Consider a sensor network to ease connections
Do industrial networks meet expectations? Bypass increasing Ethernet and other network incompatibilities at the I/O, sensor, and safety level by using a widely accepted sensor network.
Helge Hornis, PhD
Using a sensor network can save time, money, and steer clear of increasing Ethernet protocol and other network incompatibilities at the I/O, sensor, and safety device level.
Divergent PLC protocols
During the late 1990s nearly every PLC manufacturer decided to develop a networking technology suitable for industrial applications. Modbus, Profibus, CC-Link, DeviceNet, and many others were born in short order, each promising to address the fundamental requirements controls professionals had at the time: reliability, simplicity, and deterministic real-time behavior at a price that was competitive with conventional hardwiring. This led to confusion mainly because all approaches were not interoperable.
Customers preferring a particular brand of PLC were essentially forced to use the networking technology developed by that manufacturer. If this situation was not bad enough, the whole story was repeated a few years later when each of those PLC manufacturers decided it was time to promote new networks. This time the underlying wire was Ethernet. I will not claim that one such solution is superior to another, but instead ask readers if having all these options (or should I say edicts) was such a good idea. From the point of view of device manufacturers (RFID systems, drives, HMIs, etc.), it was not.
Instead of focusing on a small number of communication interfaces for RFID systems, device manufacturers had to develop a plethora of solutions, such as DeviceNet, Modbus TCP, Profinet, EtherNet/IP, Profibus, CC-Link, and the list goes on. The same was necessary for encoders, camera systems, and many other components needed to automate a complex machine. And to make matters worse, going forward, even those protocols that share Ethernet at the physical layer will need dedicated (that is, incompatible) hardware.
Initially, the situation was not all that bad. For instance, until recently we could manufacture an RFID controller that could host four noncompatible communication protocols (Modbus TCP, TCP/IP, Profinet, and EtherNet/IP) simultaneously on the same hardware device. Customers could buy one unit, and, without making adjustments or modifications, connect it to any of the four Ethernet-based networking technologies, and control it from a PLC. In this case fewer options translates into streamlined stocking, better availability, reduced ordering errors, and enhanced familiarity when it comes to installation.
In the near future, due to certain protocol changes (promoted as advancements), this will no longer be possible. The same Profinet unit will require dedicated hardware as will the system using EtherCAT. For the time being, EtherNet/IP and Modbus/TCP can coexist on the same hardware. At this point, one may ask if it offers any advantage that those networks are based on Ethernet. In a world where engineering resources are free and deadlines do not exist, this would not be an issue. But in reality, duplicate engineering drives up costs at all levels. Is this really necessary?
Interestingly enough, a solution exists at the sensor and machine safety level that is accepted and supported by most PLC manufacturers. While the solution is not suitable to interface devices that interchange larger amounts of data (RFID systems, drives, HMIs, etc.), it is an ideal method for bringing binary devices and safety components (like e-stops, door safety switches, and light curtains) to the PLC and give the PLC a way to control simple binary outputs (including valves, horns, and indicators).
In contrast to a multitude of Ethernet solutions, this system was (no surprise) not developed by a PLC manufacturer (and then made open) but instead is the result of a joint development effort among 11 industrial automation companies that was open from the beginning. After the basic technology was published and released, others joined this group. Involved companies are known for their PLCs, sensors, valves, and/or safety components. Introduced in 1994, AS-Interface is now the world’s most successful low-level networking technology, based on information from the [[get exact name-AS-Interface group]]]. With nearly 20 million installed field devices, it is the closest thing to a universal networking technology in the automation space. Approximately 300 vendors offer:
- Binary I/O modules – These modules allow the switch state of any conventional sensor to be brought to the PLC. AS-interface is a real-time and deterministic technology with a worst-case sensor update time of 10 ms. (Worst-case means that 248 input states are communicated. Fewer I/O connections means faster updates.)
- Analog I/O modules – Most industrial control systems have roughly twice as many digital inputs as outputs plus a small number of analog signals. These types of analog signals tend to be slow compared to binary data. AS-Interface allows such signals to be transmitted alongside the binary data with update times of 35 ms or less.
- Indicators and buttons – Indicators and buttons are another type of I/O signal. Timing is not the main concern, but installation simplicity is. For instance, a four-element stack light is connected in seconds using just two AS-Interface leads.
- Functional safety – For more than 10 years, networking functional safety devices has been the “killer app” of AS-Interface because it addresses the cost issue (it is much cheaper than a safety PLC) and the simplicity issue (it exploits AS-Interface installation advantages). In most cases, about 90% of the wires needed when constructing a hardwired safety system can be eliminated. Suddenly, designing a safety system is a simple, two-step process. First, the hardware is connected to AS-Interface, and then it connects to required logic using drag-and-drop operations.
Machine builder advantages
Machine builders appreciate other benefits. Because AS-Interface can be connected to a large number of PLC backplanes and an equally large number of industrial networks, an I/O and safety system designed for a machine controlled by a PLC from manufacturer “A” can easily (and without modification) be reused if a PLC from manufacturer “B” is used the next time. This feature makes navigating the maze of competing and noncompatible upper-level Ethernet solutions easy. Only one gateway between the PLC and AS-Interface needs to be swapped out.
End users have enjoyed forward and backward compatibility of the network, which is one of the guiding principles of the governing organization’s member companies. For instance, if a module on 19-year-old network fails, it takes, on average, less than one minute to replace it with a new design. And it is not even necessary to use a product from the original manufacturer. The old system may not be able to use the latest features of the new module, but it will work just as well as the old part did. Similarly, if a decade-old system that feeds into DeviceNet controlled by an older PLC now must operate as part of EtherNet/IP on a new PLC, only the gateway has to be replaced.
It can hardly get any easier. And isn’t technology supposed to make automation simpler, better, and less costly? The industrial Ethernet experiment has not failed, but it unfortunately has not lived up to its potential and customers’ expectations. At least at the I/O connection and safety level, users have a choice that is universally accepted, highly interoperable, and compatible with most PLCs on the market.
- Helge Hornis, PhD, is manager, intelligent systems group, Pepperl+Fuchs. Edited by Mark T. Hoske, content manager, CFE Media, Control Engineering and Plant Engineering, mhoske(at)cfemedia.com.
Networking articles: www.controleng.com/integration
Networking products: www.controleng.com/networks