In this age of automation revolution – technologically, culturally, and managerially – it’s critical to have a strategy, and that applies no less to the good ole sensor as much as anything else. As a sensor manufacturer, you’d be well positioned for all this change by having a dual strategy combining traditional sensor interfaces for today’s communications networks, and Ethernet to support tomorrow’s manufacturing processes. But first, a short history of sensors…
For 50 years or more we’ve been deploying sensors on the manufacturing floor. Type K Thermocouples, photo eyes and proximity switches had to be some of the first sensors ever used in manufacturing applications. In the early days, miles and miles of wires were strung from sensors back to relay contacts, which powered motors, heaters, diverter gates and everything else you can imagine. Control programs were simply banks of these relays wired using Boolean logic. Crude, expensive and labor intensive, but effective in those prehistoric days.
When PLCs were introduced, sensors were connected to PLC input modules instead of contacts. Those discrete input modules supported 0-20ma or 0-10V inputs which forced the standardization of those interfaces. The rapid growth of PLCs and automation expanded the demand for sensors and soon there was a sensor for every kind of measurement. Sensors became available for temperature, pressure, tank level, vibration, presence, thickness, length and everything else using all sorts of new technologies.
Eventually, when factory floor networking was introduced (Profibus, ASI, DeviceNet, Modbus), sensor data could be collected at the machine. Instead of wiring sensors all the way back to the PLC input module, they could be wired a short distance to a remote input module which reported the sensor status back to the PLC over a DeviceNet, ASI or a Profibus network interface. But the sensor interfaces never really changed. Instead of terminating wires, new M-type connectors were introduced with protections like IP67 and various locking styles that simplified connection to remote input modules.
As more complex sensors were introduced, those sensors incorporated technology to directly connect to controllers over a network. Complicated systems for color detection, reflectivity, sophisticated temperature detection, and vision scanning directly incorporated hardware bus technologies (ASI, Seriplex, IO-Link), sensor bus technologies (DeviceNet, Profibus DP, Modbus), some Ethernet (Modbus TCP, EtherNet/IP and ProfiNet IO) and even wireless (802.11 and 802.15 mesh).
But even with all the new ways to connect devices to controllers, most sensors deployed in manufacturing systems are still the garden variety: 0-10V, 20ma, PnP and NpN sensors (temperature, pressure, proximity, etc.) As you can see, not much has changed over the past few years. The vast majority are connected the way they’ve been connected for the last 40 years. At least for now.
The Revolution
In 2017, the automation industry is facing a host of new requirements, including new technologies, new data analytics tools and data storage requirements, a convergence of IT and OT and more. We’re in the midst of revolutionary changes and this is going to affect sensor technology. My crystal ball can be pretty cloudy, but there are some things of which I’m sure:
Ethernet is the dominant technology for the foreseeable future. We aren’t seeing the last days of ProfiNet IO, DeviceNet, Modbus and the rest, but new systems mostly won’t use these technologies.
Security is a top concern. The Director of Homeland Security has even ventured to say that manufacturers should disconnect their critical machinery from external and internal networks.
IT and operational technology are going to converge. IT will have more and more influence on manufacturing automation architectures. There will be a day when the accounting team will be on the same network as drives, sensors and manufacturing system controllers.
Manufacturers want more from not just sensors, but from every device in their architecture. The temperature of an on/off status from sensors isn’t going to be good enough. People expect more intelligence, more functionality, more data and more connectivity to feed into those advanced analytic tools.
Deterministic Ethernet is on the horizon. Bandwidth, throughput and speed are all going to increase exponentially. I find it hard to imagine, but I have heard 400gbit deterministic Ethernet networked mentioned!
Open platforms and open standards are going to find their way into manufacturing applications. Manufacturers are tired of proprietary anything.
Ethernet is migrating to constrained devices (more on this below).
A Plan for Sensor Manufacturers
No matter what kind of sensor you manufacture, customers are going to demand more data from it. Anything else you can supply (cycle counts, identification data, maintenance data) is going to be valuable. So, what should you do if you’re a sensor manufacturer? The first alternative is to simply do nothing. Systems have been deployed for fifty years with 0-10V and 20ma sensors and that will always be an option.
Another alternative is to build Ethernet connectivity directly into your sensor. Even though the cost of microprocessors is falling, the additional cost to add Ethernet connectivity just doesn’t make sense for the vast majority of low end pressure and temperature sensors. Too much cost and too little value.
The future of Ethernet in this environment isn’t known (baud rates, determinism and more are changing the face of Ethernet). How do you manage Ethernet technology issues as Ethernet technology develops? And what about Ethernet communication protocols? The hardware may be the least of your worries.
You could just add a sensor technology like Profibus DP or DeviceNet or Modbus to your product. That buys you some time – those technologies aren’t going away today but the trend is Ethernet. Why go with sensor technologies?
Low Complexity Ethernet
Another idea is to wait for something called Low Complexity Ethernet (LCE). LCE is a coming standard that RTA and others are developing that is a middle-of-the-road solution. You avoid the expensive hardware, but get Ethernet and all its benefits. LCE offers the ability – with the addition of a few, inexpensive, integrated circuits – to add Ethernet to your product. It’s something that’s being prototyped right now and may be available as soon as Q1 of 2018.
If you’re a sensor manufacturer, my recommendation is a dual strategy with traditional 20ma/10V/nPn/PnP interfaces for today’s communications networks, and Low Complexity Ethernet to support tomorrow’s conversion to manufacturing processes that are all Ethernet. Low Complexity Ethernet offers inexpensive Ethernet connectivity and provides you with the ability to support the additional data that customers require, without having to add costly parts and TCP/IP stacks to your product. It’s not a technology that is available today, but it’s in development right now with a promising future.
You can contact me for more information on LCE. We are actively looking for partners who want to prototype low-cost sensor connectivity over Ethernet.
