... on the right the real demonstrator setup.

© ISW Stuttgart

The drive setpoints generated by a PC-based numerical controller are sent via native Ethernet with TSN mechanisms instead of a fieldbus stack. These are prioritized in their own time slots (using IEEE 802.1Qbv) across two industrial switches (Hirschmann RSP35 with TSN firmware) and transmitted to two FPGA-based drive inverters. Both drives achieve a synchronicity of less than 100 ns. Parallel TCP/IP traffic from a webcam does not affect the real-time behavior in any way.

Figure 2 shows the set-up with the two data streams: the transmission of the setpoint/actual values to and from the drives under real-time requirements (RT) and the transmission of the webcam stream with high bandwidth (NRT). The best possible time synchronization of the participants is an essential prerequisite for low-latency transmission via several network components (each traffic class is processed at the right time), but is also indispensable for the respective application - in this case the exact synchronization of the two drives as required for motion applications. The IEEE 1588 standard was used for this purpose, which will be incorporated into TSN in a slightly modified form as IEEE 802.1AS-rev. Corresponding functions are already integrated in the switches.

Figure 3: Structure of the NC master in the ISW's TSN demonstrator.

© ISW Stuttgart

For a better understanding of the implementation, Figure 3 shows the schematic structure of the NC master. Standard PC hardware with a Windows operating system was used. For deterministic execution of the real-time-critical tasks, these are implemented under the real-time extension 'Intime' from Tenasys. The 'Intel i210' network adapter, which also has corresponding hardware functions for time synchronization and is accessible via the high-performance Ethernet interface, ensures real-time-capable bus access. In addition to generating the drive setpoints within the NC using 'ISG.kernel', another important task in the software is time synchronization with the other network participants, for which a PTP stack from the Zurich University of Applied Sciences (ZHAW) is used. Non-real-time critical traffic, for example for configuring the drives, is transmitted outside the reserved time slot.

Figure 4: Structure of the TSN-capable drive controllers.

© ISW Stuttgart

The drive platform shown in Figure 4 or used in the ISW demo is characteristic of a TSN-capable end device: The implementation is based on an FPGA platform in order to be able to use maximum flexibility in software and hardware. Functions are implemented at the appropriate level. For example, the PTP stack or management tasks used here also run within an embedded CPU - corresponding time-critical functions such as the necessary time-stamping of the Ethernet packets or the control loops of the drives are implemented directly in hardware.

Here, too, there is still room for further improvements. One example is the switch to cut-through switching, which minimizes end-to-end latency even in more complex networks. As part of the demonstration presented, a cycle time of 1 ms was implemented for the transmission of set/actual values, with the drive data being packaged in exemplary telegrams based on CANopen. Of the available bandwidth of the 100 MBit connection, for example, 25% was reserved for real-time data and 25% for the reliable transmission of service data. Accordingly, the number of drives - arranged in any topology - could be significantly increased; the individual telegrams only have a minimum length of 72 bytes. As the drives are synchronized via the IEEE 1588 mechanism, the exact transmission time or the latency of the set values is no longer decisive. Using pulses output synchronously with the controller clock, measurements show that the time error between the two drive controllers is less than 50 ns.

The performance that can be achieved when using TSN Ethernet for typical industrial applications is basically comparable to established Ethernet-based fieldbuses and will continue to increase as transmission rates increase. With complex network structures, however, corresponding latencies due to switching must be taken into account, which limit the minimum response time. However, as the synchronization of the participants can be decoupled from the process data, this is not a critical limitation. Further challenges arise with a large number of participants from the overhead, which is caused by the fact that even a simple Boolean measured value always occupies a complete Ethernet frame. For special use cases, such as a very large number of distributed I/O components with small amounts of data, there is still a need for research into optimal bandwidth utilization.

Another TSN demo, in which the ISW was also involved, could be found at the SPS IPC Drives at the Sercos International stand: Here, the control of Sercos drives via a comparable network structure was presented. The necessary modifications here are limited to the NC master and leave the Bosch Rexroth drives used untouched. ISG.kernel' is also used for setpoint generation, while the role of the Sercos master is assumed by 'Sercos III SoftMaster', which is available as open source. Here too, IEEE 1588 ensures a uniform time base for network components and master, so that the setpoint telegrams are forwarded to the drives at precisely defined times.

This scenario can be used to illustrate a possible migration strategy for integrating existing systems and components into TSN networks. By configuring the network appropriately, the sercos telegrams are transmitted transparently and deterministically and protected from further traffic.

What remains of the fieldbus?

As things stand today, the focus on Ethernet-based bus systems is expected to continue in automation technology and other sectors, such as the automotive industry. As part of the upcoming release of the OPC UA pub/sub model, the combination with TSN promises to eliminate the previous lack of suitability for real-time data. However, this leaves the task of porting existing device profiles to this new communication infrastructure - taking into account compatibility with legacy devices and systems wherever possible.

Many of the fieldbuses in widespread use today have specifications for the network topology and therefore restrict the arrangement and wiring of the participants. Furthermore, any required redundant connections are not possible at will, but are restricted to those topologies provided for by the fieldbus - for example, rings. The use of TSN Ethernet as a transport layer removes such restrictions. Overall, a rapid increase in compatible chipsets and protocols can be expected due to manufacturer-independent standardization and cross-industry significance.

Against this background, it cannot be assumed that the classic fieldbus will disappear from the factory floor tomorrow. However, now is the right time to think about what TSN will mean for the various players and to prepare appropriate strategies. Relevant aspects are:

• Convergent networks with simultaneous high bandwidth and real-time guarantee
• Native Ethernet instead of proprietary communication protocols
• Manufacturer-independent, future-proof and cost-effective
• Flexible network topologies and redundancy

At the same time, industrial communication still presents a number of unresolved challenges: From standardized data description, engineering and monitoring to safety and security, there are still a number of unanswered questions. In order to clarify such questions from a practical point of view, to try them out and to create model solutions, the ISW is planning to establish a corresponding interest group 'TSN for Automation' for the practical application of real-time Ethernet(http://www.tsn4automation.com). Among other things, various events are planned for the exchange of experts and user training, the provision of best practices and reference implementations as well as the joint use of a TSN lab.

Authors:
Florian Frick works at the ISW on innovative platforms for automation technology;
Peter Zahn is group leader at ISW and responsible for drive, control and machine technology.
Armin Lechler is deputy director of the ISW.