IEEE
You are not logged in, please sign in to edit > Log in / create account  

Oral-History:Walter Proebster

From GHN

(Difference between revisions)
Jump to: navigation, search

Revision as of 05:12, 10 August 2009

Contents

About Walter E. Proebster

Walter E. Proebster
Walter E. Proebster

Proebster was educated in at the Technical University in Munich. After completing his Ph.D., Proebster went to work for the IBM Research Laboratory in Zurich. From 1962 to 1964 he was director at the research laboratory in Yorktown as Director of Experimental Machines. He then became head and later director of the IBM Development Laboratories of Germany in Böblingen. At the time of the interview, Proebster was Extraordinaius Professor at the Technical University in Munich teaching computer science.

The interview mainly covers Proebster’s Ph.D. work and his years at the IBM Zurich laboratory. Proebster discusses his Ph.D. work with Robert Piloty. He describes the difficulties between the mathematicians and the rest of the team in building their machine. Proebster discusses the newness of the field and the reluctance of Telefunken and Siemens to become involved. Proebster relates anecdotes about the various difficulties they encountered during his research. Proebster describes his work at IBM’s Zurich lab. He discusses the founding of the laboratory and the interaction between IBM in the United States and in Zurich. He discusses his interest and difficulty in working on magnetic films.

About the Interview

WALTER E. PROEBSTER: An Interview Conducted by William Aspray, IEEE History Center, July 1, 1993

Interview # 172 for the IEEE History Center, The Institute of Electrical and Electronics Engineers, Inc.

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, 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:

Walter E. Proebster, an oral history conducted in 1993 by William Aspray, IEEE History Center, New Brunswick, NJ, USA.


Interview

Interview: Walter E. Proebster
Interviewer: William Aspray
Place: Munich, Germany
Date: July 1, 1993

Education and Work on PERM Machine

Hans and Robert Piloty

Aspray:

This is an interview on the first of July, 1993, with Dr. Walter E. Proebster, at the Deutsches Museum in Munich. The interviewer is William Aspray. Can you begin by telling me about your early life?

Proebster:

Basically I grew up in Munich. I studied at the Technical University in Munich from 1947 to 1951. At this time there was no computer course at the university. There were computers, but there were no computer courses.

Aspray:

What was your degree work in?

Proebster:

I was working on electrical engineering, and we had two tracks: one power, and one telecommunications. I decided after the second year to go into telecommunications. I was lucky to make my diploma thesis with Professor Piloty, Hans Piloty. I didn't do too poorly, so he asked me to become his assistant to construct the PERM machine, which you have probably seen.

Aspray:

Yes.

Proebster:

This started around 1950, and I was actually the first full-time paid assistant from the DFG: Deutsche Forschungs Gemeinschaft. The assistants usually assist the professors in conducting their studies, teaching courses, supervising exams, and participating in the lecture work. I was relieved completely of that chore and could devote my full time to the construction of the computer. Now, you probably know this story from other sources.

Aspray:

It would be nice to hear it from your point of view, though.

Proebster:

There is a son of Professor Hans Piloty, Robert Piloty — Have you interviewed him?

Aspray:

No, I haven't.

Proebster:

It may be worthwhile to get his viewpoint also. For many years he has been professor at the Technical University in Darmstadt, and he has now retired. He lives in the vicinity of Darmstadt. At that time — I think it was in 1949, but he must be able to tell you more precisely — he visited the United States and got in touch with computer work. Now, Professor Hans Piloty (his father) was quite interested in computers, not for the computer art in itself, but for the application of computers. Professor Hans Piloty — who has passed away — was an expert in filters, particularly in filter theory, but also in the construction of filters. He was working together with Caver. Do you know his name?

Aspray:

I don't know it.

Caver's Work on Telephony Filters

Proebster:

He was working together with Telefunken on the construction of extremely high-quality telephony filters for multiplexing where you have very sharp band pass and many of them in order to have multiplex operation on high-speed telecommunication links. The precise design, the high-quality design, of these filters demanded the solution of very complex operations: Chebyshev polynomials of very, very high order. Before I came and actually at the time when I came, the assistants of Professor Piloty were engaged in solving these equations by electromechanical calculators. I recall that a typical work extended somewhere between four to six, maybe eight, weeks of work with such an electric calculator of a most sophisticated design — with many, many decimal places in order to get the accuracy to solve these equations precisely: the zero points, the poles, and the zero places of the polynomial. I recall whenever my colleagues at this time caught a mistake during this four to six weeks, the whole work was for the birds. They had to restart. This was one of the prime motivations of Professor Piloty to look for a computer to do his work in an easier way. When Robert Piloty came back, he came back with microfiche of the Whirlwind computer and also of some of the text of the original work of J.V. Neumann and H. Goldstine.

Mathematicians Join the Team

Aspray:

From the Institute of Advanced Study.

Proebster:

From the Institute of Advanced Study. When I finished my diploma work, and actually even a little bit before, I was asked by them to study this material. I worked together particularly with Robert Piloty on the idea of how to construct a tube computer at the Technical University of Munich. I worked four years on that and finished with my Ph.D. thesis on this machine. If I look over the four years, the requirements increased steadily. At the beginning we had very humble goals, and the more we got into the field and the more encouragement we saw by positive results of our work, the more ambitious we ourselves were. We were also challenged by our mathematical colleagues, particularly by Dr. F.L. Bauer and by Dr. Samelson, who joined us later. Professor Bauer may recall this much better, but my recollection was somewhere in 1952 Dr. Bauer and Dr. Samelson joined. Also, Mr. Schecher joined from the mathematical side.

Aspray:

How do you spell his name?

Proebster:

Mr. Schecher is famous for inventing the indirect addressing, the address of an address. Sequence address of an address. There are many, many interesting stories of this time. But let me first start from the outside. Professor Piloty realized soon that it would be very, very valuable to engage one of his mathematical colleagues in this work. We were so fortunate that Professor Sauer from the mathematical side showed an interest. As it is the case in any development, as in any team, you cannot avoid that certain personal frictions occur.

Aspray:

Sure.

Proebster:

Some are negative in effect, and some are positive in effect. There are frictions, there is jealousy, there is the ambition to do better, and so on. By in large, I think in this field there are two poles: the mathematical center and the electronics center. I think lots of discussions went and certainly also personal affections. I shouldn't avoid to mention that. Professor Hans Piloty and Professor Sauer were like godfathers looking down at us, and to kind of see that peace, by in large, resided in our camps.

Demands of Mathematicians

Aspray:

Can you give me an example of the kind of discussion that went on between the mathematicians and the electrical engineers? Were the mathematicians looking for certain kinds of capabilities to be built into the computer?

Proebster:

Yes. Definitely so. In the beginning we were more or less interested to have a working machine solid enough. Professor Piloty thought that he was looking for a machine solid enough to have at least the minimal requirements to solve his equations. I recall that, at least in the beginning, he was certainly pleased to have a machine without exponents, just with mantisses. He was thinking of having some way to live with this constraint. When Drs. Bauer and Samelson came, they certainly pushed very hard to extend the way figures are expressed including exponents. That caused quite an extension of our work because the machine originally was relatively small. We started with just eight binary places and one sign. Then we got hungry and extended that to 40 places with a sign. Then the exponent was added, and the machine got larger and larger and larger. Initially, we had just one room in which to build the machine. It was in a building across the tower of the Technical University. The machine got larger and larger and larger. So we cut the door out of the room, and the machine extended into the hallway. [Chuckling] The toilet then had to be taken out to extend the machine because the machine got so big. This was one example, to expand things and to add exponents.

Another example, a second example, is that Drs. Bauer and Samelson were quite interested in the question of rounding errors. It was felt that particularly when solving very long equations, the errors would accumulate if you didn't do proper rounding and no useful result would then be obtained. So they got engaged quite eagerly, and asked the engineers to have several different implementations of rounding operations which you could insert in the machine in a way like modal operations. You could put the machine in different modes.

Aspray:

I see. Yes.

Proebster:

Depending on what mode the machine was in, the rounding would be performed differently. If these ideas would have been formulated in the very beginning, it would have been relatively easy to implement these demands. But they came at a relatively late point in time when the machine was relatively solidly constructed already. To add even small functions caused a lot of difficulties to implement these things. Drs. Bauer and Samelson were stubborn, so that Piloty had to shield us and said to them, "Go away. We have to have the machine running." They insisted and kind of crawled under him and talked with: "Couldn't you do it anyway? You are such a genius." [Laughter] "Find a good way. Couldn't we do it with small components? Can't you do it after all?" So we implemented these functions, and you can see these were some of the frictions between Robert Piloty, who thought he was the master of everything, while Drs. Bauer and Samelson who wanted to have certain functions added.

A third example was that Hans Schecher had — I think it's Hans his first name; we'll have to check — the idea of indirect addressing. This could also have been implemented very easily in the beginning if they had known because it was a relatively small operation. But the machine was almost finished. But we added this. We could add this function. So the machine grew and grew and grew. I don't know how much you're interested in the stories from the electrical side. There are many, many interesting stories to tell.

Magnetic Drum & Mechanical Problems

Aspray:

I am interested.

Proebster:

I'll give you some of them which strike me. This Hans Otto Leilich, now a professor at the Technical University of Braunschweig, who has retired quite recently — might be very interesting for you to interview him. He was my colleague for many years. He was almost full time working on this machine, but not quite. He was engaged also in some tasks to help on the teaching and examination side as far as I recall. He concentrated on the construction of the magnetic drum, which you probably have seen.

Aspray:

Yes.

Proebster:

So the division of the work was that under the overall guidance of Robert Piloty from the engineering side, my task was to construct everything except the magnetic drum. So I was in charge of everything from the input and output, and the central processing unit, and the conversion of decimal numbers into the binary form and back from the binary form to the decimal form, the construction of the electronic circuits and the construction of the CPU and the decoders and whatnot. It was not just the two of us; we certainly had a lot of help from other people who were surrounding us. But they were not working full time on it. Dr. Macha, for example, helped Hans Leilich very much in the construction of the drum. But certainly not full time. Some of the critical experiments were performed together with Dr. Macha and Hans Leilich, for example. The drum rotated at a speed that was very critical with respect to explosion. They tested this drum first in a cave in order to be sure that it would not hurt people. Also with respect to the selection of the bearings, it was not just Hans Leilich, but also Dr. Macha who had helped to solve some of the problems. Other critical problems at this time were the placement of the read/write heads over the drum, the coating of the drum, and the polishing of the drum. Endless mechanical problems which are similar today in magnetic disc application. But we experienced them in a way first and had to solve them step by step.

Aspray:

What did you know of work on drums at other places?

Proebster:

You would really have to ask Hans Leilich. It is certain that we had material from the trip of Robert Piloty to the States, mostly in the form of microfiche that we read for many, many weeks. The material which we got was not complete. It was only spotty. Unfortunately we had nobody to discuss difficult questions with because there were absolutely no people around us whom we could ask. Only later, somewhere in 1953 or 1954 were there were conferences organized where those researchers in Europe came together and exchanged their ideas. There were some of these conferences in Göttingen, for example. The Göttingen machine, a serial machine, was constructed. And in Darmstadt there was Professor Walther where another computer was built, and the ETH (Eidengenössische Technische Hochschule) under Dr. Speiser in Zurich. Professor Zemanek in Vienna has his machine. But nevertheless, it can be said that each of these research teams had so many problems of its own that there was very little overlap. We didn't have very many people at all whom we could ask for more literature, for example. One man who should be mentioned in this connection is Walter Hoffman, who was at this time an assistant of Professor Walther in Darmstadt. Walter Hoffman engaged in literature search and literature collection. Later he became a patent lawyer and worked for many, many years with IBM in one of the patent departments. He has now retired from IBM, and I spoke with him two weeks ago. He's now in Munich. You may contact him. He is still an active patent lawyer. He has a wealth of historical information on developments of this type and also how they eventually led to the state of the art which we have now. He helped quite a bit in getting literature and in getting literature spread around.

Newness of the Computer Field

Aspray:

When you were working, say, on the design of the drum, were these kinds of problems ones that, although you didn't have other people who had the experience, from your basic training they were surmountable? Was there something from the background in doing other kinds of work that you could apply?

Proebster:

No. It was very fascinating for us. I don't know how to express it.

In some way I found myself as a loner in the middle of people which really should know what's going to come about. One didn't have the time and the means and the words to explain to people what a fantastic new world was going to explode. It was very difficult to find resonance about these ideas with respect to other people. I will give you two examples: Around 1953 I approached Professor Hans Piloty and told him that I think this is such a new fascinating large field which we are just going to explore, that I feel it's absolutely mandatory that we have a large university effort to disseminate this information to the students and other workers. I thought even founding a faculty to consider this idea would be a worthwhile idea. Professor Piloty told me, "Young friend, young man, maybe we'll think about one course, but not much more."

I give you two more examples: I tried to explain it to my father. My father had been working for Krupp and had a large organization. He was a very open-minded man, and he more or less listened to me in a friendly way, you know: "Your crazy ideas." [Laughter] But a third one is more serious. I was conducting my studies as compared to today's standards very fast. It took only four years for my diploma and four years for my Ph.D. I finished my first exam after two years and made all tests at once. And I got elected to the Studientstiftung des Deutshcen Volkes, which is a sponsoring organization for the students who excel. I didn't need money, so I was kind of an honorary member of this club. We met about once a month in Munich, students together with their peers, plus two or three professors. Not just electrical engineers, but architects, mechanical engineers, chemical engineers, and biological people. Also students from the LMU, the Lunwig Maximilian University of fine arts. There were about 30, 40 people, and these gatherings were quite nice. We had coffee, and were invited by the wife of the professor, and it was at this time a very interesting get-together. We exchanged ideas. Each time there was a talk about something, about literature or physics or biology and so on. And I was asked to talk about computers. I recall still that the interest of my colleagues was almost nil. [Chuckling] They thought, what a crazy thing! I guess this impression continued very much through my professional life until maybe very lately, when the personal computer was invented and is propagating now throughout the world. But for some 15 years you could see that many people completely underestimated the power of the computer in the Information Age. Maybe not so much in your country, but in Europe.

Aspray:

Well, I expect so there, too.

Proebster:

Yes. Well, let me tell you some further small stories if you are interested on the engineering side.

Aspray:

Please.

Difficulty of Obtaining Digital Components

Proebster:

We were confronted with extreme difficulties to put real good digital circuits together because we almost couldn't get digital components. The German industry really concentrated after the war on entertainment and on conventional telecommunications. So the nonlinear devices were by and large not to be obtained. We had to compromise by using conventional devices like radio amplifier tubes, which were really built for amplifying audio signals and were designed to work as much as possible in a linear way. We were interested to have devices that work as much as possible in a nonlinear way. It was not easy to find those devices. Also, we couldn't get much sympathy from the industry, even when Professor Hans Piloty realized the extreme importance of this new technology. He was an adviser to Telefunken. I give you one example. I approached Professor Hans Piloty and said that we needed a more nonlinear radio tube, and this would quite easily be accomplished by changing the characteristics of a radio tube by changing the pitch of the control grid inside the radio tube just a little bit. This should have been quite easy for Telefunken, which was a tube manufacturer at this time, to change at least for one or two days the pitch of one of their machines. We would then get tubes that would allow us to build much better tube components — and/or's and flip-flops and so on. Professor Piloty was convinced that this argument was right, and I recall we made a visit to the headquarters of Telefunken. I think Hans Leilich was also with us. There were three or four of us, and three or four of the chief engineers and other managers of Telefunken. They listened to us politely and said, "That's not a market for us, and we will not do it." Kind of looking back, it was one of the biggest mistakes — not with respect to the tube, but with respect to the realization of the possible change in strategy — of Telefunken at this early time.

To give you another example, from a later period, on how little comprehension existed in the industry at this time, when they finished the computer in 1955 — the end of 1955 — the industry got more and more interested in this. Professor Piloty went to Siemens here, and said, "We understand that you are interested in computers. We have now here a working machine. We have now here at least a handful of people, who have had intensive training for several years, who are experts. Why don't you capitalize on this machine? Why don't you capitalize on these people in order to get a head start in this new field?" They considered it, and considered it, and at the end they said, "No, no. This is university-type work. We would like to do it our way completely." I now know much better because I was in IBM (I've got to tell you a little bit about my IBM career): If you start out with a team and you take the best people, and you take the best attention by management on the field, you need five years at least before you come up with the first useful, worthwhile product. Maybe ten years. So by this wrong decision, Siemens wasted at least three or four years. Telefunken did similarly. Telefunken got into the computer business and electronic exchanges as you know, and they had tremendous difficulties.

Solving Drum Problems=

Proebster:

Let me talk about one story on the drum, and then I will go over to other areas. The drum problem was to have a very close distance between the read/write heads and the surface, as you know, and you get the high density. The floating heads were not invented at this time, and we had rigid heads with respect to the surface, and the calibration was extremely difficult. So we had differential screws. Do you know what a differential screw is?

Aspray:

No, I don't.

Proebster:

You have two screws — an inner screw and an outer screw. I don't know the special mechanical words in English. They have a slightly different pitch. So if you turn one turn, then the slightly different pitch gets translated to the forward or backward motion. You don't have only very fine pitches, but instead you have two pitches which differ very, very little. When you turn them, the forward or backward motion is very minute. In a way that was invented here. I'm not saying it was invented here, but they were applied here.

At the end, it was a parallel machine. We tried to build a machine which was as fast as possible. From the beginning we had the idea to build the fastest machine in the world. This meant that we were going very much parallel and very fast with high density. The distance between the read/write heads and the surface was decreased and decreased and decreased. In the end it was necessary to temperature-stabilize the whole thing. The temperature stabilization was done by putting a big hood over the whole thing and watching with a thermostat. One day we forgot the turn on the heat control, and the drum started up and got too hot. There was a sound like a whoo-whoo-whooo-whooo. When we opened the hood, the heads had touched the surface, and there were head crashes over the whole drum. All the heads had touched, and the whole drum was damaged. These were not heads that were fabricated with automatic machinery, but which were wound by our students by hand under the microscope. So it was weeks and weeks before this was repaired and it was in operation again. It was a major disaster.

IAS Machines and the Parallel Adder

Proebster:

Now, the electronic side. I told you that the machine got bigger and bigger, and at the beginning we had to have an experimental metric unit working, one box of eight register places. We were looking very much into the design of the Institute of Advanced Studies machine at Princeton. We built experimental circuits that were very much influenced by the Institute of Advanced Studies. The Institute of Advanced Studies machine works with a relatively safe concept — a secure concept — that you have always an empty register and a full register. Between the full register and the empty register, you have then logic. You perform the logic and transfer the results....

Aspray:

From one to the other.

Proebster:

To the other one. If you do so, in a way you waste a lot of time. We thought about how could we do better. Could we do it in a way that we have two full registers? The first full register works on the set of logic circuits, and the second full register works on the second set of logic. But by the time we evaluated the logic of the first register and transferred it into the second register, there is so little delay by the circuits that we still can't do it. We helped this concept by adding capacitors which stored the results for a certain time. We designed the RC constants so that it was just possible. In a way this was my invention. By this we saved by and large one half of the tubes. So instead of 4,000 tubes, we had only 2,000. Just roughly. We increased the speed of the machine by a factor of 2. But Professor Hans Piloty was very unhappy because the first experiments were not without flaws. From time to time errors occurred. He told me, "I cannot tolerate any error. One more, and we go back to a more conservative scheme, and you cannot have your way." This was one part of our deviation from the Institute of Advanced Study machine. I carried this idea to the extreme with respect to designing a parallel adder of very high speed.

We implemented in this machine an adder where the carry is done in the same way. The worst-case carry can occur if you have just in the second register only ones and you add a single one. It has to propagate in our case through 40 stages. We did it in the way that everything was prepared for the addition. When this carry came, we had a pulse which found the last stage ready to propagate the pulse. It received the pulse and closed the gate. But by the time the pulse has already passed the last stage and came to the next stage, it was ready to let the pulse pass and then close up after the pulse passed. The danger was that the pulse would be widened too much by these 40 amplifiers, or it would be shortened too much by the closing of these registers. I worked for about two weeks on that, and it didn't work. Professor Hans Piloty told me, "You have two more days, and then we rip the whole thing apart." [Chuckling] We had already at this stage 40 registers going. My concept worked on eight places. But I had difficulties with 40. I worked one full night, and the next morning it worked. That's the machine that worked reliably for about 11 years. Professor Bauer may have told you that it was the workhorse for the Technical University for quite some time, about 11 years. That was one story.

Power Fluctuations

Proebster:

Another story which you may not have heard is that our power supply was not very reliable. After the war the power utility companies just barely could provide the electrical power needed by the country. As a consequence, we didn't have stable frequency or stable voltages provided by the power companies. There were many, many disruptions, power outages, during the day. It was impossible to test the computer, the very sensitive computer that we designed as I told you. The circuits already could work very, very fast, and we had provided relatively little margins with respect to voltage variations. So the power companies really caused a lot of problems. Now, what to do? Today you would go to a company and buy power-regulating machines. But there was nothing like that on the market. One of our colleagues at this time who was a student got charged with the diploma work on this topic, Mr. Kösar, who has helped Professor Bauer to build up his exhibition on "Informatik" at the Deutsches Museum. You can also interview him. He is in Munich as well. There were no motors and no generators at this time, so he went out to the pool of the Munich Tramways, where they dismantled demolished tramways which were damaged during the war. He got eight motors from this pool, brought them here, and constructed four motor-generator sets. The motors were driven from the power provided by the utilities, and they operated the generators, and in between we had regulators. So when we had power outages and the phases were stopping out and the wattage surged, then we could have control and have a pretty stable d.c. voltage provided to the machine. Problem was, again, the space to have room for these huge machines, which were also noisy.

What eventually was done was to close a ladies' toilet and bring the whole machine into the ladies' toilet. At this time the women's lib was not as strong as it is today so we didn't have any protest. Another problem was that the d.c. voltage generated by these generators was not smooth enough. We had to filter it with condensers. Reliable condensers were not to be obtained at this time. We had condensers from army surplus. A very large volume was filled up with these condensers, and they were not good. One day they exploded. The full ladies' toilet was filled up with the material of these exploded capacitors — problems which you wouldn't experience today.

Semiconductor Diodes

Proebster:

Maybe one or two stories more. Soon after we started work on the basic tube components, we realized that in contrast to earlier tube work, we could rely on new components, the semiconductor diodes. But they were in their infancy. We had long discussions with Telefunken about our specifications and what we demanded for our machines. We were quite satisfied with our cooperation with Telefunken, and we could save a lot of tubes by incorporating quite powerful semiconductor diodes. One day our effort experienced something which we had never seen, namely, that a circuit, a flip-flop, didn't work. We tested it — Hans Leidich was with me — and said, "It must work." So Hans Leidich said, "Go away. I will do the test." And the failure didn't occur. Then I said, "Well, let me try again." So we changed seats, and the failure occurred, and then it didn't occur. We were looking and looking for days. At the end we found out what was at this time was not even found out by the semiconductor specialists, that there was a charge-storage effect in the semiconductor which plagued us. The charge-storage effect was not in the literature; it was just in the front end in the semiconductor case. So we are really in front of the research at this time.

One final example. A flip-flop didn't work but each time we tested it, it worked. If we stopped the testing, it didn't work. Eventually we found that there was a high-frequency oscillation which together with the diodes produced a d.c. bias which distracted this element. Basically, what we did and what other workers didn't experience, we increased the speed of our circuits to the extent that our tube circuits eventually became microwave generators. We had to damp the microwave oscillations in order to get reliable operation. That has not been published. Maybe in my Ph.D. thesis I published it. I put down in my Ph.D. thesis on the design of the central-processing unit of the computer. That is my Ph.D. thesis.

Aspray:

I see.

Proebster:

Subsequently I published also an article on the high-speed adder in several magazines, and I got the NTG (Nadin'chten-technische Geselleschaft) paper Award for that parallel adder.

Career at IBM

Proebster:

Now, on my further professional life. When they put our machine in operation, for several weeks it was the fastest machine in the world. [Chuckling] But only for several weeks. Then IBM came along and had a faster machine. All of my Ph.D. work was done, and I was looking for industry work. After some consideration whether I should join Telefunken or IBM, I joined the IBM Research Laboratory in Zurich, and I first did work on components. Then I worked on thin magnetic films, which eventually got the thin magnetic film memory into high-speed IBM computers. They were the fastest memories for about ten years before semiconductors took them over, with 60 nanosecond access time at 120 nanosecond cycles. We explored nanosecond switching at the Zurich laboratory. From 1962 to 1964 I became director at the research laboratory in Yorktown as Director of Experimental Machines. Then I was nominated as head and later as director of the IBM Development Laboratories of Germany in Böblingen. For about ten years I was head of that. I then devoted my life more toward university work. I am retired now from IBM for four years. I am now Extraordinairus Professor at the Technical University teaching computer science.

Aspray:

Here in Munich?

Proebster:

Yes. I'm teaching computer peripherals, computer CPU technology, and computer networks. I enjoy that.

Aspray:

Do you have a little more time?

Proebster:

Sure.

Aspray:

What were the factors in your decision to join IBM rather than Telefunken?

Proebster:

It was a very difficult decision. I was interested in responsible work, but I was also interested to widen my horizon. I got a very good offer from Telefunken to head a team. They were not quite sure what to do, either to go into the computer field or, more likely, to go into electronic telephone switching. On the other hand, I was applying for a job at the IBM Research Laboratory in Zurich. The standard of living at this time, 1956, was still enormous in Switzerland. Switzerland was decades ahead of the standard of living of Germany. And I saw that I was surrounded with very capable colleagues: Professor Speiser had very attractive offers, not only with respect to salaries, but also with respect to working conditions. He attracted the cream of the cream of the computer experts in Europe at this time. It was a very marginal decision because I saw on one hand, if I go to Telefunken, I would have more managerial responsibility and maybe more freedom to implement my own ideas. On the other hand, I saw in joining IBM that I would be surrounded by international colleagues, being close to the development and research work of IBM, which I had seen in the literature, and which intrigued me. At the end it was decided by an evening get-together. My wife was with me at the Lake of Zurich. It was a very, very friendly atmosphere, and I could see that I would feel at home and could profit quite a bit. I felt that I was getting more with IBM than I would be staying with Telefunken. I wonder what would have happened with Telefunken if I would have stayed with them, if it would have gone down as much. And what would have happened with IBM if I would not have joined them. You cannot play these experiments. [Laughter]

IBM Zurich Laboratories

Aspray:

Can you tell me what the status of the IBM Zurich labs was? I don't really know the early history of those labs. How did they get founded? What were their status when you joined them?

Proebster:

It's now in the IBM history, but I'm sure it's published somewhere. But everyone has a different touch to this story. I think the basic principle was that after the late — the first — Thomas Watson gave powers to his two sons, Thomas was concentrating on the U.S. and Arthur was concentrating on the worldwide operation. There was rivalry between the two brothers from the very beginning. When the father died — and this was about 1956 the mother had a lot at her hand to bring peace to these two brothers. Arthur was trying to challenge his brother with building up IBM worldwide in the same way with respect to IBM U.S. IBM headquarters in Paris were very prestigious. Getting a product out for the worldwide market designed by a team of Arthur Watson was another of his ambitions. And having a research laboratory under worldwide auspices was certainly very much also Arthur's idea. At this time IBM made a lot of money — they were made extremely rich — by the first big very powerful computers. The profit margins at this time for IBM and also for the salesmen were just unbelievable, particularly when you look at the profit margins of today. I know that salesmen got so prosperous in very little time that they could afford things which we can only dream about. So money was available. The company was expanding. One was looking also from the other side, from the technical and scientific side, to attract new talents in order to get more push on the frontier to the IBM products. A little bit later it was the two brothers who hired Manny Piore. You know him?

Aspray:

Yes.

Proebster:

He had tremendous vision to separate the research efforts completely from the development efforts. The research efforts were in the beginning when I joined very much linked to the development efforts. I recall that Jerry Haddad, one of my later peers and heads, came in the first month of the existence of the Zurich Laboratory on a trip to Zurich and explained to us what we possibly could do. These were ideas which were in a way what you would term today "advanced technology," extending the art which the development engineers are just experiencing. Manny Piore had a different approach. He said, "Reach out as far as possible. Be superior in all ways. Do not stick to the ideas of today." It was about one or one-and-a-half years later that Manny Piore visited us. As I still recall it, we had three satellites sent out before he came, very famous people in history, and they interviewed us and explored the terrain and had reports ready. So when Manny Piore came he could concentrate on the essential problems. John Gibson was one of the satellites. He was the founder of the semiconductor empire of IBM. And Gardiner Tucker was the second. Later he became very famous as a research director, and was one of the leading people of NATO. The third man was Clayton Andrews. He was working in communications. So there was a big change, and the Zurich Laboratory was changed. By that time it had already increased in size. The more development-oriented fellows, colleagues of mine, some left, but the majority of them joined the Development Laboratories in San Jose or in the Poughkeepsie area. So, for example, Hans Schläppi joined the Research Laboratory in Yorktown. Fritz Wiedmar is in San Jose. Kai Kinberg is in San Jose. And others have left.

Logic Circuits & Ferromagnetic Memory

Aspray:

Do you want to talk in more detail about some of the projects you worked on while you were at Zurich?

Proebster:

In Zurich?

Aspray:

Yes.

Proebster:

In the beginning we had only a small group of about six of us. We had weekly meetings, and we were asked how could we create new ideas, and what would we really like to work on. Glätti, who left IBM fairly soon who was a very capable Swiss scientist and engineer — more of an engineer than a scientist — started to work in hydraulics, hydraulic logic, and hydraulic amplifiers. He invented a garage opener, at this time a completely new idea, and then went on and started his own consulting and engineering business. I was interested in new logic circuits. This was a time where deviation from tubes was visible, and we were thinking about solid-state circuits. An idea was visible in some of the magazines to work with ferroresonant circuits, which are bistable circuits, depending on the characteristic of the magnetic material. The magnetic material shows hysteresis behavior, high permeability with small signal levels, and small permeability at high signal levels. You can design circuits that have then bistable behavior, which have either small currents associated with high permeability or have high currents associated with low permeability. I was working on flip-flops, bistable circuits, and logic circuits, and published my work at one of the first European computer conferences in Madrid.

I was involved in a very funny experience in my IBM life. I was then interested in miniaturizing devices. Ferroresonant circuits are used also in the power business to stabilize generators. It's used today for stabilization. But they are large devices, and we were interested in very small ones. At the Zurich Laboratory at this time they had very little experimental facilities for non-electronic devices. So we were looking around to find out who could provide us with materials that could lead to miniaturization. Now, who was around? Phillips and, at this time, Steatit Magnesia. We had offices in Zurich where we could talk. I was invited to Steatit Magnesia. After a year or two of not leaving my desk, to be invited on a trip to another city was just fabulous. We were hosted very nicely and treated as kings coming from IBM. What do you offer? What kind of ideas do you have? We didn't find the right material. Our second trip was then to Philips, to Eindhoven. The same surprise. Unbelievable. Hans Schläppi and I received the red carpet two of us, two IBMers came like that to Philips. What they didn't know is that Hans Schläppi could speak Dutch. [Laughter] So they had some inside discussions in Dutch which we were not supposed to understand. We were interested in the following idea-this was something in which both Schläppi and I were interested — that the main memory was ferromagnetic. Most of the current memories were going in the direction of smaller and smaller core size, faster and faster speed and cheaper and cheaper in production. The handicap was the winding problem, to bring two or three or four wires through these tiny cores.

We were thinking about new concepts: How about having not a closed core but two pieces of doughnut, having the wire in the first place and then putting the doughnuts in place. In Zurich we didn't have any possibility to experiment on that, so we went to Eindhoven and explained this idea to the Philips people. They were quite intrigued, and said, "We'll try it." And as good IBM citizens, and being admonished many times by a very strict Ambrose Speiser, our boss, we wrote an extensive trip report. In order to do as good IBM citizens do, we disseminated this trip report widely. The effect was fantastic. Two vice presidents came by plane to Zurich and explained [to] us how to behave. [Laughter] Do not disseminate confidential information unless an agreement was signed, and unless all patent problems had been solved, and so on. From then on we behaved properly. [Chuckling] But in retrospect these ideas didn't prove to be practical because then what we did not know from our university studies, that you create damaging zones in the magnetic material. So even when you precision polish the surfaces, when you put these two half doughnuts together, you don't get the hysteresis loop which you want to get. So, it's for the birds. But at the same time we got the idea of how to work on this material in different ways, with high-precision machining. For my ferroresonant circuits I could design circuits that were very, very small, for my machine. So I could put the active magnetic area in a very minute magnetic area. This again put us then to the idea of how to extend the miniaturization further, and we came to magnetic films. There was the Balzer Company in Liechtenstein, close to Zurich, and again, Hans Schläppi and myself, we were visiting Balzer. We were amazed about the open perspective of films as such. It was a new world that was opened to us.

Soon after that Dr. Piore came again, heard our story, and said, "No way. You are not qualified to do this type of work." We were a little bit put off by his remark, but certainly he was right. He said, "In order to continue this type of work, you need experimental physicists and theoretical physicists to join you." Ambrose Speiser was asked to add these people to his staff. I was asked to look for qualified theoretical and experimental physicists in Europe. So I looked up the books of the universities, and Professor Speiser and I made several European tours to universities to find qualified experimenters. We found at the Bergakademie Clausthal Drs. Methfessel and Thomas, who at this time were just assistants, and they joined the Zurich Laboratory. Professor Methfessel is one of the most gifted experimental magnetic experts. He has just resigned from the University of Bochum and has a large university chair there. Professor Thomas, after serving many years with IBM in research, and in a way also fathered some of these early ideas on the Nobel Prizes of the IBM Zurich Laboratory, joined the University of Basel working on compound materials in semiconductors and magnetics. So they overlap in the area of magnetics and semiconductors.

Relations between Zurich and US

Aspray:

What kind of communication was there between the Zurich lab and IBM senior administration about what kinds of areas you should be researching? How did the researchers at Zurich identify the problem areas?

Proebster:

That's a very interesting question. Nobody has asked me that. I have to describe a little bit the personality of Ambrose Speiser. He is a Swiss-German, very correct, very correct. He is gifted in explaining complex facts very precisely. He is very cautious. By the way, he has also written one of the early computer books in German. I recall that he was very cautious in each and every one of his decisions. He has asked his U.S. peers about the possibilities and what he should do. I recall that in many instances, he got the word back: Decide yourself. Do it yourself. Do what you think is best. That was, by in large, the way we were guided at the time that I was in Zurich. That's the six years. We were, ourselves, certainly very much interested to learn what our colleagues in the U.S. were doing. We were quite interested to get in touch with a lot of people in order to sell our ideas. So we were interested in visits to the U.S., which were, at this time, unbelievably expensive. They were only reserved to the deluxe class of people. IBM regulations were at this time quite different from today, [Chuckling] precisely now when money is very scarce. I recall that I was given free permission without asking to fly to the U.S. in a berth. Do you know what that is? In an airplane they had a limited amount of beds.

Aspray:

I see.

Proebster:

So at night they could put down, and you could sleep there.

Aspray:

I see.

Proebster:

On top of the already very high fare, another very high price was added. I was in a way embarrassed because there were ladies and old people in the plane who did not have a berth. As a youngster [Chuckling] I had the possibility to sleep in a berth. I couldn't sleep very much because it was my first trip to the United States in very rough weather over the Atlantic. These were not high-flying planes; they were low-flying planes. You got to New York half sick. The money question was at this time under consideration, and our friends from the States were asking many times. Professor Speiser was concerned about his budget, whether these trips were really necessary. More often, our colleagues in the States were the ones who questioned the value of our trips to the States, which we had to defend. I recall that Professor Speiser said, "I have to go to at least once in a year to the States in order to keep up with the progress and the way the projects go in order to run my laboratory in an efficient way." We were given the permission to go at least every second year to the States in order to make friends, disseminate our information, and to get more information. This was the most efficient way to coordinate the work between Zurich and the States.

Work with Thin Films

Aspray:

Were there researchers from the American labs coming to Zurich?

Proebster:

Oh, yes. The continuation of my work was when I saw that the magnetic film work was so interesting that I got on the thin film work together with my colleagues. I was heading the engineering group, and Mr. Methfessel was heading the physics group.

When we got interested in thin film work, our emphasis concentrated on thin magnetic films rather than on ferrites, because we saw that the ferrites intrinsically had the lower speeds than the thin magnetic films. First, we thought just because of the reduction in eddy currents. First we thought of the larger magnetization as compared to the ferrites. I had worked here in Munich on thin magnetic material already for the pulsed transformers. I was working on very efficient pulse transformers, together with Vakuum Schmelze Hanau, which is a subsidiary of Siemens near Frankfurt, and had contacted their magnetics experts already during my work at the Technical University in Munich. So I knew at this time that the thinnest magnetic sheet you can produce by mechanical means is somewhere in the 3-micron area. But it is very expensive, very delicate, and not really good for mass production. I was intrigued by my visit with Balzer to learn that you can produce much thinner magnetic layers at low cost. We were thinking first to expand this to magnetic circuits, and I was considering in particular magnetic logic circuits. Certainly we were also interested in magnetic storage.

This was the time when our colleagues from the physics side joined us. It was a cooperation between several colleagues at the Zurich Laboratory, where we really got to the idea to investigate and capitalize on the high-speed behavior of the thin films. Hans Schläppi, whose name I have mentioned several times, had the idea of how to develop a high-density store based on the new material. In his investigations he was interested in the high-speed performance of a new material. In order to investigate this, the oscilloscopes at this time were not fast enough. He launched an effort in the Zurich Laboratories together with Dr. Louis now at IBM in Stuttgart. Dr. Louis, together with Hans Schläppi, designed a laboratory-type sampling scope with which you could see change phenomena in the sub-nanosecond area by very sophisticated high-speed components. It was a large machine, but it worked. A team of about three or four of us we proved for the first time experimentally the fast rotation of the magnetization. The thin film that is produced has the magnetization either in one direction or the magnetization in the other direction. You cannot put the magnetization perpendicular to the plane without an extra magnetic field that flops back. You can switch the magnetization from one side to another in the time frame of a nanosecond. A nanosecond was not invented at this time. We said "millimicrosecond" at this time.

After this experiment was done, we got articles in Electronics. It was fascinating, the interest we received in this experiment. Because we showed not only that it switched in this direction by just the usual way, but we also observed the magnetisation changes perpendicular to the preferred direction, and we saw these wave forms. One of my colleagues, Dr. Wolf, put it further, that he saw even oscillations of the magnetic vector. The magnetic vector is not fully damped during switching, but it oscillates before it settles. We could also observe these experimentally. It was a very interesting experiment which brought us to the idea of investigating high-speed logic elements and high-speed magnetic elements, which at this time were about a factor of a thousand faster than the ferrites. The ferrites switch by wall motion somewhere in the vicinity of about a microsecond and the thin films switch in a nanosecond. These were fascinating times. I was asked to give a report to an IBM management committee in the States. We built a small-size experimental memory in Zurich. This project was then transferred to the States because of its commercial interest, first to Yorktown, and subsequently to the development laboratories with a very colorful history. But Emerson Pugh — you know Dr. Pugh?

Aspray:

Very well. Very close personal friend.

Proebster:

Dr. Emerson Pugh can tell this story much better than I can because he was manager then at this time until the final stage, when it was put into production and put into the machine. I can only tell you the very beginning. It was at this time where I got acquainted also with Emerson Pugh and his colleagues, with Lee Shevel. You know Lee Shevel?

Aspray:

No.

Proebster:

That would be another man.... A man with very pronounced ideas. I hope he is still alive. He was particularly engaged in ferrite memories, but also in later developments, one of the key memory experts at IBM.