Sunteți pe pagina 1din 7

The Root Locus Paper

Walter R. Evans and the Story of Root Locus


On Saturday, May 7, 1949, at the Polo Grounds, the New York Giants delivered a 9-1 drubbing
to the St. Louis Cardinals, hometown team of a twenty-nine year old electrical engineer and
baseball fan, Walter R. Evans. A third generation St. Louisan, Evans now lived with his wife
Arline (also third generation St.Louisan) and two young sons, Randy (age 4) and Gregory (age 1)
in Whittier, California, about 15 miles east of Los Angeles.
Despite Saturday being a day off, the impatient Evans was mad. His anger had nothing to do
with the 9-1 drubbing and everything to do with a frustration born of six months of waiting. In
December 1948 he had submitted his second servomechanisms
1
paper to the American Institute
of Electrical Engineering (A.I.E.E.). Just as had occurred in 1947, he had become frustrated by
the agonizingly slow review process. He had already missed out on the 1949 winter general
meeting because one critic complained about the papers length. Now, two months since he
resubmitted a shorter version, he remained in the dark about its acceptability. Would it require a
year, as it had with his 1947 graphical analysis paper, before his new root locus idea would
receive a hearing at a general meeting? Were open-minded reviewers assigned his papers, or, as
he feared, were close-minded ones critiquing it? At work, open-minded engineers gravitated to
his new idea; some senior ones seemed annoyed at being asked to think differently.
He sat down and vented his anger toward the one person in whose hands he believed the fate of
his paper sat, Professor Gordon S. Brown, founder and director of MITs Servomechanism
Laboratory and chairman of the Technical Program Committees servomechanism
subcommittee.
The A.I.E.E. has now had the Root Locus paper for nearly six months. In my humble
opinion this pathetic record for an Institute which is alleged to have dissemination of new
theories .. as one of its fundamental aims. This is a subject on which I might write in
undeniably clear language if there were reason to believe it might do some good.
He finished the draft of the letter and set it aside for the time being.

1
Servo in servomechanism means slave in Latin. One example of a servomechanism is the
complex machinery that causes the Hubble Telescope to point exactly at a particular galaxy when an
astronomer enters the galaxys coordinates. The servomechanism tries to reduce to zero the difference
between the direction the astronomer has told the telescope to point and the direction its sensors tell it is
pointed. Machines take time to respond to commands, however, and these delays induce an oscillatory
behavior. Servomechanism designers want to the amplitude of these oscillations to always decrease
over time. This behavior is stable. Oscillations that grow over time are unstable. Evans root locus
method provided a new way for engineers to be confident their designs moved into stability.
2
That same Saturday, about ten thousand miles away at the Kapustin Yar launch site, Soviet
scientists celebrated the first successful launch of their R1-A Scunner rocket, a version of the
German V-2 rocket that had been used with devastating effect in the war. They had overcome
their most vexing problem -- flight control -- to achieve a successful 270 km flight. Five more
successes would follow in the month of May.
Although this Soviet success was, of course, unknown to Evans, he was well aware of Americas
competition with the Soviet Union. Three years had passed since Winston Churchill had
famously declared in Fulton, Missouri, From Stettin in the Baltic to Trieste in the Adriatic, an
iron curtain has descended in the continent. In May 1949, memories of the US-Soviet alliance
that had brought victory over the axis powers were fresh, but relations between the former allies
had grown tense. Less than a year before, the Soviets tried to block western access to Berlin.
Unlike the world war recently ended, this new cold war would not be led by generals nor
fought by soldiers on battlefields. Rather, it would be led by scientists and waged by engineers
in their respective laboratories and rocket test ranges. Evans was a new recruit on the front line
of this war, having emigrated nine months previously from St. Louis to join North American
Aviations Aerophysics Laboratory in Downey, California.
Four years before, as the war was coming to an end, North Americans president James Howard
Dutch Kindelberger had formed a small, secret Technical Research Laboratory in Inglewood
for several staff to study rocketry. Once the war ended, the militarys needs shifted immediately
from high volume aircraft manufacturing to long-range ballistic missiles. By 1949, their
requirements for range had increased to 1000 nm on their way to 5500 nm and the requirements
for accuracy had become a very significant challenge. New technology was needed across a
wide range of engineering discipline, but none was more important than design of the internal
compasses that put the guided in guided missiles their inertial references and associated
flight control systems. Kindelberger and his successor as NAA president, J. Lee Atwood,
created a new larger organization called the Aerophysics Laboratory and challenged it to become
the best in every technology associated with ballistic missiles.
To lead this new organization they needed someone with proven leadership, managerial, and
technical ability. Atwood chose William C (Bill) Bollay, a former Ph.D. aeronautics student of
Caltechs Theodore von Karmen. John Moore, an associate professor of mechanics at
Washington University, joined the laboratory to form the engineering team to build small,
accurate, inertial guidance systems. Among the first people he sought out was a twenty-eight
year old engineer he had helped become an instructor at Washington University --Walter Evans.
In June 1948 Evans joined NAA for summer employment between academic years. He found
himself intellectually challenged and, given his growing family, attracted by the pay differential.
He returned to St. Louis and drove his family west, where they would settle in Whittier.
Evans found himself among an elite team of engineers working for John Moores organization.
North American Aviation (NAA) has a contract with the Army Air Corps to develop an
American version of the German V-2 rocket to become known as the NAVAHO. (North
American liked to see their products begin with the letters NA) Evans area of expertise
feedback control systems used by servomechanisms was key to solving the technical problems
3
that his Soviet counterparts had struggled to overcome automatic flight control of a ballistic
missile using an inertial reference system.
In early 1940s Evans and Moore had studied together while at the General Electric Companys
Advanced Course in Schenectady, NY. Through GEs technical library they followed advances
in principles of servomechanism design, including MITs latest prepublication papers distributed
in advance of A.I.E.E.s quarterly meetings. In professors Charles Stark Draper and Gordon S.
Brown, MIT had unchallenged leadership in the fields of servomechanisms and inertial guidance
systems. The American military had invested heavily at MIT during the war to develop the first
radar, and later in the antiaircraft fire control systems that used radars as sensors --military
examples of servomechanisms. Many years later, Evans kept a prepublication copy of a 1945
paper -- No 45-A-20 Dynamic Behavior and Servomechanism Design in his files. Its authors
were Gordon Brown and his associate, Albert C. Hall. In fact, Evans first reference in his root
locus paper was to Gordon S. Browns 1948 book on servomechanisms.
Later that same Saturday, May 7, Evans returned to the draft of his letter to Gordon Brown.
Should he send it? What good would it do? Evans mind churned. Creative but impatient by
nature, his thinking process seldom stuck on the first solution that came to his mind. Rather he
would continue to seek out an alternate perspective that might expose a better solution to a
problem. He knew that John Moore and Gordon Brown had met within the last few days.
Among the topics he knew they would have discussed was his root locus paper. He knew he
could count on getting a report from Moore soon.
We cannot know what was in his mind. What we do know is he decided to rewrite the letter with
a strained attempt at humor and a suggestion that amounted to a threat. Dr. Professor Brown,
he wrote again. Trying a dozen plays on words, he wrote, in part:
The root locus idea has been chasing around for six months in some multiple loop with no
visible output. The only feedback has been rejection and your recent letters. There must
be something nonlinear in the system. The obvious alternative is to submit the paper to a
different organization.
He sent this second letter airmail to Brown and scrawled, NOT SENT on his first draft.
Sunday May 8, would be a better day, at least for St. Louis. The Cardinals trounced their rivals
the Dodgers 14-5 at Ebbett's Field. The 1949 baseball season was still in its early stages. And
so, it would develop, was the review of Evans paper. The Cardinals, recent (1946) world
champions, led by 1948 batting champion Stan Musial, were hoping to at least win the pennant.
To do so, they would have to subdue the Brooklyn Dodgers.
Monday Morning, May 9, Evans drove down Imperial Blvd. from his modest Whittier home to
the Aerophysics facility in Downey. There he learned from Moore that the March 28 version of
the paper was still unacceptable to Brown. Moreover, Evans got the impression from Moores
comments that Brown did not yet see the value of his root locus method. Anxious to understand
what specifically Brown did not grasp; Evans wrote another letter to Brown before Brown would
have received his May 7 letter. Dear Prof. Brown, he wrote for the third time in as many days,
4
John Moore returned with the news that the Root Locus paper is still unacceptable to you.
Therefore I would appreciate a copy of the paper marked with your questions. He
finished the letter as he had the last one with I see no alternative to starting some action in
parallel with your committee.
At home that evening, he decided to take a more proactive approach by at taking his manual
typewriter and banging out four letters in one evening, all to former colleagues at GE and Wash
U Gordon Walter, Phil Michel, Louis (Doc) Radar, and Orrin Livingston. He could express
more freely to them what was really bugging him. Evans also realized he could use his friends as
conduits to get his idea into the institutions he had recently left by including a copy of his paper
with each letter sent. To Doc Radar he wrote,
Remember the low opinion that I had of the A.I.E.E. technical paper system in Pittsburgh?
Well, it has now sunk even lower. The enclosed paper was first submitted to the winter
meeting. Every bit of information I have received has been the result of needling them.
The fellows out here at the Aerophysics Lab at North American have adopted it in preference
to any of the standard methods. If you or any of the boys have any questions about the
method, I would be glad to answer them.
Tuesday, May 10, Gordon Brown received Evans letter of May 7. He responded with a carefully
worded letter that was both professionally polite and at the same time a stinging rebuke. He took
the unusual measure of asking his secretary to send the letter airmail.
Thursday evening May 12, Evans opened it Brown had enumerated his remarks from one to six,
starting five of the six with the word please. That he took offense at Evans comments is
apparent in item number four:
4. Please remember that if anyone seriously takes the job of reviewing manuscripts, they
spend a great deal of time in the interests of the authors professional reputation that is
purely voluntary and part of the code of professional ethics.
The sixth comment advised Evans Please talk to Mr. Moore upon his return to California and
get his advice before you let this matter go to far. Below the closing salutation, Evans saw that
he was not the only one to receive the feedback. Brown had sent both letters to three members of
A.I.E.Es Technical Committee and John Moore.
Browns structured rebuke left Evans was both contrite and angry. Immediately, he drafted a
point-by-point, typewritten draft response. To Browns fourth comment (quoted above) Evans
wrote,
You will perhaps excuse my overlooking the time volunteered by reviewers when you realize
the too long and show contribution to synthesis was essentially all I learned from the
initial rejection. The several hundred hours which I spend (sic) developing the method was
fun, but has been matched by now with write-ups and revisions. But most important, think of
5
the thousands of man-hours wasted by A.I.E.E. members on inefficient methods during a
publication delay like this.
By the time he finished the evening was late. He set the letter aside to sleep on it.
Friday, May 13: Evans followed Gordon Browns suggestion and went to John Moore to get his
guidance. This time Evans learned that Brown had told Moore that he himself had devoted hours
of his own time with root locus paper and had even considered rewriting it himself. From
colleagues he learned that they had complemented him on the paper in part because of their
knowledge of the hours Evans had put into writing it. Evans anger dissolved into a mixture of
contrition and personal embarrassment. He filed away yesterday evenings defiant draft and
composed a short letter of apology for his secretary to type and send to Brown. Dear Professor
Brown, he wrote for the fifth time in six days,
Nothing seems less funny than an attempt at humor which is out of place. I wish to apologize
for my letter of May 7, it was an ill-considered outburst releasing long accumulated
frustration. Any effort which you can make to ignore the recent letters would be
appreciated. I have been completely deflated. The only thing Im sure of at this time is the
sense of duty to revise the paper until acceptable.
Over the next two weeks, Evans did just that, and on May 29 sent the A.I.E.E. in New York his
rewritten paper. He dropped servomechanism, and chose the title Control System Synthesis
by Root Locus Method. He sent a copy to Gordon Brown at MIT.
The next letter he received came neither from the A.I.E.E. or Gordon Brown. Rather, it was
from GEs Orrin Livingston, who was responding to one of the four letters Evans had banged out
on May 9. Livingston confessed his bewilderment with the paper.
I struggled through a few pages am now referring it to some of the other boys in the hopes
they can explain it to me in words of one syllable. Incidentally, if you have anything of a
more elementary nature, we would like to receive a copy
By this time, of course, Evans had already made another pass at explaining root locus.
Although it was Brown, not Livingston, who controlled the fate of his paper, Livingstons
bewilderment bothered Evans; another idea began to percolate in his mind.
A week later, the letter carrying an MIT return address arrived. This time Browns comments
were considerably more favorable. Both Evans letter of apology and his rewritten paper had
had their hoped for affect on Brown, who wrote:
I am pleased to acknowledge receipt of the carbon copy of the final (italics added)
manuscript of your paper. I believe you have done an excellent job at working out the
suggestions made by the reviewers. I assume you sent. the necessary extra copies for
review.
Evans, of course, if he knew anything, knew to expect more review; he had sent copies.
6
Meanwhile, Evans worked up a response to Livingstons request for something of a more
elementary nature. The formal style expected of the A.I.E.E. papers had been a strain for Evans
to compose. His wife, John Moore, and other colleagues were drawn to him in part out of
respect for his intellect, but mostly for his sense of humor and naturally informal style. (Years
later, for example, he informed an applicant he was hired while playing touch football with him.)
Now, realizing he had never described root locus in a more informal way, he used Livingstons
letter as an opportunity to do so. By June 13 he had completed four handwritten pages he entitled
The Root Locus Idea, divided into paragraphs on Problem, The Classical Approach, The
Frequency Response Technique, Roots Their Lives and Habitats, and The Root Locus Plot.
The only extant copy is one Evans kept in his files; he never had it typed or published. In his
cover letter he included to Livingston, Evans wrote,
Ive long thought is would be a sporting idea to write up the root locus idea the way I
came to understand it free from approved terminology. Forget the synopsis and
introduction. Lets concentrate on the simple cubic system. Thats just tough enough to
illustrate the idea and no more. Incidentally, it took from 1946 to 1949 to hit upon the idea
and all the rest was worked up in 3 weeks so Im sure you can work out all the rest anyway.
Good luck with the enclosed write up.
That summer Evans put aside any further formal work on his paper and turned his attention to
developing a book about root locus. In a letter to John Wight, Editor of Engineering Books at
McGraw Hill in New York, Evans explained that the main purpose of the book is to
demonstrate the root locus method. The importance of his work at Aerophysics Laboratory was
surely brought home to him and all of Bollays NAVAHO missile team when word came that the
Soviet Union had exploded their first atomic bomb Joe 1 on August 29
th
on the steppes of
Kazakhstan in Central Asia. It was a virtual carbon copy of the American weapon. The Soviet
scientists had received Americas recipe from sympathizers and spies. Three days after the
Soviet atomic test, Evans signed a memorandum of agreement to prepare and supply to the
McGraw Hill Company a work entitled Control System Synthesis.
With regards to his root locus paper, still in the A.I.E.E. review process, Evans received some
help in August from Caltechs G.D. McCann, who wrote to reviewer Sy Herwald:
I believe this is a such an important contribution to the art of steady state analysis of linear
systems that every effort should be made to have it accepted by the A.I.E.E. as soon as
possible.
In September, the baseball season ended. Although it went down to the wire, the Dodgers edged
out the Cardinals for the National League championship by a single game. Stan Musials 0.338
batting average fell a few points short of the 0.342 posted by the Dodgers new leader, Jackie
Robinson, who won the batting championship. Just as it had with his 1948 Graphical Analysis
paper, an entire baseball season had come and gone while he had seemingly had struck out with
the A.I.E.E. review committees.
7
Finally, in mid-October, Evans learned that the A.I.E.E. would recommend the root locus paper
publication, but, as the fall meeting schedule was too full, it would probably (italics added) be
the 1950 winter general meeting before it would be presented putting its publication in
Transactions into 1950s baseball spring training season.
Much remains to be told of the root locus story. Yet to be written is the story of how Evans
came to write his book, of the 100,000 Spirules sent to 65 countries and of decades of
applications of the root locus method, including North Americans inertial guidance systems for
the NAVAHO missile and the Nautiluss voyage to the north pole.
Fifty years later, as dawn breaks on the twenty-first century, root locus continues to be part of the
standard curriculum of every introductory graduate course in control theory. In 1994, George
Thaler, Professor of Engineering at the Naval postgraduate School in Monterey, California,
complied a book of classic 21 classic papers on Automatic Control: Classical Linear Theory.
The last papers he chose to include were those Evans had persisted through three years of
waiting, wrangling, and rewriting to have published in the A.I.E.E. Transactions. In his
introduction of these two papers, Thaler wrote:
This group of two papers comprises the only post-World War II contributions in this volume.
In these papers Evans introduced the now-famous root-locus method. The first of these
papers is essentially background, in the sense that it shows the kind of thinking that later led
Evans to the root-locus method, and it also shows the elementary form of protractor which
later became the Spirule. Some of the conformal mapping ideas are useful and informative.
Our last paper is, of course, an exposition of the root-locus method itself. Little need be said
here, except to point out that the paper is very concise, yet contains a wealth of ideas. On
careful reading one realizes that, at the time of publication, Evans understanding of his
method and his ability to use it was at a level that most of use did not reach for another
fifteen years.
We have terminated this volume with the classic papers by Evans primarily because they
mark the last of the major, fundamental contributions to that is commonly known as the
classical theory of linear, continuous feedback control systems. There have been many
contributions to classical theory since 1950, but such contributions have largely been
expansions, clarifications, and applications of the fundamental ideas. Shortly after 1950 the
explosion in technical publications began, and in the field of automatic control, countless
papers have been published that deal with linear, non-linear, sampled data, adaptive, and
other types of systems. Many of these papers are indeed of major importance, and perhaps
collections of such papers may be of value, but none seems to be classic in the same sense as
the ones presented here.

S-ar putea să vă placă și