July 2009

The Rosetta stone of controller tuning 

By David W. St. Clair

It has been a long time. Most persons practicing in the field today were not even born when it happened. I was only 18 and hadn't a clue.

It was in November 1942 when the transactions of the American Society of Mechanical Engineers published the seminal paper on controller tuning, Optimum Settings for Automatic Controllers, by J. G. Ziegler and N. B. Nichols. 

Their paper actually came out in December 1941. I first learned of them when I went to work (fresh out of college) in the instrumentation group at Eastman Kodak, in Rochester, N.Y., in 1947. Nichols was director of research for Taylor Instrument Co., which was also in Rochester, so we met.

How does the saying go? If I knew I would live this long, I would have taken better care of myself. What I write here will be from memory, not from notes, and at my age that can be chancy.

My boss was Gerry Akins. He and the instrument maintenance supervisor, Frank Kowalski, prepared the first movie for ISA, with an example of controller tuning. I wonder if the film still exists. Anyway, I had gotten the job because my course work showed I had taken a course in servomechanisms, which are control loops.

I think it was the first college course given on the subject. This was a subject heavily related to the control of anti-aircraft guns developed during WWII. It involved a lot of math … a lot of frequency response analysis. Nichols had been involved in some of that work. One of the tools used in the trade, called the Black-Nichols chart, bears his name. It did not really help me tune controllers. The Ziegler and Nichols (Z/N) paper did.

As I remember Gerry Akins telling me, there were about three or four self-proclaimed experts in the industrial controls field in 1942. At first, the Z/N paper met great skepticism. Some time later, there was a contest among the "experts" to tune a particular control demonstration, and the Z/N method won easily.

By the time I entered the stage in 1947, industry was finally accepting the Z/N approach for what it was, the Rosetta stone of controller tuning. I think most, and perhaps all, of the self-tuning controllers, now 67 years later, base their algorithms on the Z/N controller tuning principles.

While Nichols was in Rochester when their paper was written, I think Ziegler was there for only a short time. For most of his employed life, he was in sales for Taylor Instruments on the west coast. I did not meet Ziegler until 1990, when Larry Braun, of Chevron in Richmond, Calif., introduced me.

I had a son in the area we saw once a year, and that made it possible to meet Ziegler in person. He had retired from Taylor and had his own business, living near Braun. What a treat! Nichols had long since moved on from Taylor Instruments and was then with Aerospace, in El Segundo, Calif. Braun said the two kept in touch and tried to get together every year. Wasn't that great!

Ziegler and Nichols both died in1997.

It is interesting to ponder what the course of industrial controller tuning might have taken had not Ziegler and Nichols written their paper. Of course, we will never know. Would we have been trying to teach frequency response analysis, or phase-plain analysis? Certainly. I was doing it (servomechanism training) even after I understood Z/N.

Heaven forbid if we were dependent on the Routh stability criterion. Would we have had self-appointed experts for another decade, all disagreeing on how to go about it? Possibly. Certainly, after the war we were in an age of enlightenment, so my bet is something would have come along.

Yet the two tuning methods provided by Ziegler and Nichols, the open loop and the closed loop, were both quantum leaps forward. We are so lucky!

ABOUT THE AUTHOR

David W. St. Clair (dwstclair@aol.com) has been a member of ISA since 1947. He is author of Controller Tuning and Control Loop Performance: "PID without the Math" and there are over 30,000 copies in print. He is principal of Straight Line Control (www.straightlinecontrol.com).