Oral-History:James Meindl

About James Meindl

Dr. James D. Meindl, IEEE Life Fellow and recipient of the 2006 IEEE Medal of Honor for "For pioneering contributions to microelectronics, including low power, biomedical, physical limits and on-chip interconnect networks", was Director and Pettit Chair Professor of the Joseph M. Pettit Microelectronic Research Center at Georgia Institute of Technology in Atlanta developed micropower integrated circuits for portable military equipment at the U.S. Army Electronics Laboratory in Fort Monmouth, New Jersey. He then joined Stanford University in Palo Alto, California, where he developed low-power integrated circuits and sensors for a portable electronic reading aid for the blind, miniature wireless radio telemetry systems for biomedical research, and non-invasive ultrasonic imaging and blood-flow measurement systems. Dr. Meindl was the founding director of the Integrated Circuits Laboratory and a founding co-director of the Center for Integrated Systems at Stanford. The latter was a model for university and industry cooperative research in microelectronics.

From 1986 to 1993, Dr. Meindl was senior vice president for academic affairs and provost of Rensselaer Polytechnic Institute in Troy, New York. In this role he was responsible for all teaching and research. He joined Georgia Tech in 1993 as director of its Microelectronic Research Center. In 1998, he became the founding director of the Interconnect Focus Center, where he led a team of more than 60 faculty members from MIT, Stanford, Rensselaer, SUNY Albany, and Georgia Tech in a partnership with industry and government. His research at Georgia Tech includes exploring different solutions for solving interconnectivity problems that arise from trying to interconnect billions of transistors within a tiny chip.

In this interview, Dr. Meindl discusses his career through Carnegie, Signal Corps R&D labs, biomedical electronics, the establishment of fabrication and research in integrated circuits at Stanford, and the adoption of semiconductor research by the industry sector.

About the Interview

JAMES MEINDL: An interview Conducted by David Morton, IEEE History Center, 3 May 1996.

Interview #266 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 Oral History Program, IEEE History, 445 Hoes Lane, Piscataway, NJ 08854 USA or ieee-history@ieee.org. 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:

David Meindl, Electrical Engineer, an oral history conducted in 1996 by David Morton, IEEE History Center, Piscataway, NJ USA.

Interview

INTERVIEW: JAMES MEINDL

INTERVIEWER: DAVID MORTON

PLACE: ATLANTA, GEORGIA

DATE: MAY 3, 1996

Early life, education, Carnegie

Morton:

Dr. Meindl, I thought we might start just by talking a little about your early life — where and when you were born, how you got interested in technology and science and what your educational experience was.

Meindl:

Fine, David, it'll be a pleasure to do that. I was born in 1933 in Pittsburgh, Pennsylvania, and attended Carnegie Institute of Technology. The reason I selected electrical engineering is that my father worked for Westinghouse Electric Corporation. He was not a college graduate, but he always spoke highly of the engineers with whom he worked, so it seemed to me that that was a pretty interesting field to get into. I liked science and mathematics, so I had the right inclinations. It turns out that I rejected mechanical engineering among all the fields of engineering because of all the tests that I took. It seemed as though my three-dimensional spatial perception was the weakest part of my capabilities. I thought, "Well, I don't have to use that too much in electrical engineering." So instead of going into mechanical or civil or some other, I went into electrical.

What really intrigued me more than anything else about engineering, as I started to understand a little bit about it in my years of college, was that it was a means of predicting the future. That was what was really intriguing to me. If you took laws of physics and applied them intelligently and designed something—and if you were correct in the way you designed it and measured it—you could actually do the experiment to see how well it would agree with what you predicted, what your calculations said it should do. To me that was an exciting idea, predicting the future.

I'll just connect it with today. What I find the most satisfying is to look back at the history of integrated circuits and to try to understand the key features of the technology that have gotten us to where we are today, to understand the physical basis for them, and to extrapolate, using laws of physics, to understand how far we should expect to be able to go. Certainly, economic factors enter the picture in a major way, but you really hit a wall if you don't have the laws of physics on your side. So, even today, I find this very, very fascinating. It's where I focus my research interests—and my teaching interests, for that matter, since they are closely coupled to my research interests. So I can see a link between what first intrigued me about engineering and electrical engineering in particular.

But going back to Carnegie Tech — I was interested at that time in power engineering, specifically sixty-cycle electrical power engineering, the kind of engineering that brings us electric lights and motors and so forth. I would say that part of the reason for that is that my father worked for the large unit of Westinghouse Electric where they made power equipment: motors, generators, switch gear, and so forth. And my home town was East Pittsburgh, Pennsylvania, which was near the major Westinghouse manufacturing plant; and, of course, their headquarters is still in Pittsburgh.

I did reasonably well at Carnegie as an undergraduate. I had a few jobs in steel mills along the way through my undergraduate years at Carnegie, and did pretty well. I was encouraged by some faculty members, including the former chairman of our department at Carnegie Mellon, whose name was Everard Williams, to consider going into electrical engineering. He said, "Why don't you take our masters' program?" So I did that; and then, of course, he said "Well, why don't you take our doctoral qualifying exam?" And I did that, and they said, "Well, why don't you stay, and we'll offer you a research assistantship." So I stayed through a Ph.D. I never expected to do that when I started college; I never expected to do it until my senior year, really, when one thing led to another.

Morton:

What was the curriculum like there? Was there a lot of solid state electronics being taught, or were they still teaching vacuum tubes at that time?

Meindl:

We really didn't have any solid state in the kind of curriculum I went through. We didn't have any semiconductor physics, to put it that way, in the curriculum. I can remember a senior course that had some transistor physics in it, but it was kind of treating it as a black box whose characteristics you would measure from its terminals and then use those measured characteristics to do things, to design and analyze, as opposed to getting inside the semiconductor chip and understanding the physics of the device.

So I spent some years after I got out of Carnegie really devouring the books that I could get my hands on, published articles that talked about semiconductor physics. I had very little of it. My doctoral dissertation was an application of Maxwell's equations which, if anything, are the core of electrical engineering. Working with Maxwell's equations turned out to be a lifelong advantage. I obviously became familiar with how to solve particular problems, using a set of the most extraordinary expressions of natural behavior which you'll ever find. I mean, they're incredibly brilliant, they're incredibly useful, and so working with those was something that is still valuable to me today. Perhaps, really, because they're so general, they're so useful. As I went through graduate school, I worked for two summers for Westinghouse, and spent that time in a unit of their central research laboratories. And there I became convinced that I wanted to get into semiconductor circuits.

Semiconductors at Carnegie and Signal Corps R&D labs

Morton:

Were you working with semiconductors there?

Meindl:

Yes, I was. I had an opportunity to work a lot with them during both the summers I was there.

Morton:

What were they doing? What were you working on?

Meindl:

Actually, we were working on control systems. They wanted to design reliable industrial controls for large motors and generators. They wanted to get rid of vacuum tubes because the tubes weren't reliable, and they were burning out all the time. Westinghouse wanted to substitute solid-state controls, and therefore they were trying to push a lot of power through their semiconductors. That meant you were taking them to the edge of their capability and blowing out very expensive devices. I can remember paying seven and eight hundred dollars for transistors and burning them out in a few minutes. We sort of pressed the limits of what they could do.

I did that for several summers. Then one summer, I went to California to work for North American Aviation. During the summer I was there the government canceled a large missile contract for an intercontinental missile called the Navaho. North American Aviation had set up a recruiting tent on their own grounds; it was like a show, like an auto show or something like that. There were booths, and all the major contractors in the defense industry came. Their representatives there were hiring engineers from North American Aviation. I thought it was interesting because they offered me a job at the end of that summer; they wanted me to stay. I wasn't finished with graduate school, though, so I went back to Pittsburgh. But I learned that I'd like to live in California.

I finished and went to work for Westinghouse and got into the application — again, of power and semiconductors—of controlling various pieces of equipment. In fact, I worked on controlling the position of control rods in a nuclear reactor in a submarine, and I met Admiral Rickover. So that was interesting.

I had been in ROTC at Carnegie and strangely, after I was out and working at Westinghouse for a year, the Army said, "We want you for two years of active duty." This was between Korea and Vietnam, and there wasn't anything going on. Most people who had taken ROTC went in for three months or six months. Somehow I was designated a two-year person, which I thought was really a low blow. But what happened is that I was sent to Fort Monmouth, New Jersey for signal school, Signal Officers' School, and then they assigned me to the Signal Corps Research and Development Labs in Fort Monmouth. And that's where I ran across the integrated circuit.

At that time the Signal Corps was the major electronics research laboratory for the entire Army. It was as big as any one the Department of Defense had. They did a lot of internal research, but they mostly did a lot of contracting with industry to develop the techniques, because this was in 1959, right after Sputnik. The Cold War was really very, very much on the minds of everyone. So the Signal Corps was very active in developing electronics for the Army and actually had an astroelectronics division where they were designing the first satellite signal for the U.S.

When I arrived at the Signal Corps R&D labs in the autumn of 1959, they had just decided they were going to award a contract to Texas Instruments in Dallas, Texas for development of integrated circuits for use in portable field equipment which have to be small, light weight, low power, that sort of thing. So they said, "Well, here's this new guy with a Ph.D. We'll put him in charge of this contract." So in autumn of that year I flew down to Dallas and ran it, as I remember. There I met the co-inventor of the integrated circuit, Jack Kilby. He was the person who was the individual at Texas Instruments responsible for this work; that was a year after he invented the integrated circuit. And I've known him since that day in November of 1959. Shortly after that, because the Signal Corps was interested in all the important advances in semiconductor technology in the country, I visited Fairchild and met Bob Noyce and Gordon Moore when they were still with Fairchild. Bob Noyce, as you know, is the co-inventor of the integrated circuit along with Jack Kilby.

So I met the two inventors about a year after they conceived the integrated circuit and documented their original inventions, and I've been involved in integrated circuits ever since. It's been a tremendously interesting experience. I don't know that anyone in engineering could ask for a better career, because it's just been fantastic, as you know, in terms of its impact on society. Today, everybody knows what the chip is, and I've seen it go from being something nobody knew about it to all the way through to being commonplace. It's been very, very interesting, and I continue to be very fascinated by it. If you don't get the modern history of me, I'll say that before coming to Georgia Tech, I was seven years the provost, or chief academic officer I guess they'd call it, at Rensselaer Polytechnic Institute but I decided I did not want to end my career as an academic administrator. I wanted to end by following a little more closely what was happening with integrated circuits, so I took the opportunity to come here, and that's what I'm doing.

Fort Monmouth

Morton:

I'd like to get back to the work you did at Fort Monmouth and immediately after. I take it they were working on some sort of integration, obviously different than what Texas Instruments was doing. What sort of technology were they thinking about using at that time?

Meindl:

Well, first I'll say that I spent 1959 through 1967 at the Signal Corps R&D laboratories, which turned into the Army Electronics Laboratory during that period. I was in a group that was called the Electronic Components Laboratory. Our focus was on developing components that would be used in radios and computers and radar systems and helicopters and other remote sensors. anything that you could conceive of using on the battlefield that involved electronics. Because the lab was so much in the forefront in the development of integrated circuits during the two years I spent on active duty, I chose to accept an offer from them to stay, as a civilian. I didn't want to remain in the army as a career, but I decided the work so interesting that I stayed another six years as a civilian. We were working on developing families of integrated circuits for use in portable field equipment and in vehicular field equipment. And we developed the first family of very low-power integrated circuits for use in the helmet radio receiver for individual foot soldiers. We did that with Motorola, and I still know the people who I worked with at Motorola. Several of them are still with the company today, and in the last year I've seen one of them a couple of times or so.

Another thing we did was develop a family of digital integrated circuits for use in battlefield computers, the computers they used to direct artillery fire, for example. This family of integrated circuits was developed with a new startup in Silicon Valley that was a spin-off of Fairchild and was called Signetics. Subsequently Signetics was acquired by Philips. Philips has a unit in the Santa Clara Valley now that's still in the location that the Signetics originally occupied.

So as a result of being at the Signal Corps, which was the major Army unit doing research on integrated circuits, I got to know the industry very well and met a few key people in the industry. And for historical data collection, I think I could say that I believe that the investments that the Department of Defense made in semiconductor manufacturing technology, manufacturing methods (these were made by the Army and the Air Force and the Navy) were made with dozens of companies in the United States, to try to set up production capabilities so that in case of need, or in case of war, there would be an immediate availability of a production capacity for semiconductors that the Army would be able to utilize in fulfilling its needs. World War II wasn't that far behind at that time, and so there was an interest in being prepared. So I think the Department of Defense just did a great service for the country by investing in these manufacturing capabilities for semiconductors. Of course the companies immediately turned around and were able to use for commercial activity. So it really launched the U.S. to a world leadership position.

Morton:

In terms of research, when you were at Fort Monmouth, you were in a relationship with people in the industry. Did they see the work that you guys did as sort of competition? What sort of relationship did these commercial interests have with this government laboratory?

Meindl:

I don't think they ever saw it as a competition, because we would simply explore ideas and never take them into manufacturing or to the product stage. I think the philosophy was, and is, that if the Department of Defense is going to be an intelligent, informed buyer of high technology, it was necessary to have an internal capability; they needed some understanding of what it was. A lot of money was spent on that side.

I ended up at Fort Monmouth having a division of about eighty people reporting to me. It was very interesting because I had just got there at the precisely correct or advantageous time to grow with the field. I was promoted three times, from a section chief to a branch chief to division director. And all this was because integrated electronics was growing and no one knew about it. I was new enough to learn it quickly and didn't have other things to shackle me. So, companies didn't look at what we did as competition. I think they looked at it as something that was what we had to do, and we were spending three quarters of our budget or better at that time externally, so this wasn't a problem. I don't think it was. The difference between what we were spending and what we might have spent was, if you took away all of our internal capability and just left the salaries of the people, you wouldn't have made an appreciable difference, I mean, not in the money we were spending on the contracts.

Morton:

I'm interested in is the way knowledge gets transferred around. All these companies were sort of developing this technology, it sounds like, with outside funding. Were there also sources of outside knowledge? Was the government participating in transferring technology from the old laboratories to companies or even somehow indirectly from company to company? Was there any of that going on?

Meindl:

Oh, certainly. We were eager to share any knowledge that we generated, and one of the mechanisms that we were encouraged to use was publication in the literature. That was an important thing to do. In evaluating people for promotion or raises, you always look at what your external publications and external activities were. One of the ways that any knowledge we generated in the laboratories was put to use was to publish articles and people in the industry could talk about that with us freely. Now, when we awarded a contract to a company, we also had certain requirements on the information that had to be reported, which were pretty demanding, and so these reports were widely available, and we had a large list of groups that received copies of the reports. They generally included all the other contractors that we had. So there was certainly a transfer of technology.

The companies, of course, guarded against giving away too much. But on the other hand, they were obligated to provide information because of the contract that they had signed, and payments they received; therefore there was transfer of technology, and certainly a good deal of it. The one thing that I believe was probably the most critical single act — and this is a matter of personal perspective, I think — that made integrated circuits into the reality within the Department of Defense and the commercial products was the decision of the Air Force to put integrated circuits into the Minuteman missile.

It was about 1965 or so. This was an ICBM, and it was by everyone's measure, I think, the most important strategic weapons system the United States was developing. And a decision was made that integrated circuits were to be used in the missile. The opposition within the DOD community said that they weren't reliable enough; they know that because they didn't know what was happening inside this chip with all these transistors packed together. But to me, to put such a key system in the arsenal of the United States — I thought that was a breakthrough. Thereafter, there could be no opponents — and there were probably opponents who were tied to vacuum-tube technology, or discrete-part technology, who said, let's not be too rash about converting to this new technology. But the decision on the Minuteman missile system was, I think, pivotal. And once the Department of Defense started using integrated circuits very, very freely, of course, there was that much more confidence built up in them, in what they could do in commercial applications.

"Micropower Circuits" book

Morton:

That's interesting. I had a look at your book, Micropower Circuits from the 1960s. A lot of historians write about what's important about microelectronics is the drive to miniaturization. And a lot of people at the time who were working on this were talking about advantages and so forth and problems of miniaturization. But your book talks about power. You were doing something that seems different and has proven to be an extremely important part of all of this. It doesn't seem like something that many people were talking about at the time. Is that a reasonable observation?

Meindl:

I think it's quite accurate. Quite accurate.

Morton:

How was that book received at the time?

Meindl:

It wasn't widely used, because the interest in low power was a small fraction, really, a niche in the whole semiconductor picture. There were interesting applications — hearing aids, for example, as you can appreciate. That was probably the main commercial application at the time, for very low-power circuits. But the Army really wanted to use electronics for foot soldiers, to gain every advantage they could. So batteries and battery drain were a huge issue for portable field equipment. Actually we focused on this in our internal research at the Signal Corps because we realized that this was not the mainstream of the development at that time of integrated circuits. Companies weren't going to be paying much attention to low power issues. But the Army had an extraordinary interest in low-power electronics because it was the only service that was going to have hundreds of thousands of units of something used — portable units used by individual people. It was just a very interesting area to explore, from an engineering perspective, and was quite relevant to the needs of the Army at that time. So that took us into low-power applications.

The other place where low-power applications were interesting was in satellites. NASA had been formed in that period of time, and there were some people in NASA who were very keenly interested in low-power as well. There were also some medical applications; there were some consumer applications, like in a portable radio, but the batteries could be pretty large there. The hearing aid and different pieces of electronics for the foot soldier was where the application for low-power electronics was extremely important. It's quite ironic that today, as the Army looks ahead — and I'm still in contact with people from the Army Electronics Laboratories, in fact, I visited there two months ago — I don't visit often, but I just did two months ago — to the early years and decades of the twenty-first century, they see more and more electronics that they want to put in a whole multi-media system in the breast pocket of a soldier. So low-power electronics are absolutely essential. They need to go far, far beyond what the base capabilities are; and, of course, the projection for the semiconductor industry today as a whole is that portable boxes are going to be the dominant form of electronics. Certainly most of the market and most of the components built are going to be used in portable and battery-operated, pocket-carried electronic devices. Low power is going to become a big, big issue, there.

And another reason it's an issue is that packing density now is so high that heating effects are a major design issue in high performance systems, in desktop boxes. So getting the circuits to operate at lower power levels while still maintaining high performance is extremely important. It's become the mainstream now.

Morton:

It looks as if the book was mainly about circuit design rather than hardware or about developing new hardware. Was there any thought to that, or is it a naïve question to be thinking about that?

Meindl:

No, it's a very good question. There were a couple of reasons for that. Number one, there had been, prior to this book, no systematic exploration of just how you would go about minimizing power drain in a circuit without compromising the required performance. There were circuits around that were using ten, twenty, fifty times more power than they needed to use in order to perform the functions that they should perform. So this was an effort to take the major circuit functions and the major types of circuits that were in use, and to systematically ask, "OK, if I want this kind of performance from a circuit, how can I achieve it with minimum energy consumption?" And that included what transistor characteristics—which were the heart of it—what transistor characteristics do I need in order to get this performance at the lowest possible power? So it does not deal with particular applications. That's a correct observation. It deals with, you might say, the functional blocks that would be used in all of these applications. To a certain degree, that's because of the place that I was occupying in the Signal Corps research labs. It was a components group that looked at circuits, but in a kind of generic way, because we wanted to look at them in terms of a whole spectrum of applications and computers and telecommunications and radar systems and surveillance systems and so forth. And so we didn't specialize too much in any particular system.

Transistors

Morton:

I know that when the transistors came in, in a very general way, there was a lot of translation of certain generic types of circuits from vacuum to transistor. Do you think that maybe some of these circuits that were drawing more power than they had to might have been holdovers in some sense from some kind of tradition? Was there some reason why there were these circuits out there, commonly used, that were using more power than they absolutely had to?

Meindl:

I think that there was a reason. That's an interesting question. I think that the main reason was that the power requirements of a transistor were so much less than those of the vacuum that there were orders of magnitude reductions in power just by switching from a tube to a semiconductor. And this was not done in a blanket fashion; it was done on a piecemeal basis. Semiconductors at first couldn't work at high frequencies; and then they couldn't work at very high frequencies; then they couldn't work at ultra high frequencies; and so forth. So there was a march of acquisition where semiconductor technology kept conquering more and more of the applications of vacuum tubes, and as this was happening, you reduced the power by a couple of orders of magnitude, just by virtue of using a transistor. So that was a big step ahead, and people were more focused on trying to get a level of performance along with this reduced power, a level of performance from the transistor that was comparable with what they had been doing with vacuum tubes. They were making big gains, and they were having challenging requirements because a lot of features of the semiconductor terminal characteristics were different than the vacuum tubes, in terms of electronic behavior. So there wasn't a lot of focus on very low power circuits.

And again, it seemed to me that the question was, well, how low in power can we go? And today, that's what I think about all the time. What are the limits? How far down in power can we go? How much energy do you need to execute a binary switching operation that can run a full computing operation? And so the question that seemed intriguing to me at that time was to explore the limits on low power. And really, the limits depended on what I want to do. And so with each of those chapters in this book, a widely used generic circuit was examined, and the question was, "Well, if I need this performance from this circuit, how can I design it? What's a rigorous way of approaching the design to achieve that performance but minimize the power?" And this was, I think, novel at that time.

Biomedical electronics

Morton:

I believe it was. I’d like to switch gears now. I don't know much about your work in the biomedical electronics field. How did you make this transition from military electronics to this new field?

Meindl:

From Fort Monmouth you really were able to talk with people in industry and universities and other government laboratories throughout the country. And it was very clear to me, from where I could see the picture, that the leading university in semiconductor electronics was Stanford University. Their work was just outstanding, and the people at that time from transistor electronics were very good. I got to know them by mingling with them at IEEE activities. Really, definitely, it was IEEE activities: serving on program committees for this International Solid-State Circuits Conference, and the International Electron Devices meeting. During the time I was at Fort Monmouth, I became the founding editor of IEEE Transactions on Solid-State Circuits. So I met a lot of industry and academic people, and Stanford stood out. Their faculty members had the idea that there were some very interesting new opportunities to use semiconductor technology and medical applications. It really was a germ of an idea that they had. And so it seemed as though that was virgin territory at the time. We knew that they were important in hearing aids, but we felt that there were a lot of other opportunities. So I went to Stanford as an untenured associate professor with the thought of getting into the applications of integrated circuits in new medical instruments and for use in medical research and other areas. That turned out to take the next twenty years of my life, and it was an extremely rewarding period.

The first project I started to work on at Stanford was an extremely interesting one. It was to develop a set of integrated circuits to be used in a portable reading aid for blind people. And it's even more of a human interest story to understand the background. One of the two people who recruited me to go to Stanford, was the chairman of the department at Stanford at that time, his name was John Linville. He had been at Bell Labs and had been involved in transistor circuit work. Then he was recruited to go to Stanford by Frederick Terman, who is described as the father of Silicon Valley, as you know. So, John Linville had a daughter, whose name was Candice, and his daughter, shortly after birth, was affected with retinoblastoma with damage to her cornea, and she was blind, totally blind. So he wanted some way to allow her to read printed material and to do personal notes and so forth, the way a sighted person does. God bless him, he's now seventy-seven years old and living in Porto Valley in California. And so, he said, "Well, come on and help us do this reading aid for the blind." So my first doctoral student at Stanford developed this chip. And that was the first artificial retina, a silicon retina, for a reading aid for the blind. It's a little image sensor, and each one of those little cells you see there is a kind of a pixel, and it picks up a bit of information. And so this chip was in a tiny camera about the size of a tube of lipstick, and a blind person scans this camera across a line of print. The image sensor picks up one letter at a time, and then, through a little cable connecting it to a box, translates this optical image to a tactile image. Meaning that, if you think of a bed of nails, that people use sometimes for holding flowers in place in a vase, the bed of nails is really a stimulator array, and each one of these stimulators, these electric stimulators, is individually controlled. So if the camera is looking at an "O", you'd feel a circle of vibrating pins. If the camera is looking at an "I", you'd feel a single column of vibrating pins with a separate dot above it. So as you scan across the line of print, you feel the characters passing under your finger on the other hand. This was the principle of the reading aid called the Opticon. And so my first Ph.D. student produced the first retina, and this turned out to be the key element to the success. And I can remember Candy Linville being our most important member of the research team, because as soon as we would build something, she would come in the next day and test it.

Morton:

That's interesting. So was it more difficult for people with sight to use it?

Meindl:

Oh, yes. The tactile sense of a sighted is really undeveloped compared to that of a blind person. A blind person's tactile sense is so much more acute. It's absolutely amazing. Because, you know, we don't use it and they do use it. So their other sensors are just heightened because of practice in using it.

But the greatest thrill, I would have to say, of my career as an engineer, was in the 1969 IEEE International Solid-State Circuits Conference. No pun intended — it’s the blue-chip conference of the integrated circuits industry. We submitted a paper on the Opticon, and the key to it was this retina. And so as a part of the presentation, which I made, at the end of it, Candy Linville, who was on stage with me, gave a live demonstration of the Opticon. You can see I'm pausing a little bit here because as I think about this, you know, it arouses my feelings. But it was the first time, and only time, I've ever heard and witnessed and participated in a technical paper that received a standing ovation.

Morton:

That's amazing.

Meindl:

When the audience heard Candy Linville read, they were just overcome with the feeling of, here's something we did, and look how it's helping this person.

Telesensory Systems

Morton:

What has happened to that technology since?

Meindl:

It became commercial. Actually, John Linville and several other people and myself founded a company and tried to get venture capitalists to back it. They wouldn't do it, but we were fortunate in that we had some people we knew well and who knew us well in the government, and so the government, the U.S. Office of Education, placed an order for fifty Opticons, with this little startup company that, you know, met in somebody's living room every night or somebody's garage on weekends. But based on that order, we were able to deliver, and that launched a company that was called — and is called today — Telesensory Systems. We started this in the early 1970s — 1972 — and it really was aimed at using high technology, electronics technology, for prosthetic aids for handicapped people. It didn't have any boundary, but we started out by dealing with blind people. And for many years it was — and still is — probably the largest producer of high-tech aids for the handicapped. It's dedicated to that and doesn't do anything else. It's a freestanding company today. It has changed; it has undergone evolution over the years, but it's probably in the range of, I don't know, annual revenues in a range of thirty-five, forty million dollars right now. I served on its board of directors for a dozen years or so, and then I resigned from the board shortly before I left Stanford. But it has evolved and is still, I think, the most active company dedicated to using advanced electronics for prosthetic aids for blind people.

Morton:

Let's follow that up a little bit. When you started this company, I take it you guys had the chips manufactured somewhere. How did you go about actually getting these things built?

Meindl:

Actually, we did just what you said. We had a small contract with a custom chip manufacturer. There weren't very many around at that time, but we had a contract with a small manufacturer. In one wafer, you could produce a couple hundred of the chips, so we did that. And then the other capital equipment needed to put together the original fifty Opticons was pretty minimal.

Morton:

So were there particular manufacturing problems with this chip? Or was it something that the company could sort of lay out and put into production without additional research?

Meindl:

That's correct. We were using what was state-of-the-art technology, I must say, for silicon transistors at that time. Of course, we knew how to make it: we made it at Stanford. I didn't say that, perhaps. But this first chip was made by a Ph.D. candidate, who's now the president and CEO of a semiconductor company named Seek. His name is Bill Salisbury, and we're still in contact; we're good friends, our families are good friends, our sons became good friends with each other. And so we made the first chips, designed them, built them, tested them, and had Candy Linvill read with them at Stanford, so we knew we had something that was viable. We couldn't convince the venture capitalists that that would make money, and probably they were right. But this company obviously was viable because it's a public company today, and it's profitable, and has been.

Stanford and fabrication and research in integrated circuits

Morton:

I'm glad you brought that up. You got an award in 1990 for the development of an academic program in fabrication and research in integrated circuits. I take it that was at Stanford. Maybe we could talk about that a little bit. That's a nice example of this issue of the university's place in innovation and in new technology. When did that get started?

Meindl:

In 1967, actually in parallel with working on the Opticon at Stanford. I started a research program to apply integrated circuits, not to prosthetic aids, but to medical operations. I can remember the first three or four months I was there, going over to the medical school, which was on the same campus, just a short walk away from the engineering laboratories, and talking with a cardiovascular surgeon, whose name was Norman Shumway I don't know if that is a bit familiar to you, but he did the original research in the early 1960s on transplanting parts using dogs, and one of his trainees at that time was Christiaan Barnard who did the first human heart transplant in South Africa. So, Norman Shumway was the dean of surgeons who did heart transplantation. I went to talk with Norman Shumway and several of his assistant professors from his group. One of the problems they were having was rejection of the heart. They wanted to know how to treat this with drugs. And so they said that one of the things they really wanted to measure was cardiac output — how much blood the heart was pumping. They wanted to be able to measure this over a long period of time, a year after, or more, after the transplant had taken place. There was no way you could do this from outside the body, especially over a year's time. So, they said, "What we'd like is a little package of telemetry. If there's some way, we'd like to be able to measure this." So I said, "OK, well, I just came from working on micropower circuits for use in portable field equipment. Let's go ahead and try to build a little package of electronics that we'll implant, surgically, in a dog that has had a transplanted heart, and we'll measure the blood flow."

So I worked on blood flow meters for the next twenty years. These were little ultrasonic transducers and they worked on a principle called Doppler effect. You've experienced this: many people have stood beside a railroad track and when a train is approaching, blowing its whistle, it seems as though the pitch is getting higher as the train is approaching you, it goes from a low to a higher pitch. And that's because you're kind of compressing the sound. waves as they are traveling towards you; you're making them shorter, and so the pitch seems higher. By measuring this change in frequency, you can actually calculate and measure the velocity of the approaching train. Every submarine uses sonar, and so does every large ship these days, and so we built a little sonar system and the targets for the sonar system were red blood cells. We put the ultrasonic crystal just outside the blood vessel and a cuff around the vessel, just the way your shirt cuff fits around your wrist, and beamed the ultrasound through the wall of the vessel and at the moving red blood cells, and they scattered and caused the frequency shift, called the Doppler shift, in the frequency. By measuring that frequency shift, you can calculate in effect and can measure how fast the blood cells were going. Knowing how fast they're moving and something about the profile velocity across the vessel, you can actually measure cardiac output. So we produced these little implantable telemetry packages and used them in research animals for all the twenty years I was at Stanford. We were able to convince NIH — National Institutes of Health — that if they gave us the resources to build these chips, we could build instruments that could be used in medical research to capture data that was otherwise unavailable and was completely novel. And that turned out to be a very good basis for building a center for integrated electronics and medicine, which I was able to participate in as director for — and leave as director — for most of the time I was at Stanford.

We had lots of other applications. For example, in addition to measuring rejection, I can remember, we used to work a bit inter-operatively. I can remember one case where a patient was getting a heart valve transplant. It was a donor valve, the new valve had been installed, and they wanted to test it. So we had another kind of ultrasonic blood flow meter that could actually measure the profile and could sense the blood velocity across the vessel. Immediately after the valve was installed — I know I'll never forget this — we did a scan of the blood profile and noticed that there was too reversed flow, which meant the valve wasn't closing. The blood, instead of being pumped out from the valve closing, it was flowing back into the left ventricle of the heart. And this was not expected, this wasn't correcting the problem that they had originally set out to correct. So they went back in and lo and behold, discovered that there was a calcium deposit on the valve and that's why it wasn't closing. So they had to go through another surgical session, right there, continue it, and corrected this by putting in an artificial valve, instead of a human replacement. That was a very memorable occasion, because this was an instrument that we had just built in the last couple of months or so before the operation. So that was exciting. And then we went on to build those that would be implanted.

Now let's see, one of the most memorable experiments was that we built a telemetry package that could sense the electrocardiogram and the EEG, the electroencephalogram, and the respiratory rate and one other variable that I can't quite recall at this moment. But this was a package that we implanted in a fetal monkey. In other words there was a monkey, a fetus, six months after conception, in the womb of the mother. So surgically, what people did was make an incision in the mother and through the wall of the uterus and then in the monkey and put in our little package of telemetry in the monkey and sewed him up. Our goal was to understand how these physiological systems of the baby monkey made their transformation from being on life support provided by the mother to freestanding independent functioning. And we did that work with people in obstetrics and gynecology at Emory University. So that was an interesting period, and that was the first time that was ever done.

So we did experiments like that for twenty years. And the goal, David, was to use integrated electronics technology to influence the welfare of mankind in a way that no private corporation could afford to do. Because the number of chips that you would ever be able to sell to do this kind of medical research could never justify the investment needed by a private corporation in developing those chips. You only needed a few dozen of them at most. So there was just no profit in making those custom chips. But, since forty percent of the male population over age forty has hypertension, if you could understand hypertension by following it chronically in a research animal and have better drug treatment of hypertension, the payoff for economy is enormous. The quality of life improvement for a lot of people can be enormous. So we tried to engage those kinds of problems in the research we did on applying novel integrated circuits and sensors in medical applications. And that lasted for all the twenty years I was at Stanford. And they still have a program in this area, as do a number of other universities as well.

Morton:

These were more or less innovations that stayed within the research community. These were more along the lines of scientific instruments rather than something that's also sort of biomedical technology like the reader that was more of a commercial product, or became a commercial product.

Meindl:

That's correct. We emphasized instruments that would be used for research because to take on the job of developing an instrument that could be used in treating human patients was a bigger job than we were prepared to tackle. It required more reliability, let's say, and more compliance with government requirements for safety for medical devices. We weren't prepared to undertake that sort of research in the university environment. So we tried to pick projects that would have very, very high leverage in terms of working with animal models, and then the medical community could extrapolate those results in their treatment of humans. In most cases, only few were implanted. But we tried to pick high leverage areas.

Morton:

You mentioned some resistance to integrated circuit technology in earlier years, from the discrete components people, and other people like that. Were the medical researchers ready to jump on this, or did they feel like there were better ways to do this? Did you have to sell this technology to them?

Meindl:

It wasn't a hard sell. The medical researchers that we worked with at Stanford were generally pretty ambitious in terms of wanting to do new things, in wanting to pursue opportunities that were unique. They were energetic and talented people. So if you could tell them, "If we join forces in this project, we can give you an instrument that nobody else in the world will have at their disposal" — that was an immediate entree. You could almost get anyone's attention if you can make that kind of statement. Almost any collaboration that we wanted to make we were able to make. Not with everyone, because some people didn't want to complicate their agenda with new instruments; they had enough opportunity and a set of resources using commercial instruments that they didn't need anything novel. But there were certain problems that couldn't be attacked if you didn't have these special instruments, and that's where we tried to intersect with the medical research community. And it worked pretty well.

Morton:

A general question about the university research in integrated circuits, not only at Stanford but elsewhere. What do you think the place of university research has been in this industry? Has it led developments?

Meindl:

I feel, looking back on it, that there's no question that the university has done its traditional function well: that is, produce educated people. Universities are equipped to build Ph.D. research programs and therefore provide the means for faculty to stay and become current, remain current, and therefore be able to pass along the information in courses to graduate and undergraduate students. So I think providing educated people that were really ready to contribute upon graduation or shortly thereafter is something that the U.S. academic community has done superbly. Throughout the history of integrated circuits, we were certainly doing it, and I think even to this day we have a big impact on industry in the United States; and that's become true in Japan and Europe as well. There are some good examples out there in those two places, although technological transfer from the United States had an impact as well.

Now, what did university research — the knowledge generated in universities — contribute? Well, there I think the contributions are not as easy to point out because I think that industry, by and large, over the years has been ahead. Big advances in the technology have come from industrial development laboratories. The universities have contributed, but not in ways that have been critical to the success of industry, in terms of research achievements or research contributions. I think lately, in the last fifteen years, as computerized design tools have become more and more critical, the universities have become a major producer of the first versions of computerized design tools that have then been commercialized by small companies and become widely used in industry. I would say, for example, that the most famous piece of software in the integrated circuit fields is a circuit-modeling program called SPICE, and it came out of the University of California at Berkeley. Today there are lots of commercial suppliers and there are different versions of Spice, but it started in the 1960s at Berkeley. And likewise at Stanford they developed the computer-aided design programs such as AutoCAD, T-CAD. Stanford also produced some very long-lasting and very important programs in understanding and modeling the physics of transistors. One of their programs is today called PISCES, but there were previous generations of this program, modeling transistor physics, and it's widely, widely used in succeeding generations. Finally, Stanford has also produced programs for modeling the fabrication processes for integrated circuits. For example, the programs that are in the family called SUPREME (for Stanford University Process Emulator) are very, very widely used in industry, not only the U.S. but elsewhere.

I think in the arena of software, universities have really come to the aid of the semiconductor industry worldwide. The capital investment that you need to do this is within the capability of the university. So there, as the importance of software has grown, the relevance and the immediate application of the product of research in universities, has been amplified. It's interesting to see that. I spent the twenty years that I was at Stanford founding and really managing a large integrated circuit fabrication laboratory, which I think was the model that was followed in other institutions in the U.S. and abroad. It's very expensive, as you go from generation to generation of fabrication equipment, to keep such a laboratory funded. At Georgia Tech, I have a simulation laboratory which just needs some good work stations and some software. Once you have those, it's a relatively nominal capital investment, compared to an experimental facility. I think that universities still need to stay involved in experimental work, and should, but it takes a very judicious selection of the projects that you attack in order to contribute and do it on an affordable basis.

Industry and semiconductor research

Morton:

We've hit the major players in this story in terms of universities, and industry and government. You've mentioned some very interesting specific examples of the sort of unique ways each can contribute — or each did contribute — to innovation, generally, in the industry. Are there any other examples like that you might mention? Is there something else—like the missile project that the government mandated would have ICs, or Stanford’s development of SPICE—that sort of was crucial in the history of the integrated circuit?

Meindl:

I think that one of the things that has been very significant for semiconductor research in universities has been the entry of industry into support of this research. There's an entity called Semiconductor Research Corporation that is located in Research Triangle Park in North Carolina and funded by the companies in the computer, semiconductor, and semiconductor manufacturing equipment industries. I think that the support of university research throughout the country by industry through this research consortium, the Semiconductor Research Corporation, has been very significant in bringing industry and university into contact and giving universities the resources and also the technical direction that is helpful to them in order to be productive.

I think that the other big issue today is that we don't really have a Bell Laboratories, a Watson Laboratories, an RCA Laboratories, or a General Electric central research laboratory in the way we did twenty years ago. These were industry laboratories, but they were doing relatively basic research. The United States doesn't have that national asset anymore. And how industries and how universities are going to cope with this remains to be seen. I think a current issue is how universities can be brought to bear on this vacancy that we've created. How they can fill this vacancy, and the research needs, generation of new technology, creation of emerging technologies? How can universities be brought into this picture in a very significant way? I think that the paradigm of interplay between government, industry, and universities — especially government and industry and government and universities that we developed after World War II — is changing. So is the way that long-range research in this country is being supported, and where it's being conducted. This is changing now in a way that is uprooting arrangements that we've had for the past almost fifty years. I think the main issue that we face in universities and industries is how to provide this new technology that's going to be able to generate the integrated circuit industry of tomorrow. I think that's a major question that is unresolved now. We're groping, we're searching as a community for answers to that question. I don't have the answer, but I think universities need to be brought to bear on the problem in a different way than they've been working in the past.

Morton:

Well, I think I've used up most of your morning. I don't have any more specific questions. If you've got any further comments, I'll be happy to go on.

Meindl:

No, I don't think I do. I think I've used up all of my information, too.

Morton:

Why don't we call it quits, then?

Meindl:

OK.

Morton:

Thanks a lot.

Meindl:

You're welcome.