2003-2004 NAA Letters from Colleagues

In January 2003 I distributed draft manuscripts of chapters I had written in 2002 to nine of Dad’s colleagues. I received responses from John oore, Robert Cannon, Jeff Schmidt, Hal Engebretsen, and DeWitt Lyon. I followed up with interviews in the homes of Robert Cannon and John Moore. Here are the responses and notes from meeting with Cannon.

-Greg Evans

2003 February 27

Dear Greg:

I am delighted to hear from you. Your Dad was one of my best friends and I brought him from G.E. to Washington U. and from Washington U. to North American Aviation. I was very familiar with his root locus method and taught it in my classes from 1950 to 1956 while I was "moonlighting" as UCLA's first Visiting Associate Professor of Engineering in their Graduate School. 

One of Walt's major characteristics was his sense of humor. In fact, when we were at G.E. There came out of the University of North Dakota a song, sung to the tune of "The Battle Hymn of the Republic" called "The Love Life of an Engineer". I added three stanzas to it and your dad added the last two. I am keeping it alive today every chance that I get.

I would be delighted to meet with you if our schedules can be synchronized. I have always felt guilty that I lost your mother's telephone number so couldn't respond to her just after Walt passed away. 

I always had strong affection for Walt and Arline and would like to get to know you and your brother, if that is convenient. Unfortunately, I am scheduled for a TURP (a prostate ream job) on 3-12-03 which will certainly incapacitate me for a couple of weeks. This may inhibit my driving, but if you can visit me at my home, my address is 15980 Meadowcrest Rd., Sherman Oaks, CA 91403-4714 and my telephone number (which I think you have from Jim) is 818 981-1258. 

I look forward to getting together with you.

Best personal regards,

John Moore

2003 March 14

Dear Greg:

The following are comments about your hard copy that I received today. There is much interest in the paper and I am glad that you are writing Walt's story. My comments are only meant to be constructive, not critical and, of course, my views can very well be wrong.

I am not sure why you want to belabor Walt's trouble getting his paper published to the extent that you have. Historically, it does no good to point out the frailties of Gordon Brown, whose name will continue to be revered, if for no other reason than his starting the MIT Servo Lab. 

The fact that the review committee problems were so far in the past will only seem to be important to today's readers if a similar situation presently exists, since the Root Locus method paper was finally accepted and has continued to be widely used. Otherwise, the paper seems to be more about the problems of getting the paper published than the intrinsic advantages of the control system synthesis methods. More about the advantages of the root locus method for control system synthesis compared with the classical Nyquist and Bode methods which it displaced would seem to be more interesting to readers.

Also, the Navaho was a Mach 3.2 ramjet cruise missile--not a ballistic missile. In fact, it was the advent of the ballistic missiles Atlas and Titan, which brought about the cancellation of the Navaho in July 1957. Also, I may be wrong but I thought that "Rader" was Louie's name's correct spelling--not "Radar". 

By the way, MIT under Ridenour and Getting did a lot to improve fire control radars, but the radar was developed in Britain and played an important part in winning "The Battle of Britain" before we ever got into the war.

Best personal regards. John Moore

2003 November 3

Dear Greg,

I met your sister Nancy at the Autonetics Group 69 luncheon last week. She indicated that you were writing a book about your father, Walt, and requested people send your comments on his life that might be interesting.

I started work for North American Aviation in Downey, CA, on January 18, 1954. My supervisor was Sam (Robert S,) Carlson and the Group Leader was Walt Evans. At that time, all that later became Autonetics was called Department 93. Dr. Edlefsen was the Department Head and John Moore was the assistant Department Head. It was a pretty small organization so we all knew Walt as the Group Leader. There is certainly no question that Walt was an outstanding technical talent.

However, he was also an outstanding manager of people. Either explicitly, or perhaps implicitly, everyone in the group knew that Walt’s position was that we worked for him and that he had complete faith in us. As I recall, he said in effect, “If anyone outside the group has any complaint about how you are doing your job, send them to me. I gave you the job and I am responsible for how you do it.” This kind of management support resulted in people being unafraid to stretch their talents to the greatest extent. You learned to act and complete work without fear of possibly making a mistake. 

Walt’s support and management style created some of the finest leaders in the Navigation Systems Division (NSD) and Autonetics. Sam Carson became the head of NSD later and went on to manage the start of the Autonetics electronics components business, now mostly nonexistent. George Leisz, another supervisor in Walt’s Group 64 followed Sam as head of NSD and later became head of all of Autonetics when John Moore went to the corporate office. As you know he was a great man.

Hal Engebretson

2003 November 20

Dear Greg,

One of the advantages of being old is that people who might question your memory are either dead or also have questionable memories. With that said, let me recall a few things about Walt Evans.

In the summer of 1948, I was working on the NATIV missile test program in New Mexico. Ray Curci would come to New Mexico to work on the missile autopilot. Ray told me that John Moore had hired Walt Evans and that he was starting a servo course at the plant. Being interested in servos, I asked Ray to try to get me transferred to the servo unit. Walt was just starting his course when I arrived in Downey. As compared to my earlier servo courses, with a lot of mathematical theory, I felt that Walt had, and was conveying to the students, a true physical understanding.

The state of the art at that time was that "Transients in Linear Systems" by Gardner and Barns provided a solid basis for determining a systems behavior using Laplace Transforms. In most cases the open loop Laplace transform could be derived in factored form without too much effort. The problem was that when the servo loop was closed a new equation in the form of (1+ the open loop transfer function) resulted. This required factoring of a high order equation. If the loop gains or any other parameter was changed a different equation had to be solved. Methods existed for factoring equations of the fourth order, but higher order equations required extensive trial and error.

At Bell Labs, much work had been done on feedback amplifiers. H. Nyquist published a paper in 1932 on "Regeneration Theory" where he talks about stability in terms of avoiding the -1 point in a polar plot of gain versus frequency. One text I remember using was "Network Analysis and Feedback Amplifier Design" by H.W. Bode. The basic idea was that if the "loop gain" became less than one before the phase shift reached 180 degrees it would not oscillate. Elaborate schemes were developed to obtain the desired closed loop frequency response and later to predict the transient response for RADAR pulses. At the MIT Radiation Lab, the same methods were being developed for servos to aim guns, point RADAR dishes etc. All of the work was done using relationships between amplitude and phase in the frequency regime. Working in the frequency regime had the advantage of making it easy to incorporate test data in the analysis and avoided the problem of factoring high order equations.

Walt was apparently the first to see that the special form of the closed loop equation allowed it to be investigated and understood without first generating and factoring a high order equation. The roots of the closed loop lie on a line (locus) starting at the open loop roots and extending away a distance determined by the gain. The locus could be obtained by summing the angles to the locus from the open loop roots.

Unlike gun directors and feedback amplifiers with constant gain that could be tweaked, I was working on synthesizing an autopilot for a missile. Initially control was provided by jet vanes in the rocket. As speed increased, control was shifted to aerodynamic surfaces. The control surface effectiveness and other aerodynamic forces changed with both Mach number and dynamic pressure. The mass and center of gravity of the missile changed as fuel burned. There was no tweaking of the autopilot after launch. The only digital computers at this time were large mainframes used for number crunching. My task was to determine an analog autopilot that would be stable throughout the flight. Using all the classical analysis methods I was floundering.

When Walt first started discussing Root Locus in his class I saw a ray of hope. While the problem was still very difficult, I could at least get a feel for what I was trying. Initially Walt used a sheet of semitransparent paper to add the angles for a root locus plot. As more people started to use Walt's method many ideas for automating the plot were thought of:

One idea was to place small fish in a tank of water. If voltages were applied at points corresponding to the open loop roots, the fish would line up with lines of equal potential to avoid being shocked. These lines were the lines of constant gain in root locus and the locus of roots would be perpendicular to the fish. At this time Walt was having difficulty getting his first Root Locus paper published. In true Walt Evans humor he commented, "These fish smell as bad as the publishers think my paper smells". The Publishers and their reviewers were apparently looking for some complex mathematical stuff and failed to see how a graphical method giving approximate solutions was of any value.

Another idea was to stretch a large sheet of rubber over a frame. If the rubber was pushed up a fixed distance at each open loop root then contour lines on the surface would represent constant gain. A simple second order system was tried by poking the rubber up in two places. Walt exclaimed, "That looks just like Sally". Sally was a very voluptuous young lady working in the department.

None of these early ideas worked out and I was looking for something better than transparent paper to add angles. I went over to the engineering shop and made an "angle adder" out of a circle of plexiglass with a straight arm held on with a small bolt. This worked much better than the transparent paper for determining the locus, but I still had to measure lengths and multiply them to get loop gain. Walt and I were kicking this problem around one day and we came up with the idea of adding a logarithmic spiral to my angle adder. This worked well and became the first Spirule. I applied for a patent but the company decided a copyright was more appropriate. 

In the fall of 1948 John Moore was teaching a graduate servo course at UCLA. He was covering all the "Gain-Phase" stuff but adding some new twists whereby you would not plot gain and phase versus frequency (omega), but against alpha plus j omega. Alpha could be chosen to insure a minimum damping or minimum damping ratio. This resulted in many complex graphs which were very difficult to interpret. At lunchtime Walt would help me with my homework. In most cases, I would use root locus to understand the situation and then do it John’s way for the homework I turned in. John gave me an A in the course. I know of at least one other who failed the course and also had his employment with NAA terminated. Without Walt's help, my NAA employment might have also been short lived.

Meanwhile Walt was having difficulty getting his first paper published. The editors kept wanting him to make it shorter, as they didn't think it very important or that it contributed much to the state of the art. One day he told us that he sent them a version stating that something was difficult to explain but that in the interest of keeping the paper short the readers should just believe him. Word of mouth and Walt's first paper gradually started spreading the advantages of root locus. In his second semester John Moore even had me give a couple of lectures on how to use a Spirule.

After the first Spirule showed its worth, a design was made for a more precise model. A bunch were made for use in the department. After my demonstration in John Moore's servo class, employees of several aerospace companies asked for a set of drawings. Hughes had several hundred made, but the person doing the assembly looked at the assembly drawing and cemented the parts together in that position. Since nothing could rotate they were worthless. Walt commented that Howard (Hughes) could afford to build another set but to advise him not to try making them out of wood like his Spruce Goose.

As you know, Walt designed a much simpler Spirule to enclose with his book on servos. He asked me if I wanted to be his partner in the venture. I was busy with other things at the time and declined. I will always remember his kind offer and his desire to be fair.

Speaking of his book, I looked for my copy to see if he had included some things we worked on. I found several servo books but not Walt's. No doubt, someone at Rockwell borrowed it. It must have been very useful because it was my only servo book that was not returned. What I was looking for was whether he included such things as "a locus for most anything such as the value of a capacitor in a network" or the "pinball method of factoring high order polynomials"

Walt had a way of naming things. The pinball method was a good description of how the factoring was done. In teaching root locus, he would start with the locus for a simple quadratic. He would add another term and show its influence. He would explain that, if the locus only slightly curved the added term was "in the outfield" and he called the equation "slightly cubic".

His names were not limited to root locus. Harry Newman placed a shroud around the gyroscopes so the air flow would not change as the gimbals moved. Walt named it a "Newman Coffin". Norm Parker came up with a way of using precision mechanical drives on NAVAN gyro pairs. Instead of using accurate gyro torquers one gyro would be mechanically driven at velocity plus earth rate for a fixed time interval. In the next interval it would be caged and the mechanical drive returned to zero. Meanwhile the second gyro would be driven at velocity plus earth rate. Walt named this "The Parker Shuffle". Len Dozier suggested leaving the azimuth gyros untorqued rather than keeping the platform pointed in a fixed azimuth. Walt named this "The Dozier Wander". The first platform used air jet torquers. To balance the platform, Walt placed a penny on it and moved the coin until the air jets pulsed at a very slow rate. He called this "The Penny Fix". We had a short track where the platform could be tested under acceleration. He called this "The Short Snorter".

Walt was a walking engineer's handbook. You could ask him anything from heat transfer to hydraulics. He would pull a few fundamental constants from memory, apply a few conversion factors, do some math in his head while telling you what he was doing and come up with an answer.

In summary Walt was a very creative person with a great sense of humor. He was an excellent teacher and a fine engineer. I feel that my life was enriched both personally and professionally by knowing him. Jeff Schmidt

Dear Greg,

I have been reviewing the Birth of "The Spirule Company”. I found it and the GE story both very interesting. My memory of events is a bit different in a few places. Although it does not change the basic story, you might want to consider the following:

Under A "Plastic Device": September 1948-November 1948:

The statement, place a transparent sheet..........is as I remember things. At this time there was noninvolvement of a spiral curve for multiplication. The transparent paper was used ONLY to determine a locus. Gain at a specific point was obtained by measuring the lengths with a ruler and multiplying them with a slide rule. I would guess that I made my "Angle Adder" in early September and the Logarithmic curve was incorporated by October. I think DeWitt came up with the name Spirule the same week.

Under Word of Mouth: March 1949- December 1949:

I offer the following: Many hand -fabricated "root locus plotters" WERE NOT MADE. I think I was the only engineer at NAA to make my own. The advantages of the root locus method and the benefits of a Spirule were enough that within a few weeks a drawing of a much better (and more expensive) model was made and several were built for the key servo engineers. I remember Osborne's 1950 appeal for 500 Spirules, but believe the reference to, "a variety of their own hand-fabricated Spirules" to be stretching things a bit to get the needed money. In 1948 a few of us needed Spirules and they were built. I don't know how they were paid for but expect that they were either charged to the Navaho program or to a John Moore overhead account.

The statement that the oldest engineering drawing of a "root locus plotter" being a Northrop design is incorrect. John Moore's servo course at UCLA started in the fall semester of 1948. I am not sure, but I think Duane McRauer may have been in the class. He would have heard about root locus there. In the spring semester I gave the Spirule demonstrations and all the aerospace drawings. This was about March 1949. I am not familiar with the Northrop drawings or what improvements they may have made. One improvement would be, "Don't Glue the Pieces Together", as Hughes had done. 

You may also want to check the X-10 time reference. In 1948-49 I was working on NAVAHO-1 flight control. .It was a ballistic missile with a range of around 500 miles. At this time there was a big debate in Washington over the role of the army and the Air Force. I think the Army (and Von Braun) won out on ballistic missiles in that range and Navaho was changed to a 5000 mile range ramjet cruise missile. The X-10 was a turbojet test missile leading toward the G-26 and G-38 ramjets. This is a long way around to say that maybe the X-10 was not started until after 1949.

John Moore started teaching his UCLA servo course in the fall of 1948 not 1949. I was not aware of him teaching navigation and guidance at UCLA. All the Inertial stuff was highly classified. The Navaho inertial system was called a "Path Governor" to avoid references to inertial guidance.

Fred Rentz prepared Dr. Bollay’s Wright Brothers Lecture. Fred was one of the early users of root locus at NAA. He had been a B-17 pilot in WW2 and besides being a good engineer was an excellent "word mechanic.” Unfortunately, NAA lost him as an employee when he decided to return to the family newspaper business in New Castle Pennsylvania. To be fair, as I remember helping Fred, in addition to root locus, the lecture gave a lot of emphasis to the use of electronic analog computers for flight simulation.

I hope you find some of these comments useful. You can use any of the material I have sent except I would not want you to say that John Moore fired someone for failing his course. I am sure that it happened but don't think it adds to your dad’s story.

If you are interested I could describe the NAA Spirule design. It was built in 2 sizes. One was quite large, for use on a drawing board, so you could see the influence of roots and poles in "The Outfield" on the locus in "The Infield".

Jeff Schmidt,

48000 Airline Highway,

King City, CA 93930

2003 November 22

Hi Greg,

It's been a long time! Last month at the Autonetics reunion luncheon I had the pleasure of sharing a table with Nancy and Gary. They told me of your project of producing your dad's biography and said that you could email me a copy of the draft manuscript. I would appreciate an opportunity to see it. I was first introduced to your dad at a restaurant in Downey in December 1958 when I visited Autonetics on an interview trip.

In 1959 I went to work in the Preliminary Engineering Section of the Inertial Navigation Department. That was where the math whizzes worked out elegant systems-level solutions based on models given them from "on high" where the real engineering was done. The Electromechanical Systems Section was where the real action was ongoing and your dad was Assistant Section Chief. I transferred ASAP and stayed on the steepest learning curve of my life while working with your dad until I returned to Purdue to finish my Ph.D. I had dropped out of school to get some economic breathing room after my GI Bill benefits had been exhausted and my kids were mortgaged.

Your dad wrote me a wonderful recommendation and I was able to finish my degree on the NAA Science-Engineering Fellowship program. If you would like, I might be able to add some insight or background for your narrative.

Meanwhile, what have you been up to since we last saw one another? I retired from Autonetics Aircraft and Sensor Systems Division in January 1990 and went on the road for a year+ as the IEEE Distinguished Lecturer in Signal Processing where I picked up lots of frequent-flyer miles as I wandered through classrooms in major universities of 13 countries. In 1994 I stopped teaching at UC because faculty members were warned by the UC business office to not give bad grades (especially to mid-eastern students) in order for UC to avoid lawsuits! Can't work in that environment!

I established my own company, Signal Processing and Controls Engineering. It became too successful. After a very busy 10-year run I retired again ... but kept getting pulled back to consult for some of my old clients. I charge an obscene rate and they willingly pay!

I suppose this is ideal retirement: sitting high on a hill overlooking the ocean, working when I feel like it, just enough to stay in the thick of things, but not enough to have to worry about meetings, schedules and budgets. I have the time to read most of the latest good books, write for the McGraw-Hill Encyclopedia of Science and Technology, get in a fair amount of fitness swimming, remain active in my church and do volunteer work at the local food bank. Ah, the Golden Years! Cordially,

Stan White

Walter Evans was a person of remarkable insight, and a major leader in the understanding of automatic control and how to design excellent systems very well and very quickly.

At the heart of the design of any automatic control system is what the natural dynamic behavior of the controlled system will be: how stable it will be and how quick and well-damped its natural motions will be following any disturbance.

Evans’ inventions of the Root Locus Method (RLM) and the Spirule enabled designers to see instantly the natural dynamic behavior a linear system will have in terms of the control parameters at the designer's disposal. The RLM presents, in seconds, a plot of the system's stability, speed of response, and the damping quality of all of its natural motions.

The Root Locus Method surely ranks among the greatest contributions to engineering of the aerospace industry.

March 24, 2004

Dear Greg,

By this time, with no "Thank you" response from me, you may well have wondered if I had expired, but such is not the case. It has been purely a matter of failing to get some priorities right.

Thank you ever so much for your thoughtfulness in sending me copies of the first phases of your record of your Dad's life and work. What an outstanding job you have done in compiling both the spirit and detail of those years! It is also a tacit testimony to his and your mom’s care in record keeping. Walt was always so careful to remember who did what and to encourage and give credit where credit was due. The combination of humility and intellect was exemplary, and of course always coupled with good humor. One impressive example of his outstanding memory was not only the ability to remember humorous situations and jokes, but also to remember to whom he had told them.

I have nothing of substance at this point to add but have enjoyed immensely the privilege of going back over some of those years and being able to count him and you of the family as treasured friends. Not only was there the pleasure of working together but also that of returning from time to time from our Japan ministry and being welcomed to "pick up where we left off" years before.

Just overlapping the end of your "The Spirule: The Early Years" volume was a particular period that stands out to me. That was the flight testing period of the X-1 inertial navigation system. Walt, Jesse Bowman, and I rotated as system engineers with the X-1 in the C-47 (military DC-3) flying out of Downey. Each one would work two shifts, flying with the system in the day followed up by working up the data at night. The next day the next system engineer would do the same, with the previous day's data as reference, and the previous engineer would have the day off. To the best of our knowledge at the time, that was the first flying inertial-navigation system anywhere. You probably have lots of information on that period, including the incident when unbeknownst to us a parachute backpack (Walt's as I recall) inadvertently tripped a B+ switch in the gyro power supply. The stabilized platform started to drift, but feedback through synchro’s made the scopes monitoring the gyros look as if they were still synched in. Puzzling!

You have probably received valuable comments from others on these two volumes. I am very impressed with the contents and find at this point nothing to contribute except from my nit-picking proof-reading tendencies. So, I have simply sent with this the sheets with suggested corrections of typos that I noticed.

Of particular interest to me in the "General Electric Years" volume, beyond the valued story of Walt's life there, is the detail regarding Dr. R. E. Doherty. He was president of Carnegie Tech in the years I was there (1940-43). Dr. Doherty's engineering philosophy which you describe so well re: the G.É. Advanced Course Was much in evidence at Carnegie. My greatest engineering influence there was from my mentor professor, Dr. B. R. Teare, Jr. His methods and courses in Engineering Analysis always started with naming the primary principle to be applied to the problem and proceeding systematically from there.

Again, thank you for your thoughtfulness, both personally and in the diligence given to your dad's biography--a story that certainly needs to be told.

Cordially, DeWitt Lyon