Where’s the disconnect?

Control applications have changed significantly in the past decade from PID, CNCs, and PLCs to systems running advanced control algorithms, sporting sophisticated GUIs and built-in intelligence, and offering an extremely high degree of programming flexibility. However, much of what is taught in undergraduate and graduate controls courses is still not implemented in the bulk of smaller, simple c...

Control applications have changed significantly in the past decade from PID, CNCs, and PLCs to systems running advanced control algorithms, sporting sophisticated GUIs and built-in intelligence, and offering an extremely high degree of programming flexibility. However, much of what is taught in undergraduate and graduate controls courses is still not implemented in the bulk of smaller, simple control systems such as those employed on the manufacturing floors. In many cases, only complex and expensive systems employ the control techniques learned at universities.

For example, tuning a PID using the Zeigler-Nichols technique is an approach that all students learn in their undergraduate controls courses. Yet, tuning a PID loop on a process controller is usually accomplished by a seasoned veteran “tweaking” the proverbial gain knobs. When was the last time that you used root locus or drew a Bode plot? Odds are that it was some time ago.

Indeed, it seems that, upon graduation, someone presses the Ctrl-Alt-Delete key sequence when it comes to control theory. Why is this? Where is the disconnect between the undergraduate and graduate controls classes and application in industry? Much of this disconnect is due to the lack of pragmatic tools that allow analysis as well as implementation. However, new tools from companies such as National Instruments (LabView) and Mathworks (Matlab) are providing a gateway between analysis and application. One does not have to look too far on the production floor today to find evidence of a variety of systems running programs with excellent GUIs, such as LabView. Indeed, LabView has revolutionized data acquisition and process monitoring with its programming language.

For example, programming a Fourier transform or performing a power spectrum analysis with a Hanning window is as simple as clicking on a few blocks and connecting or “wiring” them together.

Ten or 15 years ago, such a task required an individual who was capable of real-time programming and also knowledgeable about the real-time implementation of the specific algorithms. Matlab provides engineers with an extremely strong set of analysis tools for a wide variety of controls and modeling topics. With Mathworks’ Real-Time Workshop, a Simulink model can be directly downloaded to a DSP or embedded processor. Clearly, today’s controls engineers have some powerful and easy-to-use tools at their fingertips.

So what does this all mean? It means that the job of the controls engineer will become easier due to these new tools that integrate analysis and application. At the same time, these jobs will become more demanding as more is expected from the engineers because these new tools are available. Just as multi-media technologies have increased students’ interest and made chalk board lectures a thing of the past—new controls tools have increased engineers’ and will make ladder logic programming, manual PID tuning, and numerical outputs a thing of the past.

The future holds much simpler graphical programming, automatic and advanced tuning implementation, and colorful and informative graphical outputs.

So, is there a cloud to this silver lining? We must be careful to continue to understand the fundamentals of control engineering and what we are doing. We cannot treat controls as a complete black box where we let computer programs tell us how to monitor and control our systems. They are excellent guides and tools, but we still must have a good foundation in control theory to properly use them. Will these new tools be taught in our classes? You bet they will! However, I guarantee that our students will still learn about Evan’s root locus plots, Bode plots and Nyquist plots, as well as all of the other tools that are so important to understanding the basis of controls engineering.

Author Information
Thomas R. Kurfess is a professor at Georgia Tech’s George W. Woodruff School of Mechanical Engineering.