSN will become a must-have basic knowledge for engineers, technicians, and skilled workers. Retraining for different fieldbuses will no longer be necessary.
What about “Brownfield”, the existing agreement?
In almost all TSN-related working groups, a recurring theme: How can the transition to TSN and the supply to existing installations (such as “brownfield” applications) be secured at the same time?
In all respects, the emphasis is on making the transition to TSN easy and smooth for customers. Today, we can safely say that existing industrial Ethernet protocols will not disappear overnight. Conversely, anyone who is currently using PROFINET, EtherNet/IP, EtherCAT, or similarly ubiquitous Industrial Ethernet protocols can safely assume that in 10 years they will still be able to run a network using these protocols, with support and Replacement parts.
All Industrial Ethernet organizations provide models to describe how existing plants work with new TSN-based devices. The interfaces of existing industrial networks consist of gateways (Sercos), interfaces with couplers (EtherCAT) or interfaces without any special hardware (PROFINET RT). In particular PROFINET and EtherNet/IP plan to use their full protocol as layer 2 for TSN.
This makes a gradual transition to TSN possible.
Figure 4. Brownfield: TSN network combined with PROFINET and EtherCAT.
In summary, TSN will be ubiquitous in new installations and gradually introduced into existing installations in the form of islands or segments.
However, with TSN, new protocols will emerge in the field of Industrial Ethernet. OPC UA with the new transport protocol PUB/SUB, together with TSN, is already seen as a competitor to the legacy protocol. For manufacturers of field devices, this means they have to support both traditional Industrial Ethernet solutions as well as TSN and new protocols.
TSN and Analog Devices
Analog Devices acquired Innovasic, a leader in Industrial Ethernet more than a year ago, and has successfully completed the integration. With the integration of Innovasic, the fido5000 series of industrial dual-port switches has been incorporated into ADI’s product line. The switch supports all relevant Industrial Ethernet protocols and supports TSN.
With fido5000, it is now possible to transition product planning to TSN while meeting current requirements (PROFINET IRT, EtherCAT, POWERLINK, EtherNet/IP, etc.). With the fido5000 it will also be possible to implement OPC UA PUB/SUB. With fido5000, TSN can be planned as an update to an existing system.
The fido5000 series is still being enhanced. We will provide new products for 1 Gb applications, but 10 Mb/100 Mb products will continue to be available to meet customer requirements.
With the flexibility provided by the fido5000 series, TSN migration can be performed reliably and efficiently.
TSN is an opportunity
TSN makes it possible to create a unified foundation for all industrial communications. Once TSN is introduced, layers 1, 2, and 3 of the ISO seven-layer model in industrial applications will be unified. This has the potential to bring scalability and performance to a whole new level.
Based on this, will upper layer communication be standardized as well? Is it possible to have a unified OPC UA PUB/SUB? possible. With the fido5000 series, users can be prepared for all scenarios.