Ethernet Powerlink Communication

Powerlinks nowadays serve to replace or supplement the field buses which (for example) connect field devices / actuators etc. in an installation with a control unit. As such, the Ethernet Powerlink belongs to the real-time Ethernets. They make it possible for example, to create a computer network of computers at the office level with those in production installations and in doing so, to connect all of the important components.

In this example, the computers in the production installation would be called “field devices”. The networking of these computers is called “vertical integration of the automation technology”, and it is very essential in automation technology, where this technology has become indispensable. However, it is problematic for standard Ethernets as they do not transmit their information in real-time, but a delay of approximately 1 ms must be expected.

The Ethernet Powerlink transmits data in the microsecond range. Originally, it was produced by B&R, and nowadays it is developed further by the Ethernet Powerlink Standardization Group.

Currently, Ethernet Powerlink is being offered in two versions.

Ethernet type 0x3e3f: Produced by B&R, this was previously used as the basis for further development. As such, it was only a kind of intermediate solution.

Ethernet type 0x88ab: This is the current version, which in comparison with the first version already had been developed further, e.g. CANopen device profiles, Powerlink Safety, electronic data sheets, master poll response.

The advantage of Ethernet Powerlink over other standard Ethernets is the fact that it is guaranteed to transmit the data, and it does this in very short, isochronous cycles using configurable time response. Ethernet Powerlink also can synchronize all network nodes in time (this is done in the microsecond range). Currently, it reaches cycle times below 200μs and a jitter of less than 1μs.

For data exchange with nodes in networks, it is possible to specify a communication protocol similar to CANopen through Ethernet Powerlink. When both elements operate together, they are treated by a so-called Powerlink protocol stack that requires no special hardware. In this way it is possible to realize master and slave nodes with standardized Ethernet components. As such, they are also available for different operating systems. Ethernet Powerlink also uses the device profiles of CANopen.

Ethernet Powerlink is located in the OSI layer model. This model was developed as the design basis for communication protocols in computer networks. It has communication tasks in seven layers that build upon each other (application layer, presentation layer, session layer, transport layer, network layer, data link layer, bit transfer layer). As such, each layer has special requirements.

How does the data transmission function?

Ethernet Powerlink is normally operated with twisted-pair cables and in this case, it is a Fast Ethernet. The readily available 8P8C/RJ-45-, the M12 plug-in connectors and glass fiber cables can be used with this. However, in the case of the latter, additional delays as a result of media converters can be expected.

Wiring in accordance with the Ethernet Powerlink standard (according to the IOANA guidelines) guarantees that planning and installation function correctly. So-called repeating hubs instead of switching hubs should be used in order to minimize the delay and the jitter (if desired).

Naturally, collisions in the network must also be prevented. Deterministic data transmission can only be guaranteed in this way. This is ensured by so-called controlled nodes (CN) which are only permitted to transmit when requested. This controls a managing node (MN).

The start of a cycle forms the start of cycle and/or SoC. Now each node is polled from the managing node with a poll request (PReq), whereby the controlled node is answered by a poll response (PRes). Here, other Powerlink devices can “listen”, as the answers are sent as an Ethernet multicast. The advantage of this is that, this way, the controlled nodes can communicate with each other.

Not every device has to be polled. This avoids long cycle times. The asynchronous phase starts after the cyclic phase. It starts with the start of asynchronous (SoA). Here, it is possible that in each case, one controlled node selected by the managing node sends data. Data from normal, i.e. non-deterministic networks and the Powerlink network can be exchanged via special gateways.

The Powerlink Packet

The Powerlink packet consists of the header and the payload. Everything is contained in a normal Ethernet frame, which should have 64 to 1500 bytes. It is important to know that so-called jumbo frames, i.e. those with a size in excess of 1500 bytes, are not permitted in a Powerlink network. Such jumbo frames are not standardized and are excessively large. They are only practical when the protocol overhead can be minimized by them. However, this must first be tested by checking if switches or routers can handle a jumbo frame and if this results in a speed advantage.

A Powerlink header consists of…

  • 1 bit reserved
  • 7 bit message type
  • 8 bit target node number
  • 8 bit source code number

Defined message types are SoC, which defines the start of a new cycle, PReq, which polls cyclic data of the CN, PRes, which sends current cyclic data of the CN, SoA, which signals the start of the asynchronous phase and ASend, which sends the asynchronous data.

Ethernet Powerlink

This system, a real-time Ethernet, was developed in order to be able to transmit data in the microsecond range. It is used mainly for the transmission of process data in automation technology. It was developed first by B&R, and today it is specified as a standard by the open user group EPSG, the Ethernet Powerlink Standardization Group.

Developed for standard conformity, it introduces mixed polling and time slice mechanisms for the deterministic transmission of data. This way, amongst other this, a guaranteed transmission of time-critical data is achieved in short cycles with configurable time behavior. Furthermore, a very accurate time synchronization of all of the connected networks is possible.

Additionally, there is transmission of less time-critical data in the asynchronous channel. With the instantaneous implementation of Ethernet POWERLINK, it is possible to achieve times of under 200 microseconds and a time precision (jitter) of less than one microsecond. It also specifies a communication protocol for payload exchange with nodes in the network. Both are handled by a POWERLINK protocol stack. However, no special hardware is required for this. As such, the master and slave nodes can be realized with simple Ethernet components. Open-source master and slave stacks are also available for different operating systems.

The data transmission of Ethernet POWERLINK

As the link on layers two and seven exists in the OSI layer model, it is normally independent of the physics that are applied. In practice, it is often used with twisted-pair cables to the fast Ethernet. With this, commercial 8P8Cs are approved as M12 plug connections. The use of optical waveguides is also possible, although additional delays as a result of media converters cannot be excluded.


Ethernet POWERLINK was included in the standards 61158-300, IEC-61158-600 and IEC 61558-500. Different frames can be sent in the asynchronous phase. For example, this makes it possible to use IP-based protocols on higher layers such as UDP, TCP as well as within Ethernet Powerlink networks.

Transmission of safety-critical data

For safety-critical data, the Ethernet POWERLINK can be expanded with the openSAFETY open safety protocol. With this, the safety-critical data are stored by openSAFETY with checksums and are then divided into two frames. The safety functions of the network are provided by the network’s own safety control mechanism. Secure and non-secure members can exist together in a network as well as exchange data that is not essential for the security function. OpenSAFETY is added as a protocol for the application layer. As such, it can be implemented on a high number of industrial Ethernet network topologies. OpenSAFETY has been tested by the TÜV and has been released for use with safety-critical data in accordance with IEC 61508 SIL 3 of the European standard 954-1.

The Data Format of Ethernet POWERLINK

Each packet is divided into a header and the actual data. This is implemented in a normal Ethernet frame, which can and must have a size of 64 bytes up to 1500 bytes. Jumbo frames, i.e. frames larger than 1500 bytes, do not exist and are not permitted in a Powerlink network. 0x88AB was assigned by the IEEE as an EtherType for the Ethernet POWERLINK. The Powerlink header is composed of one bit, which is reserved, seven bits as message type, eight bits as target node number and eight bits as source node number. For deterministic data transmission, collisions on the network must be suppressed.

For this, the data transmissions are controlled by a specific subscriber, the managing node (MN). The other participants, the controlled nodes (CN) may only transmit when they have been specifically requested to do so. A cycle starts with the message’s start of cycle (SoC). The managing node then controls all nodes individually with a poll request (PReq) whereby the controlled node is answered with a poll response (PRes). The answers are sent as multicast. This way, they can be accessed by other Powerlink devices. In this way, lateral traffic between the CNs is fully possible. In order to keep the cycle times as short as possible, not every device has to be controlled in one cycle. The asynchronous phase with the start of asynchronous (SoA) starts after the completion of the cycle. Now, in each case, a CN specified by the MN can transmit non-cyclic data. In this phase, data from a normal, non-deterministic network and from the Powerlink network can be exchanged via special gateways.

Profiles were defined for a number of device classes. These define the structure and the functionality of the directory for the various devices. Greater independence from manufacturers is obtained through the use of devices with specific profiles. At the same time, the Ethernet POWERLINK uses the CANopen profiles. The transformation rules specify the objects that are to be used with this.

Versions of the Ethernet POWERLINK

Currently, there are two different versions of the Ethernet POWERLINK. The first version, the Ethernet 0x3e3f, is a proprietary approach by B&R, which was used at an early stage as the basis for further development. On the other hand, the second version (Ethernet 0x88ab) is the current standard of the EPSG. It has been expanded through the addition of various characteristics. Both systems have many characteristics in common. In spite of this, the first generation is only a temporary solution.