Controldesign.com is revisiting an article that they published in February 2001 where they debated PLC’s versus PC’s in industrial settings. I saw the latest update on this topic today. I remember the debate raging with several customers when I went to work for Rockwell Automation as a sales engineer in the fall of 2001. I remember Steeplechase Software (now a part of Phoenix Contact) having a marketing video with a PC on fire and the machine still running as a show of robustness. I could only think to myself that if the PC was on fire and I had no way of shutting the machine down, then I wasn’t convinced that was a good thing that the machine was still running. Rockwell of course, had a “me too” product with SoftLogix and I remember selling one of them on a test system at a plant when they were replacing a homegrown VB based test system that had integrated LabView and an SLC500 processor.
We at LSI actually sell and use PC based, PLC based (PAC also if you are into that acronym), and DCS based controls and we are of the opinion each has its own merits. When we are selling our test systems (DALS Solutions), PC based control is the only way to go and we actually talk customers out of a PLC based solution because of speed, configurability, and database integration. PLC’s just don’t make sense. We also believe that there are many more mission critical tasks that PC’s just have no business trying to control. It is simply a matter of picking the right tool for the right job. Why would you use a screwdriver to hammer a nail?
The right tool for the right job is critical for any job you are doing whether it is a controls project or just something around the house.
I also take issue with a few statements made in the article as I think there have been some advancements over the last 10 years that the author has overlooked. Note the statement below:
PLC proponents have long claimed that PC-based systems running a Windows operating system (OS) are not a good solution for real-time operations because a PC is not deterministic, i.e., you cannot guarantee a response time to interrupts and tasks. A PLC, on the other hand, has a guaranteed scan time, ensuring its I/O and interrupts are serviced on a regular basis.
Recent developments in Windows and other operating systems may be close to making this a non-issue. Even worse for PLC manufacturers, some people think that PCs are now more responsive for real-time control applications than PLCs. Isn’t that heresy?
This stems from the PLC’s penchant for processing relay ladder logic sequentially. Historically, a PLC started at the top of a list of logic. It refreshed its I/O, and then ran through the entire sequence from top to bottom. This is called a scan time, and it is fixed. Many PLCs still run this way; others may use different languages such as Function Block or Flowchart to avoid running all the logic.
In a PC, scan time is not a limiting function. A PC’s non-ladder logic control programs are often based on interrupt service and changes in I/O, so they don’t have to run through all the logic every scan cycle. Instead, they deal only with those portions of the software affected by the change in I/O. They can accomplish more in the same processing time because they aren’t evaluating all the logic all the time.
If you look at just the ControlLogix system, you can do interrupts, prioritize logic to be scanned based on a deterministic rate, or do traditional continuous scans as mentioned. That is all part of the design of the system and one that we take advantage of every day.
I know that many other processors have the same ability in the marketplace today. For that matter, you had selectable time interrupts many years ago in the PLC5 product line. So, I think that there needs to be some clarification added to the article.
I do, however, agree that there is some convergence between the technologies with solid state hard drives in PC’s; prevalent Ethernet communications on PLC’s; Webservers, Historians, and Databases on solid state devices in PLC racks and the like.
I don’t think that the situation today is a tossup as the author concludes as I see very little PC based control on the plant floor, except in highly specialized applications like test systems. The technology has converged some in the last 15 years and I am curious as to how the debate will change over the next 15. Will we see tablets on the plant floor? Will they replace PC’s and laptops? It is hard to say, but I know one thing, technology change is here to stay. Get used to it.