The End of Cyclic Communication…

It’s amazing to think how life has changed over the last ten years. Today, I make extensive use of Yelp, a banking app, a GPS app, Uber, and other apps that I never even conceived I’d need in years past. A phone was a phone and now it’s an Internet connectivity device connecting me to all sorts of data in the Cloud.

It’s hard to predict the general trend of commercial technology – I can’t tell you if online grocery shopping will ever take off and how well driverless cars will be accepted. To do that, you have to predict the feelings and emotions of masses of people.

But I do think I know the direction that industrial automation is taking. There are a lot of new technologies that are being talked about and are starting to be implemented. There’s a push for open standards, Microsoft with its Azure product seems to be pushing for control of industrial data from the Cloud, sensors are getting more connected, and on and on. I could use this whole article to just talk about all the new technologies, requirements, cultural changes, security issues, and more that are changing how we do business on the factory floor.

But there are a few things I am absolutely certain about. The first is that the physical layer for all factory floor communications in the future will be Ethernet. That’s locked and loaded. I see a steady decline for Modbus RTU, DeviceNet, Profibus DP and anything else that isn’t Ethernet based. Modbus RTU just doesn’t have the data modeling or throughput. DeviceNet and Profibus DP both require special connectors and hardware. Those won’t do in the future as platforms are going to be off-the-shelf Windows or Linux boxes. I know quite a few 20-somethings that want to use a Raspberry Pi in industrial applications (that’s wrong headed, but I digress). With the speed and ease of use of Ethernet, there’ll be a steady decline of DeviceNet, Profibus, and many, if not all, sensor bus and I/O systems.

Not only will we be doing Ethernet in the future, but network speeds are going to take gargantuan leaps in the future. We will get network speeds of 100, 200, or 400 gigs. At those speeds, we will have almost infinite bandwidth. Everything – video, audio, sensors, control data, financial data – will all run over the same wire protected by the same security systems, maintained by the same people. There won’t be a factory network, an information network, an enterprise network; there will just be the company’s network. Controllers will be talking to motors over the same network that accounting is running its end of month reports on.

With all that bandwidth and speed, the one thing we won’t need is cyclic communications. Cyclic communication over Ethernet, as implemented by EtherNet/IP and ProfiNet IO, was adopted for two reasons. One, we needed to move short (hundreds of bytes) bursts of data pseudo-deterministically with as little jitter as possible. To do that we put controllers and I/O on segmented Ethernet networks with little or no other traffic. Two, controllers of the 1990s were under-resourced compared to controllers of today’s factory floor. These controllers were incapable of ingesting and sending large volumes of data. EtherNet/IP and ProfiNet IO came of age in this era.

But that time is over. We now have extremely fast and well-resourced platforms and are moving into an age where we will have extremely fast networks that are only going to get faster and to provide nearly infinite bandwidth. This signals a move to more of a Publisher–Subscriber model of communication between controller and I/O data. The advantages of this model are the ability for controllers to pick and choose not only what data they want to get but how often they want to get it, and to modify that as events warrant. It is an entirely new way of operating that is more flexible and adaptable to the networks of the [very near] future.

In the Modbus era, cyclic communications were new, novel, and different. Now the cyclic communication era is over. It’s served its purpose well, but like every technology before it, it’s now being replaced by something faster, more flexible, and scalable.

That’s the cycle of life. The only thing for certain is that things will change.