Implementing Industrial Ethernet: The challenges of connecting a multiprotocol network

Implementing Industrial Ethernet: The challenges of connecting a multiprotocol network

By Andrew McIntyre, Texas Instruments

Originally promoted by the German government, the term Industry 4.0 describes a fourth industrial revolution currently taking place. The first industrial revolution came with the advent of steam-powered machines. The second wave dawned with the implementation of mass-production assembly lines and machines running on electricity. The third followed the introduction of electronics and information technology (IT) into the manufacturing space. In the fourth revolution, Industry 4.0 is defined by networks of connected devices able to share information, create a virtual copy of a physical space and make decisions in real time.

At the core of Industry 4.0 is the need for reliable communication networks. These networks are closely tied to real-time operational needs; therefore, they must be able to handle information in a deterministic and redundant manner. Networks must scale across a wide range of equipment operating on multiple different standards with a variety of interfaces.

Many challenges stand in the way of implementing a factory system capable of this kind of performance, but rewards like faster response times, increased efficiency and greater reliability can be well worth the work. To realize these benefits, designers must resolve issues including selecting the best communication technology, navigating a multiprotocol environment and covering a wide range of industrial equipment.


Communication technologies

Each communication technology has its own characteristics, so achieving performance standards requires a blend of multiple technologies. It is important to consider the strengths of each technology when looking at the various applications in an Industry 4.0 factory.

Wireless technologies like Wi-Fi® are good for communicating across an entire factory and for connecting machines to the cloud. New low-power wireless solutions are making it possible to put sensors in places that weren’t feasible before, and technology like Sub-1 GHz communication extends the range of these devices even farther.

Near-field communication (NFC) can provide quick interface points for operators walking around a factory to interact with machines via tablet devices. Radio-frequency identification (RFID) tags allow for the tracking inventory moving around a factory.

Historically, wired field-bus technology handled the most critical operations within a factory due to its ability to meet real-time standards. Now, it’s possible to implement industrial Ethernet in these same applications. A number of protocols already exist in the industrial Ethernet space to manage deterministic communication. However, as demand and applications for this type of communication continue to increase, standards like Time Sensitive Networking (TSN) which is an approach to unify vast variety of industrial Ethernet standards, are helping guide future development. Also, industrial Ethernet is branching into new areas of the factory, including motor drives, input/output (I/O), and sensors, to connect more data than ever before into the network. Assessing the functionality of each system within a factory is important to ensure it is matched with the most efficient communication technology for its purpose. Figure 1 below demonstrates different advantages and disadvantages of some communications technologies.

Figure 1: Each type of communication technology has its own set of advantages and disadvantages. Consider these attributes carefully when implementing a communication network.


Multiprotocol environment

Although more than 30 industrial Ethernet protocols are available, these five are most commonly used in factory automation settings:

  • EtherCAT – optimized for process data and scalable across a wide range of equipment, from large programmable logic controllers (PLCs) down to I/O and sensor-level devices. Capable of connecting up to 65,535 nodes in a system with low latency in each slave node, EtherCAT provides a versatile and cost-efficient solution for implementing industrial Ethernet.
  • PROFINET – a widely used industrial Ethernet protocol with three different classes of latency standards. PROFINET Class A is common in infrastructure and building automation and has a cycle time of 100ms. PROFINET Real Time (RT) reduces cycle time to 10ms with a software-based real-time approach and is well suited for factory and process automation. PROFINET Isochronous Real Time (IRT) is designed for motion-control operations and uses special hardware to get cycle times down to less than 1ms.
  • Ethernet/IP – an application-layer protocol on top of TCP/IP compatible with many internet and Ethernet protocols. Ethernet/IP uses Common Industrial Protocol (CIP) over TCP/IP which provides a standard set of services and messages for industrial automation control systems. Ethernet/IP also uses standard Ethernet switches and can have an unlimited number of nodes in a network. This allows for full factory coverage on one network; however, it has limited real-time and deterministic capabilities.
  • POWERLINK – can be a pure software solution implemented on top of existing applications processors and Ethernet connections or can utilize hardware accelerators to achieve more stringent timing requirements. POWERLINK is capable of running a range of systems, from communication between PLCs down to I/O and motion control. Given its open-source software stack, the barriers to implement POWERLINK are low.
  • Sercos III – often used in servo-drive controls, with cycle times as low 31.25 microseconds. Sercos III operates in either ring or line topologies with up to 511 slave nodes in one network.

With several different protocols in use, it can be a challenge to design devices compatible with the right industrial Ethernet standard. A multiprotocol industrial Ethernet device – one example is Texas Instruments’ Sitara AMIC110 system on chip (SoC) – can provide an advantage in this regard as it integrates the industrial Ethernet function into the SoC, making it capable of running all of the above standards by only loading a different software during startup. This enables communication with a broad range of devices and applications. Multiprotocol industrial Ethernet devices help reduce overall system cost by eliminating the need for external ASICs, and they can also greatly extend the potential markets for a device by allowing it to fit into diverse networks running on different protocols. Even more flexible, a software-based multiprotocol device can adapt to additional industrial Ethernet standards over the life of the product.


Range of industrial equipment

In order to connect an entire factory, gaps must be bridged between a variety of machines with different functions, technologies and life cycles.

Sometimes this is a straightforward process where individual drives and sensors feed data to devices like PLCs which in turn pass information along to the operator level with its interactive human machine interfaces (HMI) and controls. In this case, a series of industrial Ethernet networks can effectively link a factory together. Connection can be achieved by simply inserting communication nodes at each level. This process can be as simple as adding an SoC device at each node with an incremental cost and learning curve.

However, bridging the gaps can also be a convoluted process that requires a blending of communication technologies and network topologies. Noncritical networks can get away with employing leaner processes with lower reliability and determinism; whereas, safety-critical networks require the redundancy that comes with larger, more complicated networks. When looking at retrofitting older equipment to interact with more modern systems, determining the most efficient solution can be quite challenging. For example, is a single-function external sensor sufficient, or is it necessary to integrate sensors with the internal functions of the machine?

Regardless of the level of complexity necessary to connect a factory, it is best to find a solution that uses as few components as possible. SoC devices that integrate multiple functions into one part are preferable because they reduce the number of devices and failure points in a system.