Oral-History:Harold S. Black
About Harold S. Black
Harold S. Black (1898-1983) invented the negative feedback amplifier, revolutionizing the field of electronics. Black began experimenting with telecommunications systems in his youth and received a BSS in electrical engineering at Worcester Polytechnic Institute in 1921. Recruited before his graduation, Black joined the labs at Western Electric Company in July 1921. After undergoing the company’s training program, he joined the Systems Engineering Department, where his projects included carrier system design.
Independent work shaped Black’s Western Electric career and his inventions. To acquaint himself with the workings of the company, Black spent Sundays reading the company’s memoranda files, starting with 1898 documents and proceeding through the present. This knowledge, combined with socialization in the company cafeteria, netted Black more significant assignments. At home on Thanksgiving Day, 1921, Black made a breakthrough in his work with the imperfectly performing three-channel system, determining that “the requirements on a single amplifier would vary as a function of the number of channels,” increasing by the square root of the number of channels. Black describes his his communication of this model to Ralph Vinton Lyon Hartley in the Western Electric research department, who provided feedback on the role of amplifiers in third harmonics. As a result, Black replaced the push-pull amplifier used in the three-channel system model with amplifiers used in tandem.
He describes the resistance of Western Electric to his assignment of up to three thousand channels to an amplifier when a four-channel model had failed. Black, on the other hand, predicted that transcontinental telephone communications would require multiple channels, and received permission to work on linearizing the multichannel amplifier if it did not interfere with his other Western Electric projects, including his assignments to standardize push-pull amplifier design and to offset the effected of battery heat on amplifier performance. Nevertheless, Black did communicate with other scientists in the Bell Labs system who supported multichannel development, and he collaborated with Mervin Kelly on his ultimately futile efforts to produce a linear vacuum tube.
Black’s efforts culminated in the invention of the negative feedback amplifier in 1927, which he described in the influential 1934 paper “Stabilized feed-back amplifiers.” In this interview, Black details Western Electric construction and testing of this amplifier, and he describes the breakthrough sketch me made on a ferry ride during experimentation in Morristown, New Jersey. He considers the influence of his negative feedback amplifier patent on later work, and he describes the skepticism of experts consulted by the U.S. Patent Office, who asserted his device would operate. Once the patent was obtained, Bell Labs protected it vigilantly; a suit against Zenith for patent infringement required extended testimony by Black between 1948 and 1953.
The bulk of the interview covers the logic of Black’s multichannel amplifier and negative feedback amplifier development, as Michael Wolff asks for clarification in terms understandable to non-experts. Other topics include the promotion practices in the Bell System, and the influence of Lee de Forest and E. Howard Armstrong on Black’s own work. Black describes his childhood independent study and employment, as well as the economic factors which shaped his education. He describes his decision to join Western Electric and details the four-channel and three-channel carrier systems in operation when he joined in 1921. Black ends the interview by considering the social impact of the negative feedback amplifier and by describing another influential amplifier he developed, the 8-channel open wire system.
Black also wrote the article "Inventing the negative feedback amplifier,” (IEEE Spectrum vol. 14, Dec. 1977), which he discusses in this interview.
About the Interview
HAROLD S. [STEPHEN] BLACK: An Interview Conducted by Michael Wolff, IEEE History Center, 20 May and 29 June 1977
Interview #456 for the IEEE History Center, The Institute of Electrical and Electronics Engineers, Inc., and Rutgers, The State University of New Jersey
Copyright Statement
This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.
Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, Rutgers - the State University, 39 Union Street, New Brunswick, NJ 08901-8538 USA. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.
It is recommended that this oral history be cited as follows:
Harold S. Black, an oral history conducted in 1977 by Michael Wolff, IEEE History Center, Rutgers University, New Brunswick, NJ, USA.
Interview
INTERVIEW: Harold S. Black
INTERVIEWER: Michael Wolff
DATE: 20 May 1977
PLACE: Unknown
Education
Black:
[Referring to a paper he wrote] Do you want to retain the title of the article?
Wolff:
That's the last thing on my mind. We have a long way to go before that.
Black:
When it comes to my last paragraph about the historical importance of the thing, I was very modest there. It is used for very type of communication, whether by land, microwave, underground, outer space or submarine. And then it has applications in fields completely unrelated to communications. It has completely transcended application there.
Wolff:
Let's go chronologically and start by setting the scene.
Black:
Memory is a peculiar thing. Quite a long time has elapsed since I wrote this. In my patent and in all the work that I did when two lines joined together I joined them with a dot.
Wolff:
We are going to end up with a manuscript about three times the length of this. Your story is interesting and there is a great deal that is important enough that there is a great deal more we want to crowd into it if we possibly can. To set the scene, I want to know really when it all began, so let's talk about Thanksgiving Day, 1921.
Black:
I joined the laboratories on the 5th of July, 1921.
Wolff:
From which college had you just graduated?
Black:
I got a BSS in electrical engineering at Worcester Polytechnic Institute.
Wolff:
Was this your first job?
Black:
Yes. I wanted a job in Minnesota or South America where they make those giant turbogenerators that take fifteen years to make.
Wolff:
What company?
Black:
I don't recall, though I delivered a good many lectures there afterwards. However, they wouldn't take me.
Wolff:
Did you ever find out why?
Black:
That was shortly after a small economic depression. It was not as easy to get a job. J. J. Pilliod and the personnel director at 463 West Street – that was Western Electric Company – had a job for me. They picked me up right away and paid me $32 a week.
Wolff:
How did you connect with AT&T? In those days did they come to visit colleges or had you written to them?
Black:
That's a good question. About four weeks prior to graduation, which I think was the early part of June, I was approached by two people. One was the personnel director of the Western Electric Company at 463 West Street and the other one was J. J. Pilliod. Pilliod was the assistant chief engineer of the American Telephone & Telegraph Company (AT&T) at 195 Broadway. That is where it has always been and still is today.
Wolff:
Had you written to them?
Black:
No. I had not written or applied anywhere.
Wolff:
Except the place in Minnesota?
Black:
Yes. I have no recollection of how that came about. Education was very different than it is today. It is vastly superior [now] to my day. However I had hydraulic engineering, mechanical engineering, surveying, mathematics and physics.
Wolff:
All that went into being an electrical engineer?
Black:
Yes. I had to do work with my hands – pattern making, drop forge, machinery, milling machines, lathes – ergo, practical.
Wolff:
Had you always been interested in engineering?
Black:
I had always been interested in being an electrical engineer.
Wolff:
It was a pretty new field in those days. Do you remember what intrigued you with electrical engineering rather than for instance building bridges?
Black:
I didn't have very much money.
Wolff:
Nobody did.
Black:
Nobody did, but I came from poor parents. My father only went through the eighth grade and my mother was stenographer and had graduated from high school. Naturally they wanted to see me get an education.
Wolff:
What did your father do?
Black:
He was a shipper. That was his title at a shirt shop.
Wolff:
Had you always been interested in electrical engineering?
Black:
That came about after I purchased a series of small books. I think there were ten or fifteen volumes. They dealt with all phases of electricity – magnetism, motors and things like that. As is mentioned in that book that I gave you, one summer when I worked for the American Steel & Wire Company, I worked trimming arc lights. They were all 220 volts or higher. When I put a red sign over the switch, somebody would close it while I was out working. Therefore when I climbed the poles I always was prepared to get a shock and protected myself accordingly.
Wolff:
That was during college?
Black:
Yes.
Wolff:
Do you remember how old you were when you bought those little books? It says here that as a boy you had always been interested in setting up telegraph systems, telephones and microphones.
Black:
That is correct. My first telecommunications system was probably in 1914, a year before I graduated from high school. I was sixteen years old. In the attic of the second story of small house that my father rented I had electrical things I had gathered from the town dump. That was my great source of supply. Across the street was the editor of the Lamister Enterprise. He had five daughters. I made a microphone that was two pieces of wood at right angles and then from a battery I had sawed two pieces of carbon. Then I used a tin can that I cut to make a spring so the two were in contact. That was a sensitive microphone. I could hear a watch tick and conversations all over the house. I took a piece of very fine wire that I could throw across the street and catch onto the two poles and bring it downstairs. Up in the attic I had an old telephone receiver that someone had discarded. Therefore I was able not only to hear a watch tick – because they put one on the table – but I was able to hear every word of conversation all over their house. That worked fine until a little after 5 o'clock when the father came home. He destroyed the microphone, tore down the wires and said, "No more." Therefore my first telecommunications system didn't last very long.
Wolff:
It was in their house?
Black:
I could hear everything that was in their house. It was not two-way. They could not hear things in our house. That didn't please the father.
Wolff:
That's a delightful story. You said you got interested in electrical things when you bought this series of books. Do you remember when that was?
Black:
That was before.
Wolff:
Was it sometime as a teenager?
Black:
Yes. They were advertised and the price was very small, but they were full of information. I didn't graduate from college until 1921 because my father lost his job and I had to support a family of four. By good fortune I was able to earn $12 and [unintelligible phrase] a week working at the same factory at which my father had worked.
Wolff:
Working at the Steel & Wire?
Black:
No, I worked at the shirt shop ironing shirts and things. They had automatic machinery for pressing. And there were woolen pajamas for men. That was all piecework, and I did pretty well.
Wolff:
Did that give you enough money to go back to college?
Black:
That gave me enough money to take care of that year. When I was 16 or 17 there was World War I. Nottin Grinding Company, although my father had never done that kind of work in his life, gave my father very good pay and the entire family moved to Worcester. I however kept on working in the shirt shop, because even with a 10-cents-a-day carfare I could make more money that way. That was all the work I did at that time.
Wolff:
Then you went to Worcester Polytechnic Institute because you wanted to study electrical engineering?
Black:
Yes. I was bound and determined to become an electrical engineer.
Wolff:
Was this partly because you felt it was a good profession?
Black:
I guess the truth is because I was interested in electricity.
Western Electric Company
Recruitment and training
Wolff:
Let's come back to Pilliod and the other fellow from AT&T.
Black:
They made me good offers.
Wolff:
What I don't understand is how they knew about you if you didn't write to them.
Black:
They were looking and they interviewed me as well as others.
Wolff:
How did they know you were there?
Black:
They went to Worcester Tech because they turned out electrical engineers. At that time they had no graduate courses. They turned out electrical engineers, mechanical engineers and civil engineers.
Wolff:
And they made you an offer at $32 a week.
Black:
After they interviewed me and made me an offer and I accepted it right on the spot.
Wolff:
Did you start work at Bell Labs then?
Black:
It was called the Western Electric Company. I started there on July 5th. They didn't work on July 4th. There was a delay of a couple of hours. I had to pass a physical examination to settle the fact that I was in good health.
Wolff:
What was your job there when you started?
Black:
First I had to go through an 8-week training course. That was to familiarize me with the organization of the Bell System and where the Western Electric Company, AT&T, Long Lines and the associated companies fit in with it. In addition there was a little practical work. I had to work as a telephone operator and use the cards and things like that and listen to lectures by different departments at 463 West Street. The department that I chose, which answers your other question, was Systems Engineering. I became a systems engineer punching a time clock. After my other factory experience I did not care for that.
Wolff:
Was this after the 8-week training course that you began the systems engineering?
Black:
Even in the 8-week training course we punched a time card.
Systems Engineering Department
Wolff:
Did you become a systems engineer after the training?
Black:
After the training course I was interviewed [by] heads of different departments. There were 150 people total that entered that particular year. There was a systems engineering department, an apparatus department and a research department. They were rather limited. The systems department was the one that appealed to me because it involved contacts with all the others and made complete systems.
Wolff:
Did you start working in that department after the training?
Black:
Yes. That department was headed by Amos Dickson. He graduated from the eighth grade but then took correspondence courses and worked up to the head of that department in a position that would correspond today to vice president. I was interviewed by B. W. (Burton W.) Kendall.
Wolff:
How many engineers were in the systems department? Was it small or large?
Black:
It was fairly large. The only way that I could answer that question accurately would be to get out the organization chart, but that department occupied the entire 9th floor at 463 West Street. That was where I met "Hi" [Haakon I.] Romnes, who later became president of the Western Electric Company and after that became president and chairman of the board of the American Telephone & Telegraph Company.
Wolff:
Was Rommes in the systems engineering department?
Black:
Yes, but he did not stay very long. He moved fast.
Wolff:
That brings us up to September of 1921. The summer was the training and then you started. What was your first assignment in the department?
Black:
That is a good question. I went out in the laboratories to make myself useful to people who reported to someone by the name of Slaughter. Slaughter's position was later taken by Charlie (C. W.) Green. One morning I given a set of tools that I did not succeed in keeping. Then I leaned against a battery and it burned the bottom of my trousers. Fortunately I came down to New York with $75, so that when they stopped work in the middle of Saturday afternoon I had time to buy another suit of clothes.
Wolff:
Was your first day at work in systems engineering on a Saturday?
Black:
No. I got along as best I could with a poor-looking suit. Saturday was the only day I could do any shopping. I made myself useful and helped anyone who was working. As a result I was finally given a succession of problem-solving assignments. You would find it hard to believe, but at that time in the Bell Laboratories and the systems engineering department nobody ever used as much as a slide rule. Everything was empirical.
Wolff:
Do you remember some of the assignments you had? Were they empirical assignments?
Black:
Yes. I was concerned with the design of the first carrier system. It was a four-channel system that had already been designed for President Woodrow Wilson so he could communicate with Pittsburgh. That was rather unsuccessful because there were some directional filters that did not have enough singing margin between the two directions. That is where I became familiar with the terms "singing margin" and "phase margin," which were no mystery to me when I got around to inventing the negative feedback amplifier. It is explained in some of the references that I gave in this paper. So-called directional filters. In order to get two-way transmission and a number of channels involved, aside from this first example that I mentioned to you, were three. This triangle was the amplifier. In front of that was an equalizer, which I could almost have left out. That is because the attenuation of the line varies as the square root of the frequency and there is a very clever and important technical reason for pointing there. Then to get transmission in the other direction they use a second amplifier.
Wolff:
What is this diagram? Is this the four-channel system?
Black:
This is a diagram of the three-channel system, and it wasn't working at all. The second three-channel system is the one to which I am referring. The first three-channel system transmitted the carrier and so that if one talked on one channel everything on the other three channels could be heard as well. The AT&T didn't think much of that, so this was being worked upon by Slaughter. After, Slaughter was replaced by C.W. Green, who came as a professor from MIT.
Wolff:
I had better interrupt you or I'll get lost. You were telling me that you worked on a four-channel system that had been designed for President Wilson.
Black:
No, I did not work on that. I said that was prior to the three-channel system. We had a four-channel system somebody else designed. That was a complete failure because these two high-pass and low-pass filters, if this is the frequency band that is for the low-pass filter then you have a guard space as it was called, and then there is a high-pass filter. Therefore the low one would go here and the high would go here. Then of course as frequencies got very high and very low they would become virtually [unintelligible word]. If the gain at the crossover points is more than this attenuation, the thing will sing [i.e., break into oscillations], and if the gain is a little less than that but doesn't have a good singing margin such as 15 dB or higher, the result is serious distortion – which is what that four-channel system for President Wilson had.
Wolff:
This had been worked on at the lab when you started to work there. Is that right?
Black:
That's right.
Memoranda files
Black:
One thing I didn't tell you is that after I had been working for the Western Electric Company for a little less than six months the time came for a raise. Everybody who started at $27 got a $3 raise, but I got nothing. I said to myself, "By golly I'm going to resign tomorrow morning and work my way to go to Harvard Business School." Had I done that it would have been the biggest mistake of my life. I could never have been a businessman. However I thought better of the idea by the next morning. Instead I said, "I am going to go to work and learn everything that I can about the business." At that time the company had memoranda for file that had been written by everyone there. The records extended back all the way to Dr. George Ashley Campbell, the first professional mathematician the AT&T ever hired. I went all the way back to 1898 – which was the year I was born – and read everything that had been written by employees of the Western Electric Company in the form of a memorandum for file. By the time I got to the memorandums from 1921 I was hitting payday. I was learning what happened on all twelve floors at 463 West Street. In that way, and in due time, I became well acquainted with the names of the engineers and the supervisors and to whom they reported. In addition to that they had a cafeteria where food could be bought almost nothing. I ate my three meals a day there. That was very good food at a cost measured in pennies. For instance ice cream was a penny. I would sit down three times a day, particularly with the 150 people with whom I came in, and became well acquainted with people by eating there and by these memoranda. I read them all that I could without interfering with my work. I was of course gradually getting more and more responsible assignments. I would read them all day on Sundays in particular. You notice a pipe there. That was around '34. Smoking was forbidden. I had a pass that permitted me to get in and the guard would examine my briefcase or whatever I was carrying to see that I didn't take anything out. I went in as early as I could on Sunday mornings and worked up until 8 or 9 o'clock. However I would always get caught.
Wolff:
With what would you get caught?
Black:
Smoking a pipe.
Wolff:
On Sundays when you were reading?
Black:
When I was reading. I was reading these memoranda for file in my office. The first week it was reported to my immediate supervisor, who oversaw about six people. The second week it was reported to Slaughter, who couldn't have cared less. The third week it was reported to B. W. Kendall. The fourth week it went up to Amos Dixon. Dixon said to me, "Harold, I understand that you are one of the most promising engineers in the systems department. How is it that you cannot manage to smoke without getting caught?" I said, "Amos, I'll tell you a very simple solution. Why don't you let me smoke in Slaughter's office? That's a private office where smoking is permitted." He said, "That is a good idea, and that is what we will do." That solved that problem. At any rate, in reading these memoranda for file I became acquainted with all the supervisors and department heads and fondillas [spelling?] – that's the apparatus – and other departments that I was called upon to work with as a systems engineer.
Type C three-channel system and amplifiers
Wolff:
That's very interesting. Let's come back to the fall of 1921. I want to get up to that Thanksgiving Day when you actually started to look into the reasons for the imperfect performance. After your training program, which lasted July and August, you started in the systems group in September.
Black:
That's right.
Wolff:
One of the first things with which you became acquainted was this four-channel system.
Black:
Three-channel. I learned from my reading about the four-channel system that was not a success. I also knew what a failure the four-channel system, which was called a Type A system, had been. The next three-channel system was a Type B, and that was a complete failure as well. The next system after that was being worked upon.
Wolff:
When you started there in the fall was the main project this Type C system?
Black:
Yes. That is where I learned that everything in the system – and particularly the amplifiers – was causing trouble. That's the first reference that I made in the manuscript.
Wolff:
When you started there in the systems department in the fall of 1921, was this Type C system the major problem in the department and was it the first major assignment in you got involved?
Black:
The first major assignment that I had was to get involved with the Type C. Prior to that I had to fix one of the amplifiers in a Type B three-channel system in the laboratory. That took me a week. I wasn't very quick at finding trouble. They never let those go into commercial service because they were so poor. That was not a major assignment, and I was just a beginner at that point. By that time they had gotten hold of C.W. Green, and it was his job to make the Type C system work. It took him about four years to do it. However the amplifier is what caught my eye. For the three channels in the short distance that they was going – and it was going on open-wire line – it would get by and do its job. I envisioned something entirely different. I thought of amplifiers that would carry one to three thousand channels – such things as I described in the manuscript.
Wolff:
I don't want to get too far ahead here. Your first major assignment was to get involved with this Type C system.
Black:
Yes.
Wolff:
And particularly the amplifiers, which were causing trouble.
Black:
There was little that could be done about the amplifiers that were causing trouble. I was able to point out that the third harmonics generated by the input and output transformers and the iron core coils in the filters were making the thing unworkable.
Wolff:
I thought it was the second harmonics that were the problem.
Black:
Turning to the amplifier I found the importance of the second harmonics. Theoretically they would cancel out, but practically they did what I said: [unintelligible word].
Wolff:
The third harmonics had to do with what?
Black:
They are the odd order. Any nonlinear device.
Wolff:
Did the amplifier generate them?
Black:
They were generated by [tape turned over at this point; sentence not completed in recording].... [H. J.] van der Bijl's book on vacuum-tube amplifiers showed me the curve characteristic and all that was wrong.
Wolff:
It's a little puzzling. You started the article by saying you remember investigating the reasons why the amplifiers did not perform well on Thanksgiving Day.
Black:
That's right.
Wolff:
What was it about that day that you remember doing?
Black:
I did that at home. I had to do what I was supposed to do on working days.
Wolff:
What were you supposed to be doing the day before that?
Black:
I was contacting the transformer people so that they would make more linear amplifiers and also do a better shielding job for the Type C system. That was a specific assignment.
Wolff:
Good. That was your assignment.
Black:
That was my real job.
Breakthrough on amplifier performance
Thanksgiving Day, 1921
Wolff:
Then what happened on Thanksgiving Day?
Black:
I let my imagination take over. I felt that I could foresee that we were going to need many, many channels going through an amplifier. I prepared a chart, based on push-pull and this poor performance, of how that requirement would go up as a function of the number of channels. They would add in random. I assumed that the third harmonics would behave the same way. That caused me to describe that to Ralph [Vinton Lyon] Hartley. About a day later he wrote me a little private note. It was not a memorandum for file. He showed me that push-pull amplifier or not, with any kind of an amplifier the big problem was the third harmonics. Therefore the amplifier had two jobs. With this new information, I had to change my chart to show how the requirements on a single amplifier would vary as a function of the number of channels. There I let the number of channels vary from one to three thousand.
Wolff:
I am not that familiar with this material and a little confused. I'm still back on Thanksgiving Day. What was it specifically that you started to do on Thanksgiving Day?
Black:
I was investigating the reasons for the imperfect performance of the amplifiers used in the Type C system.
Wolff:
By investigating do you mean you were at home reading?
Black:
No. I was calculating what this curve would look like.
Wolff:
What curve is that?
Black:
In this case it was a three-channel push-pull amplifier. I was worrying about what requirements it would have to meet.
Wolff:
You were doing this on your on time at home on Thanksgiving Day.
Black:
Sure.
Wolff:
What did you discover?
Black:
I discovered that the requirements were very, very severe and went up as the square root of the number of channels.
Wolff:
What about the square root?
Black:
If you had four channels instead of two the thing would go up 3 dB – that is, the square root of 2 rather than by going to 4. The curve would be built up that way. It would vary [according to the number of channels]. The reason is that in a push-pull amplifier the two sides are never the same – particularly with the poor vacuum tubes available at that time, which were designed by Dr. [Mervin J.] Kelly. That was how they would go, and it was not anywhere need good enough to meet this requirement of a few percent that I mentioned. However I kept the curve going.
Wolff:
What requirement of a few percent did you mention?
Black:
I said somewhere in this paper that the second order instead of disappearing were only reduced by 10% to 18%.
Wolff:
That is a little confusing because a word is missing.
Black:
They were reduced.
Wolff:
Is the distortion reduced by 10% or to 10%?
Black:
It is reduced to 10%. That would be 1/10th. I guess the word "to" should be in there.
Wolff:
Yes. I have marked it.
Black:
That escaped my notice. Knowing that the law of addition for that kind of an amplifier was the square root of the number enabled me to draw my first incorrect curve. Step number was to consult with Hartley. Employees at the laboratory at that time were permitted walk all over the laboratory and talk to anyone they liked. Ralph Hartley was in the research department. He was somebody that I liked very much. I had read his memorandum, admired the work he did, and this was in his field. He had done some work on third harmonic modulators. I explained to him what I was doing, and he turned right around and told me what would happen. The reason that I substituted a carepegs [correct word?] for the little "n" was that I thought that having two n's in this thing was going to be a matter of confusion. Hartley told me in a single-sided amplifier the value of the nonlinear distortion in any such amplifier – regardless of the kind, good or bad – it would be twice as much. If there were four of them, it would be four; if there were three thousand, then three thousand times as much. That is more than 60 dB. Now that is getting into some sledding that is really tough. This is a technical point, but unfortunately third orbit distortion appears as understandable speech in the adjacent channel. The AT&T requirement was 70 decibel. That is 3,000 to 1 on a power basis or more than a million. That was a terrific requirement. Therefore I had to redraw my curves.
Wolff:
You paper explains that this amplifier reduced the second harmonics to 10% to 18%.
Black:
That's right.
Wolff:
You thought that the third harmonics would also be reduced to 10%.
Black:
A similar figure.
Wolff:
Okay. Then Hartley said that was wrong.
Black:
Very wrong.
Wolff:
He said that it builds up as the number of amplifiers. Is that right?
Black:
Yes. It sure does.
Wolff:
I don't understand this sentence where you say "independent."
Black:
That is independent of the type of amplifier. If there are a lot of amplifiers in parallel and big power is obtained that way, no matter how you get it, [the same rule applies]. In multiplied systems we only use a single tube. Therefore no matter how the amplifier is made and no matter what kind of an amplifier the third harmonic, the third harmonic is more serious than the fifth or seventh harmonic. When the term "harmonics" is used, it really means third order products too. There is a lot of literature on that. That goes directly. If there are a thousand amplifiers where a 60-dB improvement is needed, and there is 70 dB that is already sought, it could be seen right away that this was a tough problem.
Wolff:
Would you paraphrase this sentence you wrote to make it clearer? No matter what kind of amplifier, what is it that happens to the third harmonics?
Black:
The third harmonics. The things you didn't want would double in size when you came out on number two.
Wolff:
The harmonics would be double as big?
Black:
The harmonics. The things you didn't want.
Wolff:
After a thousand it would be a thousand times.
Black:
It would be a thousand as big, and a thousand is 60 dB.
Wolff:
All right.
Black:
Three thousand is about 70 dB. That was what hurt. That really hurt. It was bad enough to have a lot of channels going through one amplifier, but then to pile on top of that the fact that the thing was going to go up with the number, boy, that gave you something.
Wolff:
When you say the thing is going up as a number you mean the distortion, right?
Black:
Yes. I didn't think this paper would want to get that technical, but in the adjacent channels that distortion appears as clean speech. You see, you also get cross-talk between wires and things. AT&T's view was that a conversation should be private. If one picks up the telephone someone's talking that is not a part of that line or circuit should not be heard.
Wolff:
Was Hartley a Rhodes Scholar at that time?
Black:
Yes. He had come back from Cambridge. He was much older than I.
Wolff:
Was he in the systems department?
Black:
No, he was in the research department, but all I had to do was walk to his office.
Wolff:
Now we are down to the last paragraph. I think I understand this now. What were you determined to accomplish? After Hartley told you this, what did you say to yourself?
Black:
I was determined to solve this problem if it was the last thing I ever did.
Wolff:
What was the problem? To find a workable free channel?
Black:
Just what that sentence said in the paper.
Wolff:
Having "many amplifiers in tandem," that means in series, right?
Black:
Yes.
Wolff:
Was this to make the Type C system workable?
Black:
No, it had nothing to do with the Type C system. This invention was no accident, but I did not get any encouragement.
Wolff:
You started off investigating why the amplifiers didn't work well.
Black:
Yes. That's true.
Wolff:
Is it correct to say that your discovery of why it didn't work well was because of the second harmonics?
Black:
Yes. I thought the third harmonics were going to cause less trouble than the second.
Wolff:
All right. Your first discovery was that it was the second harmonics.
Black:
Yes.
Wolff:
You thought the third harmonics would cause less trouble and then you found they caused more. Hartley explained to you how serious it was.
Black:
Precisely. Yes.
Multichannel amplifiers and transcontinental telephony
Wolff:
After that you were started to think about something very different.
Black:
Yes. I had given up the idea of a push-pull amplifier. I had to find some way. There was nothing at that point in time to indicate that I could make this last paragraph in my paper come true.
Wolff:
It is kind of elementary, though I didn't realize it, that the basic point of the type C system was that it used a push-pull amplifier.
Black:
That's right. And that was good enough for what it had to do.
Wolff:
Your brainstorm, if you will, was to forget about push-pull amplifiers and to put aftertreatments in tandem.
Black:
In a long, long string.
Wolff:
All right. A string of amplifiers. What confused me was I thought that in the type C these were repeaters.
Black:
They were. It did have repeaters, but they were many miles apart because they went on something like 16-gauge. They were hundreds of miles apart and they never expected it to go any further than a thousand miles. Therefore there were not very many repeaters on a string on that system.
Wolff:
That was not what you thinking about as repeaters.
Black:
No. My idea was to use many, many repeaters.
Wolff:
In tandem.
Black:
Yes, in tandem. That is what I meant by the word "string." That was precisely why I switched to a clearer explanation of what I really meant. You may be able to improve on that explanation.
Wolff:
Okay. At this point you were starting to think about amplifiers in tandem and each amplifier handling many channels.
Black:
That's right – even up to three thousand.
Wolff:
The first thing you did on November 30th was to plot the distortion and linearity requirements.
Black:
That's right. Thanks to Ralph L. V. Harley I was able to do that. He could have prepared that chart too, but I did it.
Wolff:
I think this is clearer here. [Jake] Jammer said, "A beautiful piece of work, but why bother about so many channels?"
Black:
Three was the biggest he had ever heard anybody bother about at that stage. Four channels had been forgotten.
Wolff:
What was your reaction?
Black:
It didn't surprise me, because Jake was a happy-go-lucky person. He wondered why I concerned myself with three thousand channels when no one else was bothering about any more than three.
Wolff:
You asked to work on this on December 27th?
Black:
Yes. I had to get permission to do a thing like this.
Wolff:
Let's talk a little bit about this period. That they didn't understand.
Black:
They couldn't understand why I would be interested in anything of this sort. No one could.
Wolff:
Why were you interested?
Black:
I knew that there would be a demand for many channels. A single coaxial today transmits 1,028 and a waveguide transmits almost half a million. By making a simple terminal change that half a million can be expanded to almost a million. Optical fiber has thousands of times more transmission.
Wolff:
You didn't foresee optical fibers at that time.
Black:
No, I sure did not. I recommended the inventor of fiber optics for a research corporation prize. I got one of those awards myself.
Wolff:
Do you remember why or how you knew there would be a demand for many channels?
Black:
The shortest distance across this country was 4,000 miles, and of course in telephony there has to be regulars and spares. To use a big network efficiently a straight line is not always the shortest distance. Dr. [Frank B.] Jewett came long a year later and envisioned 4,000-mile system. That was the first job that made use of my amplifiers.
Wolff:
Why did you think there would be so many channels?
Black:
It's a big country. Certainly the wires were not going to be put overhead because grounds and everything interfere. The coaxial incidentally had already been invented.
Wolff:
Maybe I'm missing the point. When you say "many channels" is that synonymous with a coaxial cable?
Black:
No. On July 1st of 1975, 66% of our long distance communications was via microwave radio relay systems. Incidentally, I was the first to introduce that.
Wolff:
We'll get to that, but I want to press you about why you thought there would be a demand for so many channels.
Black:
It's a big country and I couldn't conceive of using open wire. I could not conceive of not wanting to hop across the country.
Wolff:
Let me put the question this way. Does open wire mean single channel?
Black:
I used the original Blackwell as a textbook in one of my courses. The disadvantages of open wire would be so many that it would never be used to hop the country. We did hop the country with open wire. That was where we got our first transcontinental thing before we had repeaters, but there you just got one channel. You didn't get a lot of channels.
Wolff:
That's what I don't understand. When I asked you how you appreciated that there would be a need for many channels and you said because open wires could not be used.
Black:
I knew that you could not have a telephone system and not be able to go from coast to coast.
Wolff:
Does that mean that an open-wire system is synonymous with single channel?
Black:
You would never get three channels. The only number that was ever gotten – and it was not very good – was one channel.
Wolff:
That's what I'm trying to see.
Black:
You see the little picture there?
Wolff:
That's a single-channel system?
Black:
Yes.
Wolff:
Are what you have on telephone poles single-channel systems?
Black:
That all depends on the pole that you look at, because this is not 1921. As you look on these things [in this photo] here, that is a group of a large number of paper-insulated wires. They are twisted together to prevent cross-talk between wires. That is actually a cable. It used to be lead, and now it is laminated with plastic in such a way to get the necessary shielding. That is the open-wire cable of today. It is used to hop only very short distances, such as to bring the telephone into this house.
Wolff:
Was the idea of putting in more than one channel very difficult on open wires back in 1921?
Black:
It was not practical. We had the one channel to Denver. That was before Dr. Harold DeForest Arnold invented the hard-vacuum tube. If you wanted to talk over open wire you could as far Denver. If you yelled to the top of your voice from Boston, the message heard in Denver would be a faint whisper. With the advent of Dr. Arnold – and there was some litigation there with Dr. [Irving] Langmuir at GE, but we were the first – with the hard-vacuum tube we got a single voice channel from coast to coast. I have forgotten the spacing, but that was a landmark. I would be able to find the answer to your question somewhere in the blue four-volume set of books from Bell Systems I have on the shelf there, but I can tell you in my own words. That was considered a great accomplish in 1915 or whatever the exact date. They got a two-way telephone conversation from coast to coast. It was ocean to ocean. It wasn't good. The bandwidth was about 2,000 cycles as opposed to 3,000 cycles today.
Wolff:
Let me see if I understand the point. It was not practical to have more than one channel in an open-wire system. Is that right?
Black:
That's right. Not the way things were done at the time.
Wolff:
Right. You said to yourself there is a need for many channels.
Black:
Many, many channels.
Wolff:
And it's synonymous with the fact that it could not be an open-wire system.
Black:
That is for sure.
Wolff:
All right.
Black:
There are many other alternatives, but it sure was not going to be open wire.
Wolff:
Going to the second paragraph on the page, you asked to work on linearizing, stabilizing and improving amplifiers.
Black:
I asked permission to do that.
Wolff:
This was so that you could have a large number of multichannel amplifiers.
Black:
And I did not get much encouragement.
Wolff:
Was this because people did not see the need for many channels?
Black:
That's right.
Wolff:
You wrote, "I studied the available material listed in topics not treated."
Black:
Yes. That's in a book that I wrote called Modulation Theory.
Wolff:
I looked that up, and what you say here is that you studied everything you could on the unwanted generation of products of modulation through the nonlinear response and the design theory of modulation and related nonlinearity. Right?
Black:
Yes. That was a natural thing to do. Having read all of these memos before, I just went back and read a particular group. There are memoranda for file, and this is true even today. Most of them are papers that are candidates for publication but for patent or other reasons they were not for the public, whether they are classified or not.
Wolff:
I don't understand is why the people there didn't see right away that there would be a need for many channels.
Black:
I cannot read their minds of course. You will see from this manuscript that I got 50 dB of feedback over a certain frequency range. That was at the end of the year after I got it all done.
Wolff:
That was six years later.
Black:
Six years later than '21, yes. That's right. What page number is that?
Wolff:
I'm on page 2 in the second paragraph where they said you could do it if it didn't interfere with your other work.
Black:
Yes.
Wolff:
You wanted to work on a multichannel amplifier.
Black:
Yes. An imaginary one.
Wolff:
And you say they did not appreciate the importance of many channels.
Black:
Well, I wouldn't say that. They said, "Go right ahead, Harold, but don't let it interfere with your other work." I was paid a salary and had to do some work.
Wolff:
Jammer did not think it was very important to do. Right?
Black:
He sure didn't. Less than eight months later he went off to Australia. He gathered with him every memorandum for file that had ever been written that he could lay his hands on, and after he did the Australian job he became vice president of IT&T and never came back to the Bell System.
Wolff:
You started this reading.
Black:
That was a first and natural step.
Wolff:
There is something here I don't understand. You asked to be assigned this task and they said, "Okay." Who was it that told you it was all right as long as it did not interfere with your other work? Was that Jammer?
Black:
That happened to be Jake Jammer. Yes. The reason for my wording is that by that time Slaughter had left and Charlie (C. W.) Green had come down from MIT. He didn't think much of Worcester. He thought MIT was better. However he was ill. He had a critical illness and had to stay home, though it didn't affect his salary, for about four months. Lindridge was my supervisor, but Jake Jammer was taking Green's place so I thought I should ask him. I showed the charts to Jammer too. Anyway, that was J. S. (Jake) Jammer's response.
Wolff:
He was your supervisor?
Black:
He was my boss's supervisor.
Wolff:
You asked Jammer to allow you to work on linearizing.
Black:
That's right. That was the exact question that I put to him.
Wolff:
And that's his answer.
Black:
That's his answer.
Western Electric assignments
Push-pull amplifier design
Wolff:
Do you remember what your main job was at the time?
Black:
Yes. They had some other push-pull amplifiers that had been designed for another system that had not been made obsolete because there were some in the field. They were also push-pull amplifiers, and there were fifty-seven kinds. In the interest of the Western Electric Company, who had to supply all of these things, I replaced those by a single design. That was a job on which I was working at that time.
Wolff:
Was that an assigned job?
Black:
Yes. That was assigned.
Amplifier interference
Black:
I also got another assigned job. There was an amplifier in use for some purpose I don't remember. While the tubes were warmed up by a 24-volt storage battery they were picking up some 60-cycle hum. They wanted to know how to get rid of that. I made a sort of Wheatstone bridge with an extremely high resistance to parallel the other and pick the two midpoints of that so that the 60-cycle did not interfere with the amplifier. I very rapidly reached the point where I didn't just walk around and fill pails with water. I was given small problems to solve.
Collaboration
Western Electric colleagues
Wolff:
Meanwhile you started to reading about these other topics in the evenings?
Black:
That's right. Saturdays, Sundays and anytime that I felt like it.
Wolff:
You said there were many other researchers in 1922 who—
Black:
That was in the research department. Dr. Eugene Peterson was one. I could name people.
Wolff:
We don't need all that.
Black:
Within the Bell System, within Western Electric, I was not the only one.
Wolff:
I understand that.
Black:
They were aware of the need.
Wolff:
You were reading and thinking about this at the same time as other people in Bell Labs.
Black:
That's right.
Wolff:
Did Mervin Kelly in particular work closely with you?
Black:
Yes. We were great friends. Every vacuum tube that Western Electric manufactured was designed by Dr. Mervin Kelly and people working for him on Hudson Street, which is about a block from 180 Varick Street. It is only a quarter mile from 463 West Street, which is where I was at that time. During this period I went to Kelly and told him what I wanted. For quite a few months he worked with me until it became evident that it was hopeless.
Linear vacuum tube
Wolff:
What exactly was hopeless?
Black:
To make a linear vacuum tube was hopeless. To give you an example, if you plotted the voltage to the grid as a function of the amplification of the voltage out, if you plotted the output against the input in terms of current – and I square this pump [correct phrase?], if you made yourself a drawing that was 1 mile square and drew it using the hardest pencil chiseled to a straight line and made that line straight, the nonlinearity would be too much to meet my requirement. Therefore there was no earthly way of getting it by a linear vacuum tube. There are a lot of tricks one can try.
Wolff:
What was your requirement again? You said you said you could not get it to your requirement.
Black:
My requirement was modest. It was 50 dB. Dr. Jewett scaled that down to 40 dB for the first commercial application.
Wolff:
When you say 50 dB are you talking about distortion?
Black:
Distortion. Nonlinearity.
Wolff:
It had to be less than 50 dB?
Black:
As good as 50 dB is what I asked Kelly.
Wolff:
Then 50 decibels is a low distortion?
Black:
That is a low distortion. I gave a figure. I think it's three thousand maybe.
Wolff:
You and Kelly were working together trying to get 50-dB distortion from the tube.
Black:
That's right.
Wolff:
And you could not do that.
Black:
No way.
Wolff:
Why was that?
Black:
It was just a fundamental limitation. It was inconceivable that one could make a vacuum tube that linear.
Wolff:
When you say fundamental, was it a theoretical limitation or was it because the techniques or materials were not good enough?
Black:
Remember this was in the early '20s. I have the date of the invention of the transistor. The techniques for making that [unintelligible phrase] used negative feedback. It would be no good if it didn't.
Wolff:
I'm talking about this 50-dB distortion. You said it was a fundamental limitation and I'm asking if it was a theoretical limit.
Black:
No, that was a practical limitation. There was no way it could be done.
Wolff:
Was it because the materials were not good enough?
Black:
We knew electrons and that they went opposite from the currents that we could conventionally draw using a triode or a pentode or any of the vacuum-tubes available at that time. No one in the world could get it, and they don't get it today either.
Wolff:
All right. I was wondering if it was because of imperfect materials or techniques.
Black:
No, it is just the fundamental property.
Wolff:
Is it theoretical?
Black:
Not theoretical. I mean no vacuum tube. I can pick up any book you want on vacuum tubes and no vacuum tube like that exists.
Wolff:
Okay. You and Kelly worked on the tubes but it just didn't work.
Black:
Yes. He was very patient, but it didn't and couldn't work. No way.
Negative feedback amplifier
Invention and testing
Wolff:
Let's talk about this photograph.
Black:
That's entirely different, but it is curious.
Wolff:
What does it show? Is this 1927?
Black:
Let me see. When did I finish this job? I think it was in '27.
Wolff:
Are you talking about when you invented the negative feedback?
Black:
Yes.
Wolff:
That was 1927.
Black:
Oh yes, by the end of the year. The beginning of the next year, in January, Dr. Jewett came along. This paper does not cover this at all. Jewett wanted to take 16-gauge paper-insulated cables, which existed to carry voice, and use the frequency band 4 kHz to 40 kHz to travel a distance of 4,000 miles with repeaters spaced every 25 miles. With that requirement it could be done if the power level of the amplifiers that I finished up with were about 8 dB to 10 dB. They were ten times as powerful on a power basis as that amplifier. To achieve that result we used pentodes in the first two stages. I had used pentodes throughout at a lower level in the first two stages. Then, because the Bell System would not hear of having a B battery – if that means anything to you, a plate voltage – higher than 130 volts.
Wolff:
Yes.
Black:
My first proposal was to use a filamentary or an indirectly heated triode. I have forgotten which one. However it was to have two grids, one positive and one negative, so as to get more swing without increasing the voltage. [J. O.] McNally, who worked for Dr. Kelly, constructed some models. I tried them out and they were not linear enough. Therefore we had to go to 260 volts.
Wolff:
You built seventy amplifiers.
Black:
Yes. This one shows their stability.
Wolff:
Okay. We are on page 15 of the Bell Systems— This is the 1934 paper, right?
Black:
I was not permitted to publish this early.
Wolff:
This photograph was taken in 1927 and it's one of the amplifiers you built for Jewett?
Black:
No. The year of 1928 was consumed me designing this amplifier in my own laboratory and getting models made. Then in 1929 the Western Electric Company made seventy of them. They made few more too. The trial was all over in 1930.
Wolff:
Jewett asked you to do this, right?
Black:
Yes. That was for Dr. Jewett.
Wolff:
This is a stupid question, but is this the negative feedback amplifier?
Black:
There it is. It isn't very small either.
Wolff:
This is your invention that you are showing then.
Black:
This is the first application of my invention.
Wolff:
Oh, well we want to use that photograph in the article.
Black:
That's a field trial. That was not made for the public. As a matter of fact it was used publicly for another purpose.
Wolff:
It was the first application of your negative feedback amplifier.
Black:
Yes. The custom was always to make a large-scale trial before they were turned over to the public for use.
Wolff:
Where was the photograph taken?
Black:
That was taken at Morristown, New Jersey in my laboratory. This picture would have been taken in about 1929.
Biconjugate circuits and distortion
Wolff:
On page 3 you said that by this time when you came back to New Jersey at 2:00 a.m. you had come to the conclusion that anything that was not part of the output was what? Do you want to say that again?
Black:
I regarded as distortion anything in the output that was not a replica of the input, irregardless of whether it was due to nonlinearity, variation in vacuum-tube gains or anything else. That's a broad interpretation that I placed upon the thing. Contrary to a statement in a book by Dr. Terman, that can only be done by using two biconjugate circuits. I think I will have to add some words here. A Wheatstone bridge is a biconjugate circuit. That means that if the arms are balanced and you apply a battery and get no current in the galvanometer circuit. That is a biconjugate circuit. Now I have a large book by Dr. [George] Campbell. He's the first professional mathematician ever to be hired by the Bell System. In this book he listed all of the biconjugate circuits used most widely in telephony at the time of which I speak. They get them in other ways with integrated circuits today. It was with a transformer. At the time, the Bell System could make better transformers than anybody else anywhere in the world. Campbell listed the maximum number of biconjugate circuits of the transformer type that didn't have unnecessary elements. The simplest one of all was the one that I used. I used them always to separate the µ and circuits.
Wolff:
Do I understand correctly that you got the idea to use them from his listing? Or did you know this before?
Black:
I knew it before I ever saw this book. The first thing dealing with voice frequency circuits, for instance an ordinary house telephone uses just a single pair of wire. It has a biconjugate circuit to separate the transmitter from the receiver. It's slightly unbalanced, so you'll hear a little head-through, but the moment you get your central office and begin to go somewhere else the transformers – or hybrid coils as they are known to communication engineers – are used to change from two-wire to four-wire. They have the property of doing that. Then as you go longer distances, unless you are mixing up and getting into multiplexing and many channels, that's where the hybrid comes from. I learned about that by the time I had been around for a couple of weeks outside the training course. At that time any engineer worth his salt would try to do something with the hybrid coil. It was impossible. I remember trying one of those things, and Dr. Harry Nyquist explained to me the error of my ways.
Wolff:
We can get into all that next time.
Black:
Yes.
Harmonics, plotting numbers of amplifiers and channels
Wolff:
We are at our second meeting now with Dr. Black on the June 29th, 1977. The first thing I want to clear up is something that it is one of the turning points of the article. On November 24th, 1921 you had just started to think about improving the amplifiers and you had gone home and were working there on your own time. We talked about what you were doing there, and I got mixed up because we kept jumping back and forth between before and after you talked to Hartley, so that sequence is not clear. Before you talked to Hartley you went home and you were trying to draw a curve. You were coming up with the wrong thing because you were assuming that third order harmonics would act the same as second order harmonics. Right?
Black:
Yes.
Wolff:
While you had this erroneous idea on the 24th and before you talked to Hartley and he disabused you of this, you said, "I prepared a chart based on push-pull and its poor performance of how that requirement would go up as a function of the number of channels."
Black:
And number of amplifiers in tandem. See, there were two things I was doing there.
Wolff:
Okay.
Black:
This Hartley thing had to do with the number of amplifiers in tandem.
Wolff:
Before you talked to Hartley, on Thanksgiving Day, 1921, and with this erroneous understanding in your mind what were you trying to draw? What kind of curve were you trying to draw?
Black:
There were two curves. The first was how the requirements on an amplifier changed as the number of channels varied from one to three thousand. That is a matter of how many voice frequency channels I was considering.
Wolff:
What was the second curve on that same day?
Black:
The second had to do with what additional requirements – linearity really – had to be placed on the same amplifier. Likewise to the function of the number of channels, but primarily with respect to how many amplifiers were to be connected in tandem. Obviously the more amplifiers that come one right after another the more severe the requirement. It was at that latter point that I made my error.
Wolff:
Is it correct that your error was that you assumed that the third order harmonics and other unwanted products would also be reduced to between 10% and 18% just like second order?
Black:
Yes, whatever the figures were on that.
Wolff:
Yes. Just like second order.
Black:
Just like second order.
Wolff:
All right. That was your error.
Black:
To state it differently, it would sort of add up on an RMS basis, but because no push-pull amplifier is perfect even the RMS spaces instead of being random addition was the figure I mentioned. Well I guess that's a technical detail we'll leave out.
Wolff:
I don't understand what you mean there. What do you mean when you say it would add up on a random basis?
Black:
If there were a hundred amplifiers in a row, instead of a three linearity requirement increasing a hundred-fold as it actually does it would only increase by the square root of hundred, or ten.
Wolff:
Wait a minute. N is the number of amplifiers.
Black:
Yes. N is the number of amplifiers. First you have to figure out how good an amplifier has to be, depending on how many channels. Number two, you have to figure out what is the additional requirement. Here is the one that hurts: when you put a big number in series one right after another.
Wolff:
Are you saying that the distortion would only increase according to the square root of the number of amplifiers if what you believed were true?
Black:
Statistics might give a little different answer, but in a rough way yes.
Wolff:
If what you believed was true, namely that the third harmonic was 10% to 18%. Okay. I think I am beginning to see where this got confused. You are talking about doing two different things.
Black:
That's right.
Wolff:
One is the number of channels and the other is the number of amplifiers.
Black:
That's right.
Effects of multiple channels and amplifiers on distortion
Wolff:
That got all mixed up last time we talked. All right. Coming back to your first curve, you were trying to find out how the requirement on an amplifier varies. What do you mean by the requirement? If n is one axis, what is on the other axis – n now being the number of channels?
Black:
The other is how linear it needed to be. To put it another way, the nonlinearity of a vacuum tube of the kind described by van der Bijl in his book on the vacuum tubes available at that time.
Wolff:
You are talking about a push-pull type C system amplifier.
Black:
That was the first one I ever saw.
Wolff:
Yes. This is your push-pull amplifier. You were concerned with how its linearity varied with the number channels? Is that it?
Black:
Yes. It was only carrying three channels and was doing a poor job, so it would be rather obvious if it carried three thousand channels it would have to perform a lot better.
Wolff:
Is the curve you drew regarding linearity versus number of channels?
Black:
That way I put it was decibels of improvement.
Wolff:
Is it correct to say that what you were trying to figure out was how the linearity varied with the number of channels?
Black:
Yes.
Wolff:
Okay. Was the second thing how the linearity would be affected by putting more than one amplifier in series?
Black:
That's right. If you had two in series, regardless of the number of channels it would have to be 6 dB four times as linear.
Wolff:
Are you using the terms linearity and distortion synonymously?
Black:
Yes.
Wolff:
Okay. Then you believed that the distortion in a string of amplifiers would increase as the square root of n.
Black:
Right. If you put the word approximately in front of the square root that would be a little bit more accurate.
Wolff:
This is what you told Hartley you had done?
Black:
That's right. I went down to his office and showed him what I had done.
Wolff:
And Hartley corrected you by saying no?
Black:
He didn't say no right away. He came to my office the next day and handed me a handwritten note that said exactly what I wrote in the paper. It's important to note that he was not talking about push-pull; it was regardless of the kind of amplifier.
Wolff:
To paraphrase what he said, is it correct to say that he told you that the distortion in a string of amplifiers would increase in direct proportion to the number of amplifiers?
Black:
That is correct. The most important thing of all is where he said "independent of the type of amplifier." From that day on, push-pull amplifiers were out. They take twice as many tubes and don't do any good. The third harmonics and all third-order distortion contributed by a string of X amplifiers is virtually X times that contributed by one.
Wolff:
I want to expand on this. Is this the same as saying that the distortion in the whole string will increase in direct proportion to the number of amplifiers?
Black:
That's right.
Wolff:
Will a thousand amplifiers produce a thousand times as much distortion as one?
Black:
Yes. And I might say voltage distortion. There would be a million if we were talking about power distortion. In other words the linearity would have to be improved by the example you gave by 60 decibels. That would mean that if you took about 10 acres and plotted input against the output, the line could not vary a thousandth of an inch. That is a tough requirement. I'd like to add another remark in passing. L5 is single sideband and is the latest and more efficient coaxial [cable]. A single tube, which carries I think 1,028 channels, has two negative feedback amplifiers every mile so that in 5,000 miles there would be 10,000 amplifiers in tandem.
Wolff:
Are you talking about a system that exists now?
Black:
The most efficient coaxial system of today.
Wolff:
I don't know how to reconcile what you just told me about distortion with what you did on November 30th. Your attachment says the business that you "then found for a string of X amplifiers, the distortion contributed by each amplifier must be divided by X."
Black:
That's right.
Wolff:
Is that saying the same thing?
Black:
Yes.
Wolff:
The thousand was when they were all added together. Right?
Black:
That's right.
Wolff:
That's the net distortion.
Black:
That's the net distortion. That is voltage distortion too. If you were thinking of power it would be a million. That statement is precise as it stands.
Wolff:
That's voltage distortion?
Black:
It is voltage or current, whichever way you want to view it.
Wolff:
That's the distortion of each amplifier. It must be divided by X, so the total is going to be X. Right?
Black:
That's right. If you are carrying a thousand channels the thing that had to be divided by X had to be awfully good to even handle the thousand. You get hit twice. The hard one is the number in series because it didn't take long before we had long strings.
Wolff:
Before Hartley corrected you, you believed that if you had a thousand amplifiers in series then the total or net distortion across the whole thousand would only be the square root of that. Right?
Black:
Yes.
Professional advancement in the Bell System
Wolff:
All right. I think I got that straight. Good. That's important to understand. Now let's clear up a couple of other points. You told me a story about how you nearly quit. You were passed over for a raise and you were angry and you were thinking of going to Harvard Business School but you didn't. Did that happen before that Thanksgiving?
Black:
That was very early, well before Thanksgiving. I reported for work on the 5th of July. My salary had been fixed ahead of time. Then along in September of 1921, out of 150 people everybody got a raise but me.
Wolff:
You had just started then.
Black:
I had just started. I started with 150 other people.
Wolff:
And they all started they got a raise?
Black:
The day they were hired all of them that I knew – and I knew virtually everyone – got $27. I got $32. That was no secret. Then in September they got a $3 raise and I didn't get a raise. It was as simple as that.
Wolff:
That's when you got angry.
Black:
I wouldn't say that made me angry. I just didn't exactly like it.
Wolff:
You were thinking of quitting, so it made you angry. Why shouldn't it?
Black:
I'll tell you another story that has nothing to do with this paper. Dr. John R. Carson is a very famous mathematician. In 1925 he heard that he was going to be transferred from AT&T to Bell Telephone Laboratory. Dr. Jewett did that when he was president and drew up a chart for the laboratories. Then Dr. Carson went home and decided that he was not going to work for the Bell System if he couldn't work at 195 Broadway at least on the organizational chart. However the next morning he changed his mind. He too decided that he was going to stick with the Bell System. I'm not the only one to do something like that. If I had been that stupid, it would have affected my life. I still hold that luck plays a part in many things, and I would never have been a businessman. Absolutely no way.
Singing margins and amplifier stability
Wolff:
You were talking about the older four-channel systems that you saw when you first came to work there. You used the term "singing margin." Is that the same as the phenomenon of singing? They talked about amplifiers singing in those days.
Black:
Yes, that is correct. The terms "singing margin" and "phase margin" were used by Dr. [Hendrik W.] Bode many decades later in his book on amplifier design. Those were two terms that I had introduced not in connection with amplifiers but with filters and two-directional transmission. I guess here I was talking about the stability of a negative feedback amplifier.
Wolff:
This was before the negative feedback amplifier. You were talking about the first carrier system that had been designed for President Wilson.
Black:
That's right.
Wolff:
It was unsuccessful because there were some directional filters that didn't have enough singing margin between the two directions.
Black:
Use of the word unsuccessful in this case is a little bit relative. The system did not fail, but didn't work as well as it should. It met his requirements. It served the President. I'll put it that way.
Wolff:
I'm trying to understand this term "singing margin." It is synonymous with the term "singing"?
Black:
Not really.
Wolff:
Well, we don't need to get into it.
Black:
To get into it I would have to make you a diagram that is not concerned with amplifiers.
Wolff:
All right.
Black:
That one system was the only one of its kind that the Western Electric Company ever manufactured. That was called type A. The next one manufactured was type B. They were both open-wire systems. The type B didn't work for an entirely different reason. A few of those were installed. They were not removed from service, but if you listened on channel 2 you could also hear channel 1 or channel 3. That was not considered good. The reason that happened was because it was a carrier transmission system. It transmitted the carrier and one sideband, though not both. With the amplifiers that were available at the time that gave a non-workable system. Therefore very few were produced. The next step was where I came into the picture really. C. W. Green was my boss's boss. I did not report to Green. I reported to a supervisor who in turn reported to Charlie Green, but Charlie and I talked together. He was working on a three-channel amplifier for open-wire lines that would work. That's where I saw and observed the poor performance of push-pull amplifiers. I did not mention it in my text, but everything else was causing trouble.
Lee de Forest and E. Howard Armstrong
Wolff:
Let me interrupt you.
Black:
Yes. You ask questions. Ask me simply.
Wolff:
One more background question about when you came to work in the fall of 1921 at AT&T.
Black:
I was working for the Western Electric Company in New York. I never worked for AT&T.
Wolff:
Okay. I should correct that. Did you hear any talk about the work of [Lee] de Forest and [E. Howard] Armstrong?
Black:
Yes. Let me put it this way. I have to think very carefully. Lee de Forest invented the soft vacuum tube. It's a delightful patent, which I have read. That came up in 1915. That was before I was a member of anything.
Wolff:
That was the decade before you got there.
Black:
Yes. When I got there, because I read every memorandum for file that had ever been written, I soon learned about all of these things.
Wolff:
You read about de Forest's work?
Black:
I read about Dr. Lee de Forest's work. Somewhere between 1913 and 1915, Harold DeForest Arnold introduced a hard vacuum. He beat [Irving] Langmuir of General Electric (GE).
Wolff:
This is Arnold?
Black:
Yes.
Wolff:
Was he related to de Forest?
Black:
No. That is peculiar. If I took you out to the Bell Telephone Laboratories you would see the Arnold Auditorium. It was named in his honor because he was the first director of research. If you have good eyesight you will read his definition of research, which is inscribed in marble.
Wolff:
You read about the work of both these men?
Black:
Yes. Incidentally, at that time at 463 West Street they had a small library measured by today's standards. Within that library was a book by [H.J.] van der Bijl on vacuum-tube amplifiers, which I took from the library and asked if I could keep it. They said yes, and I have it over there in my bookcase.
Wolff:
Did you read about Armstong's work?
Black:
My first contact there was when I received the [John Price] Wetherell Medal from the Franklin Institute in 1941. That was a very formal affair. I had to wear afternoon clothes, then a black tie followed for a cocktail hour and then a white-tie affair for the evening. I was sort of ashamed that I was taking time off from work to receive that medal, so I didn't tell anyone. I was rather delighted in the evening to see Abe Clark (A. B. Clark), who occupied a position that today would correspond to vice president and Roy Chestnut, who was the supervisor on the same level as I but in a different phase of work. In other words, friends from the Bell Labs were there. All during the afternoon I talked to [E. Howard] Armstrong, and that was where I first became familiar not only with his superheterodyne but also his invention of FM. That has a very interesting story.
Wolff:
Didn't Armstrong invent regenerative feedback?
Black:
Yes. Superheterodyne regenerative feedback.
Wolff:
That's positive feedback.
Black:
Yes. That makes the modulation worse, dB for dB.
Wolff:
He invented or discovered this positive feedback, which he applied to his superheterodyne receiver.
Black:
And he was only receiving one channel.
Wolff:
I think he did that before you came to Western Electric.
Black:
That's right.
NOTE:
[Transcriptionist's Note: Here is where Tape 2, Side A ends. Side B of this same tape is a recording of others and not this interview. However, Tape 3, Side A appears to pick up where the above left off. – Michele Deradune]
Wolff:
When you started at Western in 1921 and you were reading all the memorandum for file, did you read any of Armstrong's first papers on his positive feedback?
Black:
Yes, I did. Those papers were available to me. If you will read my basic patent, I mentioned that in the patent as something to be avoided. Well, I would have to see what the patent said.
Wolff:
Here is the patent. Do you think you can find it?
Black:
I'm not certain if I had any reason to make any reference to Armstrong's suggestion – which found wide application until better methods were developed.
Wolff:
This is all part of setting the stage. You came to Western and you read about Armstrong's work.
Black:
Yes. I could say something that worked for Western Electric for about three years.
Wolff:
After you had worked for them?
Black:
After I had worked for the Western Electric Company in New York for about three years, I rented a room in Roselle Park from a woman. She rented out rooms to two people. We were both from Worcester Tech. Across in the next house someone lived, whose name escapes me at the moment, whose brother was president of the Western Electric Company at Kearny, New Jersey. He very much wanted a high-quality radio receiver. Ed Beamus, who was a radio ham, built one for him that I didn't think was very good, so I asked if he would like me to try to build one using a cone loudspeaker. I used Armstrong's circuit. The parts were only resistors and capacitors, and as long as I didn't steal any vacuum tubes I was permitted to get that working in the laboratory on Sundays. That was a personal experience that I had with Armstrong's [ideas], but it had nothing to do with my patent.
Distortion reduction in a single amplifier
Wolff:
Let's take up now we left off last week with the manuscript. You said you were able to reduce unwanted distortion by more than 50 dB. Is that distortion in a single tube or in an amplifier?
Black:
I defined distortion there very broadly. I regarded the input as what was wanted, and then in the output, were that to be attenuated, until it would be of the same magnitude as the input. Anything that was there I would regard as distortion, whether it was change in gain due to a power supply or anything else. I regarded a perfect amplifier as one in which the input was an enlarged copy of the output. In those days the frequency bands were so narrow that the time of transmission across the amplifier was not a practical factor.
Wolff:
You explained that. You said that you could cancel out the distortion in the original output.
Black:
Yes.
Wolff:
Then you said unwanted distortion could be reduced by upwards of 40 dB.
Black:
That was the result of our experiment that next morning.
Wolff:
That was distortion in what? That was in a single amplifier, right?
Black:
A single amplifier, regardless of the number of stages or type – although my patent shows the two embodiments that I set up.
Wolff:
In a single amplifier you could reduce the distortion by upwards of 40 dB. It's at the bottom of page 3.
The problem of developing a perfect amplifier
Wolff:
On the next page you said, "Over a period of four more years I struggled with the problem." What problem was that?
Black:
The problem of developing a perfect amplifier.
Wolff:
What do you mean by a perfect amplifier?
Black:
One that would be viable to manufacture and would overcome all of the distortion, fulfilling the severe requirements that I had set out to meet.
Wolff:
What I don't understand is, this embodiment worked so well. You demonstrated it could reduce the distortion by this much.
Black:
Yes, but it was impractical for two very important reasons. Every hour on the hour, 24 hours a day, someone had to come around and adjust the filament turn to its correct value. They were permitting plus or minus or half a dB or a dB variation in amplifier gain, whereas for this purpose the amplifier gain had to be absolutely perfect. On top of that, four times a day – every six hours – it became necessary to adjust the B battery voltage because the amplifier gain would be out of hand. Those are two practical reasons. There were other complications too, but that's enough to show you.
Wolff:
That's very clearly put.
Black:
For 40 dB, one-thousandth of a decibel change in gain would throw that thing in error by 6 making it poorer by 6 dB. That was just a laboratory demonstration.
Wolff:
That was a 6 dB error in what?
Black:
Instead of getting a 40-dB improvement it was a 34-dB improvement.
Wolff:
That's clearly put and important.
Black:
The amplifier of those days.
Wolff:
It had these two drawbacks.
Black:
That's right. However for me it was wonderful, because it showed that theoretically it would be possible to get a solution.
Wolff:
A solution to what?
Black:
The problem of a superlatively perfect amplifier.
Wolff:
By that you mean that it had no distortion?
Black:
No distortion of any kind.
Wolff:
You wanted to put in a string of them.
Black:
Yes. I wanted a tremendous improvement.
Negative feedback amplifier
Brainstorm on 1927 ferryboat ride
Wolff:
Now we come up to 1927 and the ferryboat ride. This is of course the most important part of your article. What I don't understand here and want to get you to expand upon is what you were thinking about that day. What was the problem? You said you conceived the analysis of the negative feedback amplifier. Can you remember what you were still trying to do?
Black:
What I had been trying to do for years was to come up with something good. The idea of negative feedback and the drawing of the conautical circuit and the equations I wrote came to me all of a sudden as I looked back over the things I had done and published prior to that. I had produced an open-wire system that added one additional talking circuit. There I had designed the filters, and from Heaviside work I achieved superstability in many things. There I actually had a µ. If you understood it carefully enough, I actually had a µ of six or eight. However I cannot connect that in any way with a negative feedback amplifier. I have tried to remember for fifty years, again and again, why the idea came to me. I couldn't tell you today any more than I could that morning. It just came to me.
Wolff:
What exactly was the idea that came to you? This was your sketch.
Black:
This is the sketch. The idea was, here comes an input and then there's an output. The reason I call this µ was due to having read the book by van der Bijl. However, µ here is a complex quantity like Steinmetz first introduced: A + AB. The reason I used the symbol was because in some book in my library was the transmission characteristic of a circuit. I guess it was really Maxwell's Equation. It had to do with the electrical properties of electromagnetic waves. If they were on a pair of wires or a coax […ial cable], the attenuation would vary as the square root of frequency. In other words, was a term that was usually expressed E to the I, to put it in complex form. I noticed here that if was the attenuation of the line it would be an equalizer and would serve two purposes for one. Of course I didn't get any of that information on the sketch. It took a long to get that. You asked what was the reason for my choice of these two Greek letters.
Wolff:
No. I asked you what it was you actually thought. What was this flash that came to you?
Black:
It came to me that if I fed the output back to the input in reverse phase and kept the device from singing I would have exactly what I wanted and I could write this equation. The whole idea came to me.
Wolff:
What did you want?
Black:
You notice that the improvement is substantially equal to the absolute magnitude of µ.
Wolff:
The improvement in what?
Black:
In distortion of any kind – variation in power supply, you name it and this does it.
Wolff:
You said another important thing, if you could keep it from singing. What do you mean by singing?
Black:
Ordinarily when you connect an amplifier that has a large amount of gain back to the input, which has less loss, you are liable to get singing.
Wolff:
Do you mean self-oscillation?
Black:
Yes. There I do mean self-oscillation. I explain here why I knew and what I had to do in order to avoid that. That is all explained, and I would not know how to improve on that.
Wolff:
At the top of page 5 you put that in quotes. Why is that in quotes?
Black:
The quotation marks should certainly be deleted. The only thing would be to decide whether to bother to put that in italics or not.
Wolff:
Okay. You knew that you could prevent self-oscillation by having the loop transfer factor real positive greater than unity [real positive > unity].
Black:
That's right.
Professional communication on work in progress
Wolff:
A lot of people doubted that it would be stable. Is that correct?
Black:
That is correct.
Wolff:
I guess that it wasn't until [Harry] Nyquist and [Hendrik] Bode did their mathematical work that people were convinced. Is that right?
Black:
I would like to abolish that idea at once.
Wolff:
Okay. What is the story there?
Black:
Since Nyquist published first, I will point out that it wasn't until after 1930 that Nyquist even knew that I had developed a negative feedback amplifier. And the same for Bode. I thought the best way to keep others from taking credit away from me was to just go merrily ahead and do it and make believe there was no secret about it at all. I didn't tell anyone.
Wolff:
What did you not tell?
Black:
That I was building a negative feedback amplifier. The Western Electric Company had built seventy-five negative feedback amplifiers in the year 1929 before either Dr. Nyquist or Dr. Bode had even heard of it. They did not know a thing about what I did. By the way, I would like to tell you – since I have this right here and I can also show this to you in an Italian publication that is in three volumes. It lists 3600 people like Einstein – that is, they are not all alive – and I am among the list. The thing that it shows as being the most important figure of my patent – because it does have many figures – is Figure 2. The first that Nyquist ever heard about this thing was in 1930. I asked him why I got these symmetries beyond the one-zero point that I avoided. I think it's in one of my patents. I have got them all and I am going to look and find which one. I think it's my Case #18. I asked him to explain this.
Wolff:
Is that curve number 1 there?
Black:
Yes, or any of these curves, because everything looks to be so symmetrical. To get one of those, you have to have a looped-over amplifier. Of the 10,000 learned papers that I have read, I know of no one but me who used a looped-over amplifier. Bode never did.
Wolff:
I did not make my question clear. I understand that you built the amplifier before they did their work. The point was that a lot of people, as you just said, doubted your amplifier would be stable.
Black:
That is right.
Wolff:
Was it this mathematical work that Nyquist later did later on that convinced people?
Black:
No.
Wolff:
What was Nyquist's contribution?
Black:
His contribution was a paper. I read the memoranda that preceded that paper. That was an intuition and ability to think of something original that is given to very few men. However, his mathematical derivation of the thing is very poor because there are so many other ways of deriving it that are vastly simpler.
Patent; influence of Black's work
Wolff:
Let me ask the question in a different way. How does what Nyquist did relate to what you do?
Black:
In no way.
Wolff:
You mentioned him in your paper. You said, "It was tested before Nyquist developed the stability criterion which bears his name." Didn't that stability criteria have something to do with negative feedback amplifiers?
Black:
It did. It had to do with those that were like the one that I built. As a matter of generality, he introduced a second kind. Yes. He produced a mathematical derivation of what I did. What I would say about that is that it was one that I didn't have any idea existed. It was new to me. I made good use of it ten years later. He derived it, it bears his name, it's known the world over, and that part is right. However I believe that if you look at one of my patents you will find that the part of it that has been used by everyone was something I had already put in a patent. That too is regarded as a publication.
Wolff:
I see. I was just trying to relate it. Who was Blessing? Was he the patent attorney?
Black:
No. Let me explain. He was a supervisor of a group of people. Earl C. Blessing reported to me.
Development challenges
Wolff:
Before that famous ferryboat ride, while struggling with that problem with the feedforward amplifier, you said you used circuits and approaches that were far too complex to be practical. Could you give me an example?
Black:
Not offhand. I could probably sit down and draw a few, but they might not be the ones that I had tried.
Wolff:
That's all right.
Black:
[It was] completely unsuccessful. You can believe that.
Wolff:
Meyer started gathering parts for the model.
Black:
That's right.
Wolff:
You said a fundamental requirement of a repeater amplifier is that its input and output impedance must match accurately. Was this a very difficult thing to do?
Black:
Yes. It had been thought so. It's not an impossible to do. Don't misunderstand me. Before the advent of negative feedback and some of my other inventions it was accomplished at the expense of loss of signal at the input and loss of power in the output. It is a necessary thing to do. I might note in passing that Dr. K. C. Black in the research department was at the same time, around 1936, working on coaxial cables.
Wolff:
Wait a minute. We're back in 1927.
Black:
Yes, so it was not exactly the same time. When he did that I asked him to turn the job over to me, which he refused to do. He was not matching impedances. I told him that if he didn't do that the system would not work. He built a system from Philadelphia in '36. Don Qualls, who was high up in the company, found it and fired him. That's what happened to him.
1927 sketch
Wolff:
That's another story. This was a difficult thing to do, but then on the ferryboat ride you sketched the details of how to do this.
Black:
That's the figure you have there.
Wolff:
That's the figure you have given me. You say it employed a simple novel biconjugate network.
Black:
That's right.
Wolff:
Is that what you've sketched?
Black:
You have to use a little imagination here. I might find a clearer example. This isn't too bad. If you would regard this E as the voltage acting in series with the plate resistance of a vacuum tube and then envision these three things here which carries the—
Wolff:
What three things?
Black:
These are general impedances.
Wolff:
That's rr0, kr and r, right?
Black:
That's right. Those are the things that are the Wheatstone bridge with respect to the galvanometer, which is here and goes back to the input. Z is the load impedance. Then the feedback amplifier is done. It gets a little reflected signal here, and negative feedback eliminates it. The impedances are matched, and the only loss is the loss in this Wheatstone bridge which can be made about a tenth of a decibel.
Wolff:
Where is the amplifier again?
Black:
Regard this as the plate of the last tube of a three-stage amplifier.
Wolff:
Regard what? Where are you pointing?
Black:
This first point.
Wolff:
Number 4?
Black:
Right. This would be the input voltage.
Wolff:
E is the input voltage.
Black:
Yes. Multiplied by the µ of the three-stage amplifier. I warned you this wasn't too clear.
Wolff:
I just want to be able to explain it in the caption. I think that's good enough.
Black:
And you have to do the same thing on the input in order to match impedances.
Morristown experiment, 1927
Wolff:
All right. My question about December 1927. You said you reduced distortion by 50 dB. Is that a single amplifier?
Black:
That is a single amplifier. That used two tetrodes and a pentode. The output of that amplifier was about 10 dB less than Dr. Jewett required in January of the next year. He was satisfied with 40 dB of feedback.
Wolff:
Where does Dr. Jewett come in now? In Morristown?
Black:
That's in Morristown. I did not put that in there at all.
Wolff:
I think it should be in there.
Black:
That was why I wrote something for you. You may want a reference to that paper describing those experiments.
Wolff:
Before we get to 1928, let's come back to the top of page 6. You said you submitted an application on August 1928 to the Patent Office.
Black:
I submitted a drawing. I had many people working on this, so I don't recall whether it was Steve Meyer or a different engineer. I submitted a drawing of the amplifier that accomplished this, and that was the basis of the main patent.
Wolff:
The patent that was then filed in 1932?
Black:
That's right. All the work that had preceded that occupied a large amount of time. That one had the biconjugate networks and 6 dB improvement and matching impedances and everything else.
Wolff:
In January 1928 the carrier system for transcontinental cables began to be developed. I think we should describe that.
Black:
What page are we on now?
Wolff:
We're not on any page because you didn't write it up, but I found this in what you gave [Julian] Tebo, and I think we ought to include that. This is about the Morristown experiment.
Black:
Yes.
Wolff:
You have given me some material on it here. The interesting thing is that Dr. Arnold didn't believe your amplifier would work.
Black:
What do you know about that. Now you got right to the Bell Labs. Not to all the other people and not to England, but Dr. Jewett for the most highly prized possession in the Bell Labs. I thought you might perhaps want to include that. Or maybe you wouldn't.
Wolff:
I do. It's all very interesting.
Black:
It's all very true. Did I write it in a way that you can understand?
Wolff:
Was Arnold Dr. Jewett's boss?
Black:
No. Dr. Jewett was president. Dr. Arnold was not a vice president. He was the director of research, third down in the line.
Wolff:
Okay. Arnold did not believe it would work and he instructed you to build a powerful Colpitts amplifier.
Black:
That's right.
Wolff:
What is the point of this? You proceeded to assemble the parts for this Colpitts amplifier.
Black:
I got them all together. In the meantime, every engineer I had was working on this other job.
Wolff:
The other job being the negative feedback amplifier?
Black:
The negative feedback amplifier for Dr. Jewett.
Wolff:
You were working on two amplifiers.
Black:
All that I did – and I did no more – was to have A. G. Garns design the transformers and collected the condensers and inductors. I got the parts necessary to built one Colpitts amplifier if that proved necessary. I didn't feel that it was going to be necessary, but I did exactly what he told me to do.
Wolff:
That's fascinating. Meanwhile you went ahead with your own and found out that it worked.
Black:
Yes.
Wolff:
That's what that photograph is that you gave me last week. Right?
Black:
Yes. That is a physical picture.
Wolff:
That's the Morristown experiment.
Black:
Yes. I am taking out one of the condensers there. Charlie Green thought I should be fired.
Wolff:
Why?
Black:
Because I had a come-and-go trouble in it, and he thought that I had done something wrong. The people who made the condenser said that if there was anything wrong I would have to find it. It cost me about $100,000 of my engineer's salary. I found that the method of making the contact between the foil which was insulated to get a capacitor was loose. It would go on and off. I don't think that it would be pertinent to do that, but you can notice I was looking at the animal sort of seriously.
Wolff:
That's an interesting detail to put in the article.
Black:
Then I was on to confirm that with F. A. Brooks because I designed more than one amplifier.
Condenser troubleshooting
Wolff:
Wait. I'm confused. This is a photograph.
Black:
What I am taking out in that photograph looks to me to be a condenser. If that is true, then I believe that it is this faulty condenser that I had so much difficulty troubleshooting.
Wolff:
The important thing to clarify is that when you tested these amplifiers there was a problem with the condenser. Is that it? I don't think it's important whether it was that particular condenser. I just want to tell the story of this Morristown experiment.
Black:
Fred Brooks is alive, and unless he is on vacation I will be able to reach him by telephone and find out whether what I am saying is true or false. I would never put that in unless I could prove it. This part however is the absolute truth. I built the push-pull.
Wolff:
During 1928 you tested eight models of a negative feedback amplifier.
Black:
Yes.
Wolff:
Was it one of those models that had this trouble with the condenser?
Black:
No. There was no trouble there.
Wolff:
Where was there trouble with the condenser?
Black:
I don't remember. I'll have to check up on that. Regardless of whether or not that condenser was at fault – and if it was it was quickly fixed in just about a week and a half – the experiment was a complete success. It is fully described in a brief paper that you could make reference to if you like.
Wolff:
Have you got that paper?
Black:
We aim to please. [recording turned off, then back on...]
Patent application review process
Wolff:
On page 6 where you are talking about the patent, you said, "The concept was so contrary to established beliefs that the U.S. Patent Office reported that experts the world over delivered lengthy opinions stating it wouldn't work." To whom did they report this?
Black:
When an invention is submitted to the Patent Office the first thing they do is make a search. They are well equipped to do that in a hurry. They want to make sure that the same idea has not been patented by anyone else in this or any other country. Next, they want to make sure that it is not already described in any book, technical paper or article. Finally, they undertake to decide for themselves whether or not it satisfies their requirements. It was in that process that they found papers in France, Italy and other countries saying that the output could not be connected back to the input unless the loop gain was less than 1. Whereas my loop-gain was 40. Actually I experimentally got 50 in because I allowed for manufacturing variations and so on.
Wolff:
Is loop gain a product of µ?
Black:
That is exactly right. It is the absolute value of the product of µ. The U.S. Patent Office found those papers, and each one and said it could not be done.
Foreign patents
Black:
Did I tell you what the British post office said?
Wolff:
No.
Black:
Here in the United States one makes an invention of what is the old-fashioned idea of perpetual motion. Some very intriguing things have been so proposed. I was taught in school never to do that. The U.S. Patent Office then – and only then – says, "Submit a working model." That is then the last that is ever heard of that invention. My inventions were patented in many foreign countries. That's very costly and important, and they have to go into use within [a certain period of time]. Now my invention was clearly in the field of electronics. Curiously, Great Britain has a rule that in electronics that if they have reason to believe that it does not work they may request the inventor to submit a working model. In my case they said the negative feedback amplifier would not work and asked me to kindly submit a working model. Harry A. Burgess then got affidavits that seventy such amplifiers were working in the telephone building at Morristown, New Jersey.
Wolff:
That's fascinating.
Black:
He also sent pictures. In view of that and the fact that it had been described in two publications by that time – my own plus the other to which you just referred – they conceded that it could be patented in Great Britain.
Wolff:
That's very nice. All of this kind of material would be nice to work into the article. There were objections to the length and arguments about the claims.
Black:
That is right.
U.S. Patent Office objections; patent claims
Wolff:
Who objected to the length and what kinds of arguments were made about the claims?
Black:
Every time the U.S. Patent Office objected to a claim, I stoutly refused to concede the thing and immediately proceeded to add one more claim. In that way I gradually built the number of claims up to 126. Burgess would not allow what would have been Claim 127 because he thought it would not stand up and might thereby invalidate the entire patent. As to their objection to the length, well they just thought that it was complex and long.
Wolff:
All right. That's clear. Did they argue with you about claims before the patent was issued?
Black:
That's right. They – and only they – decide how many claims are going to go into it.
Wolff:
Why would they argue about a claim?
Black:
They can for instance argue that it is too broad. The patent starts out with wave translation. That can be electrical, acoustical, mechanical or hydrodynamical. Any medium that will transmit a wave is a wave translation system. Those are the first words in the description of my invention.
Wolff:
I was interested in that. That is what you meant by the new material you wrote for me about applications. You envisioned broad applications.
Black:
That's right. That of course is where the claims come in. I would say that the value of a patent lies in its claims. The reason for the long description was that this was a new field, so I gave a lot of examples. The examples I gave were amplifiers I had set up in the laboratory.
Wolff:
Are there wave translation systems in the claims that are not electrical?
Black:
When it comes to claims, there are technicalities. By the way, here is a little detail where I told in my text how much of the patent was devoted figures, how much to description and how much was claims. The first three lines of the second claim are mainly description. The last page of the claim is vacant, so that I don't know just how you would say where the description ends and the claim begins.
Wolff:
I'll figure something out about that.
Black:
I never realized that situation existed until the other day. There is another thing I would like to mention having to do with the figures. If you look at some of these figures, if they are examined and if for instance you call this Figure 35, some of these can really be different from others. If the figures were counted the way an engineer would count them it would come to a larger number.
Bell Labs response
Wolff:
Why would Dr. Arnold have been doubtful that your amplifier would work? Was it the stability concern again?
Black:
My only answer to that is that I could not read his mind, but he was convinced. The most natural reason would be the one you suggest. I boldly suggested putting a tremendous amount of gain into the µ rule. Yet I was claiming that the amplifier would not sing. He was educated as a physicist. He got his Ph.D. from Chicago and came to work for Western Electric in 1911. He was hand picked by Dr. Jewett. My answer is that I don't know. I don't even know why the British post office and or other people in other countries thought it would not work.
Wolff:
I am trying to get a sense of whether there were many people in the laboratory who were skeptical that your amplifier would work.
Black:
No, not in the Bell Laboratories. The reason for that is that I was about the only one who was concerned. I was not consulting with anyone else. In those days if I wanted a transformer, a condenser, a resistor or anything else, I got it from the apparatus department. That was under William Fondella. If I wanted to get my amplifier mounted I went to the equipment department.
Wolff:
You are saying that you did it yourself.
Black:
I assembled the things in the laboratory. First I got all parts of the apparatus and placed requirements on that. Therefore that if the apparatus met its requirements, then the system – in this case was the amplifier – would meet its requirements. It would be all right. I took into account a liberal amount for variations – manufacture, temperature of offices and all factors including changes with time. I then had to cooperate and change the way that I mounted it to conform to the way the equipment division wanted it mounted. They had the final say on that before for the Western Electric Company manufactured these amplifiers. That laboratory trial involved a great many things.
Wolff:
Are you talking now about your first one?
Black:
The eight amplifiers that we got together.
Suitability for transcontinental telephony
Wolff:
I just realized that there is a gap we need to fill. You said that on December 29, 1927 you demonstrated that the amplifier "was more than sufficient to do the job I undertook six years earlier." What job was that? Was then when you realized this was an amplifier with which you could put a hundred of them in a string?
Black:
I sure did. I envisioned hopping across the nation.
Wolff:
And it was at this point that you knew that you could do that?
Black:
I knew that I could do that. Yes.
Wolff:
With this test. All right.
Frank B. Jewett
Black:
The thing that I did not realize was that it did not have enough power. That was the main difference between the first working model and what I built for Dr. Jewett.
Wolff:
Let me get to Dr. Jewett now. You built this first model with Meyers ?
Black:
Yes, and my other colleagues.
Wolff:
What happened after that? Why did Dr. Jewett, on the next page, suggest to Arnold that you build this amplifier?
Black:
I was very friendly with Dr. Jewett, and he knew that I was working on a negative feedback amplifier. He also found out that I had succeeded. It was pure coincidence that my success, which was at the end of the year, coincided with his suggestion in January.
Wolff:
Were you still in the systems group?
Black:
Yes.
Wolff:
Was this at West Street?
Black:
That was at West Street, and the head of that group was Amos (A. F.) Dixon. I could show you that on the organizational chart.
Wolff:
That's important. That was a gap we had here. You were friendly with Dr. Jewett and told him what you had done?
Black:
Sure. You bet.
Wolff:
What was his reaction? He believed you?
Black:
Oh yes. Well, I am complimented that he had that faith in me.
Wolff:
Why wasn't he worried about stability?
Black:
I told him that I had already gotten one and it had worked.
Wolff:
Did you demonstrate it to him?
Black:
No. I went to work for the laboratories in 1921. I had belonged to Tau Beta Pi and Sigma Si while going through college. I don't know whether he was president at the time, but I believe it was AT&T that brought together a large group of people who belonged to Tau Beta Pi. That was when and where I was first introduced to Dr. Jewett.
Wolff:
What was his full name?
Black:
Frank B. Jewett. That was my first meeting with him. He was very friendly. Of course at that early date I had not yet gotten into this feedback or anything, but that was my first contact with him.
Power of amplifier models
Wolff:
Were the eight models of the amplifier powerful enough?
Black:
Yes. It took two tries to get that power. Dr. Kelly designed every vacuum tube I ever used in any amplifier I ever made or with which I wanted to experiment. Kelly gave me two high-grade pentodes having a little more power than any of the tubes I had used in December. Then we came to the stickler. What I was going to use in the last stage? At that time there existed no directly heated pentode in Morristown that would deliver the power. Fortunately they only wanted 40 dB of negative feedback. The equalizer was designed by F. A. Brooks. He reported to me. To my disappointment, however, the equalizer was not in the feedback path. Therefore all I needed was the amount of gain shown by that. Let's see. The slanting is the low loss and then the equalizer goes this way and those two losses cancel out. One could figure it out. Anyhow, I needed only modest amount of gain, practically enough for the two pentodes given to me by Dr. Kelly. When we came to the last tube we decided that would be a triode. Then I made a suggestion. Now a suggestion is not a patent, and McNally is the one who got the patent. The [unintelligible word] vacuum tube has a grid, a filament that is not indirectly heated and a plate. I proposed that we put in two grids. We worked fast on this job. One good thing about Kelly is that he was a fast worker. And he expected those who worked for him to do likewise. We put two grids in the same plane. We gave one of them 60-volt wires that were negative and the other 30-volt wires were positive. They were given the opposite polarity. In that way we got a wider swing without increasing the distortion.
Wolff:
A wider swing in what?
Black:
On the voltage that was put on the grid. If there were only one grid instead of two, only half as much swing could be achieved. We expected a 6-dB improvement that way.
Wolff:
Is this what gave you a higher output?
Black:
It gave me 6 dB higher output, but unfortunately that was not enough. The next step was to do what the Bell System had refused to do for one hundred years. They raised the B battery voltage from 130 volts to 250 or 260 volts – whatever my paper showed. In other words, we did it by brute force. That gave the required power and completed the success.
Channels, transmission, and frequency
Wolff:
To close the loop on this whole article, what you have not told me is how many channels such an amplifier could have.
Black:
Nine channels. That's given in the paper.
Wolff:
And you could put 160 of them in tandem?
Black:
Yes, with spacing at every 25 miles, as Dr. Jewett's requirement was 4,000 miles. The transmission, due to any and every cause, was not to be out by more than 1 dB. Today they permit 4 dB because of the improved handsets and other reasons.
Wolff:
Okay. That takes care of it.
Black:
The frequency allocation and everything is given in that article.
Challenges to Black's patent
Wolff:
Was your patent ever challenged?
Black:
Yes. It was challenged once. I licked it in a day. Let me see who did that. [Inaudible name; Tillet?] challenged it. They produced a patent, and I showed that the patent they produced would not work. It had never even been set up in a laboratory.
Bell Labs suit against Zenith
Wolff:
Somehow I had it in my head the idea that there was some problem with Zenith.
Black:
That was something entirely different. Zenith manufactured radios, television and a lot of electronic equipment, including some stereophonic. They had a patent attorney who told the president of Zenith that there was no patent issued by the U.S. Patent Office that he could not break. He told Zenith to advertise in newspapers, magazine articles and all over the place that they were using my patent, the negative feedback amplifiers. They did that and showed the places they were using them and so on. Then that same patent attorney died, leaving Zenith to battle it out without him.
Wolff:
Did Bell bring a suit against them?
Black:
Yes, they did. Incidentally, most of my patents have been in the communication field. [recording turned off, then back on...]
Wolff:
Your testimony lasted for years?
Black:
I think it lasted about five years.
Wolff:
Was this was on how you invented the negative feedback amplifier?
Black:
It was on this patent. That was a very peculiar arrangement. There was a long, huge table. The opposition sat on one side and I sat all alone on the other side. Then to one side was the court stenographer and the person running the tape Coincidentally, the next person there was a man by the name of Kelly who used to be in the same office with me at 183rd Street. During the Depression he got fired, went to South America, didn't make out well there and subsequently got tied up with Zenith. The first morning they asked me a question having to do with radio broadcasting. I proceeded to answer the question at great length. Every sentence, and some phrases, of my answer led to six or more questions. We began about 10 o'clock in the morning and it lasted until 4 o'clock in the afternoon. When there was a break, the AT&T attorneys would all descend upon me at once. They said, "Harold, to any and every question that you are ever asked, say 'yes' or 'no'. And before you do that, take a good long time. And if you hear one of us cough, say nothing." We adjourned for lunch, and then in the afternoon came question number one. I think it had to do with why there were so many claims, being 126. I heard a cough, so I said nothing. Then they repeated the question. After about five times the attorney for Zenith said, "The stenographer will kindly take note that the witness is highly uncooperative." Then he said a few other disparaging remarks about the patent and then said, "At 10 o'clock tomorrow morning I shall bring this to the attention of a certain judge in the City of New York." Then AT&T attorneys said to me, "Well, you just see to it that you don't appear. We will appear for you." They argued the question that day without me there. I was told that the attorney said, "Mr. Black is a very busy man and valuable to the laboratories and it was an unreasonable question," and the judge said, "Yes. I agree with you. He proves the entire patent." He said, "I will make two suggestions. One is that Mr. Black be given ten days to prepare the answer. Secondly, that he be allowed to have a typewritten copy of the answer to the question." Then six of us worked the entire ten days for 12 or 20 hours a day and typed up an answer that took me over two weeks to deliver.
Wolff:
When was this case? Was it in the '30s or '40s?
Black:
No, it was probably starting about 1948 and lasting until '53.
Wolff:
Was this a suit brought by AT&T against Zenith and RCA?
Black:
RCA and Bell (or AT&T) were suing Zenith.
Wolff:
It was not against RCA? It was against Zenith?
Black:
RCA had a couple of patents and they didn't care much about, but when it came to my patent the Bell System was prepared to carry that all the way to the Supreme Court.
Wolff:
Did Bell sue RCA?
Black:
No. Bell sued Zenith. RCA sued Zenith too, but they didn't get anywhere.
Social impact of negative feedback
Wolff:
Good. I think this takes care of it. Do you have any general thoughts looking back over this work that started more than fifty years ago? I imagine that the wide success of the negative feedback has been very gratifying to you.
Black:
Yes.
Wolff:
Was it surprising too?
Black:
No, I wouldn't say it was surprising, because I understood its importance from the beginning. Time moves very swiftly over a period of fifty years. I will give a few examples. Thanks to negative feedback, half a billion people watched man take his first steps on the Moon. It was really a lot more than that, because even the launch itself involved feedback. In the picture you have of one of my amplifiers, you can see that the length of my feedback path was almost a yard and a half.
Twelve-channel transmission
Black:
[Nowadays] the length of the feedback path in the coaxial is the thickness of a piece of cardboard and the size of two amplifiers. They couldn't do it in one step because they were handling so many channels and the thing varies as the square root of S, so they had to use two amplifiers every one mile.
Wolff:
When was this?
Black:
Coaxial?
Wolff:
No. Two amplifiers every mile. I lost the point. Were you talking about coaxial cable today?
Black:
The coaxial cable of today. The coax has a long history starting in '36, but with the coaxial of today a single pipe carries about 1,028 analog voice circuits, and at every mile they have to put in a repeater. Due to the large number of circuits, that really has to be two three-stages amplifiers having about 150 dB of negative feedback – as compared to my 60 or 65 over a 12-channel K.
Wolff:
Over a 12-channel what?
Black:
K1 and K2 transmitter, the 12-voice circuit. The coax. I'd have to check on the exact number, but I think it's 1,028.
Wolff:
What is K1? Was that your first system?
Black:
K1 is a 12-channel.
Wolff:
When was that made?
Black:
The K1 was made in 1940, and the K2 was made in 1942. Each one had the equalizer in the feedback circuit, but the K2, in order to get more out of things, used a looped-over amplifier. Twelve channels was the most I ever did, and my amplifiers were bulky.
Wolff:
This is important. I want to make sure we understand it. Your negative feedback amplifier eventually got to where in 1940 it could transmit twelve channels.
Black:
Instead of nine channels.
Wolff:
And you are saying that today the negative feedback amplifier and coax systems can transmit over a thousand. Right?
Black:
That is right.
Wolff:
You are contrasting the improvement.
Black:
And I am calling attention to the fact that the two amplifiers are about the size of the palm of my hand – 3 or 4 inches.
Wolff:
These are the amplifiers in today's coax system, right?
Black:
Yes. And the length of the feedback path was a little thinner than a piece of cardboard. The bulk would differ by more than 10,000 to 1.
Wolff:
The size of the amplifiers in bulk between 1940 and today?
Black:
Yes.
Wolff:
Differ by more than 10,000 to 1. That's beautiful. That's nice to have that comparison.
8-channel open wire system amplifier
Black:
I would like to tell you about another amplifier I designed, which was for an 8-channel open-wire system. That was the best negative feedback amplifier I ever designed, because everything was fed back through the transformers. There were three transformers. It did the impedance matching and had some very unique properties. It had equipment that went with it to deal with the fact that in some parts of the country the ice would form an inch around. These were used in fairly large quantity. It was a horrible system really, but the amplifier was a beautiful one. During World War II a hundred or more of them were given to Russia – of all places.
Wolff:
For use in what? Anything in particular?
Black:
I suppose it was due to the fact that they have a lot of open country there.
Wolff:
Did you ever have the experience that the inventors of the computer had? Initially they couldn't sell them because people didn't think they would ever be practical. Did you run into the resistance like that in the beginning?
Black:
No. It has sold from the beginning.
Wolff:
Only some skepticism by Arnold and a few others.
Black:
That and the Patent Office and the articles that had been published for entirely different reasons.
Wolff:
It's an exciting story. I appreciate all your time.
