Challenging Assumptions: Q&A With Yannis Tsividis

Nov 01 2019 | By Allison Elliott
Portrait of Prof. Yannis Tsividis.

Tsividis in his office at Columbia.

Credited with building the first fully integrated MOSFET operational amplifier in 1976, Yannis Tsividis has since built an illustrious career as a researcher, mentor, and educator. For almost 40 years, he’s guided Columbia Engineering students through the ins and outs of circuits and devices and, as Edwin Howard Armstrong Professor of Electrical Engineering, he’s been a major force in shaping the overall direction of the department’s curriculum, ensuring it reflected an evolving discipline and the varying needs of students.

In 2019, Tsividis was elected to the National Academy of Engineering in recognition of his contributions to the field. Over the years, his many other accolades include the Institute of Electrical and Electronic Engineers (IEEE) W.R.G. Baker Award, the IEEE Undergraduate Teaching Award, the IEEE Circuits and Systems Education Award, and Columbia’s Presidential Award for Outstanding Teaching. A fellow of the IEEE, he also received the IEEE Gustav Robert Kirchhoff Award and was named Professor Honoris Causa by the University of Patras, Greece. He is the author of three books, including the industry standard, Operation and Modeling of the MOS Transistor.

Tsividis spoke with Columbia Engineering about his student days running a pirate radio station in Greece, how to create an inspiring research environment, and his forward-looking approach to research and teaching.

Technology changes very fast. Fundamentals don’t.

Yannis Tsividis
Edwin Howard Armstrong Professor of Electrical Engineering

You’re known for having a strong dedication to fundamentals as a way to develop a deep conceptual understanding of a given topic. Why do you think this is such a critical skill for engineers?

Technology changes very fast. Fundamentals don’t. Thus, for example, the silicon chips in today’s computers are very different from those of decades ago and accomplish much more, yet within those chips the fundamental laws that govern the behavior of electrical quantities—such as charge, current, and voltage—still apply. The problem is how to convince students that fundamentals are not just mathematical formulas to which they have to insert numbers; at least in my field, they need to have a real “feel” for the fundamentals, which allows them to build up their intuition. As for me, my research would be nowhere without the fundamentals.

You are known for your groundbreaking work in MOS (metal-oxide-semiconductor) technology for analog circuit design. Though MOS transistors eventually revolutionized the electronics industry, at one point few recognized them as a powerful option for analog circuits. How did you see the potential of this technology?

Actually, it wasn’t me who initially saw the potential; my PhD adviser at Berkeley, Paul Gray, did. But once he mentioned it to me, he didn’t have to try hard to convince me. I saw the promise and the opportunity right away. I had gone through two other topics, with different advisers, and could not get excited. Once I was asked to come up with analog circuits in MOS technology, which back then was essentially a digital technology, I was hooked. In technology, just like in society, there are plenty of prejudices; one of the prejudices at the time was that MOS transistors are very poor for analog circuits work, which is much more demanding than digital circuits work. So I had a ball with this. I developed new techniques that bypassed the transistor limitations. I also looked carefully at the transistor properties, to see if they could do some unique things. It turned out that they could. Today, analog MOS circuits are everywhere (in your cell phone and in biomedical devices, for example).

I love teaching. And it teaches me; when I want to really understand a subject, I try to teach a course on it.

Yannis Tsividis
Edwin Howard Armstrong Professor of Electrical Engineering

Are there other highlights or turning points in your career where you questioned an assumption in your field, and by doing so, took the research in a new direction?

Yes, I challenge assumptions often, and I have two reasons for doing it. One is that there is real challenge in it; the other is that I can have some peace and quiet to do my research, without feeling that I am in a horse race with a great many capable people working in the same area. In our group, we have worked on nonlinear systems that behave linearly externally, which affords certain advantages in noise performance. Also, we have worked on digital systems that operate with ones and zeros as functions of continuous time, thus taking advantage of the time dimension. Of course, to do such things takes daring PhD students. I was lucky to have such students in my group.

You’ve described your time at Bell Labs in the ’70s and ’80s as “research heaven,” where you were able to follow your own interests in integrated circuits. What made it such an ideal environment for research?

I will give you an example. A group at Bell Labs heard of my results at Berkeley, and they invited me to join them and apply those results in what they were doing. I did join them, full time during the summers and part time during the rest of the year, for about a decade. When I would ask my department head, George E. Smith, what I should work on, he would answer, “Do what excites you.” He understood what open research means, and that you need to allow freedom in order to increase the chances that something important will result. Not coincidentally, he went on to win the Nobel Prize in physics. Bell Labs was responsible for a score of important developments, from the invention of the transistor to dominant theories about the universe. There aren’t many such places today. In my opinion, the way research is incentivized these days often inhibits creativity. The rise of bureaucratic requirements, both in the funding agencies and in the universities, doesn’t help either.

Over the years, you’ve taught multiple generations of students at Columbia, each arriving with very different backgrounds— both in terms of the subject matter and technology in general. In the ’90s, you took time off from teaching because you felt it was time to rethink how the department introduced circuit theory to students. What did you learn and how did it affect the way you teach, both then and now?

I interviewed many students back then, and I realized that what used to excite my generation just was not exciting enough for them. You could not tell them, “Study math and physics for two years, and eventually, starting in the third year, you will see why this is useful.” This was the immediate gratification generation; they played video games and saw the results of an action in milliseconds. Previously, I had succeeded in convincing my department to move the first electrical engineering course from the third to the second year; this time, I convinced them to move that course to the first year and to get students in the lab, working with real circuits, right away. That did it. We doubled our enrollment within two years.

You are also something of an early pioneer in innovative teaching methods. As a student at the University of Minnesota you took it upon yourself to create lab tapes to educate students on using equipment. Later at Columbia, you piloted one of the first MOOCs (massive open online courses) for remote learners. What inspired you to adopt these methods, and how do you see technology shaping engineering education itself?

I love teaching. And it teaches me; when I want to really understand a subject, I try to teach a course on it. This takes a lot of work, but it pays in the end. The MOOC was different, though, in that it was based on a topic on which I had written a book. I did the MOOC (called MOS Transistors) in part because the Dean’s Office asked me to do it and in part because I was curious about MOOCs. It was successful, and Coursera still runs it today. My objective with it was to present in detail the operation and modeling of the MOS transistor—the workhorse of today’s electronics. Many engineers attempt to design circuits without such detailed knowledge, and this can hurt their designs. Initially, tens of thousands of people signed up for it, with a very wide range of backgrounds—from university professors and industry engineers, to high school students. But significant maturity was required and, eventually, those who stuck with the course were able to get from it what a graduate course on the subject is supposed to offer. I keep hearing from people who took the course and are happy with what they learned. MOOCs have found their place today, and they have helped “democratize” education.

Black and white picture of Tsividis as a graduate student working in a lab at Berkeley.

Making analog MOS silicon chips as a graduate student at Berkeley.

You’ve been on the frontlines of many changes in electronics research and development. To your mind, where do you see electrical engineering research going in the future?

I have to be careful here. You’ve heard about stock market bubbles. . . . Well, we have bubbles in research, too. Something is touted as being the next great thing; then, years and hundreds of millions of dollars down the road, it proves not feasible. So I will avoid judging a number of currently “hot” ideas out there. But I will mention one idea that is currently not “hot,” although it should be. We should be making better use of the time dimension. In today’s computers, the time interval between two successive events does not convey useful information. But it could! We have now produced test chips that show that this is feasible, and it provides significant advantages. Another pet project in our group is analog computing for fast and efficient scientific simulation— an idea whose time has come (again), thanks to advances in chip technology.

Congratulations on your election to the National Academy of Engineering, a recognition of your long and successful career. What has this acknowledgement meant to you?

I was delighted by this. It means a lot to me, especially because, when I look at the Academy’s membership, I realize that I am in very good company.

You have said that your love of electronics was partly inspired by your love of radio and music; you built and ran a pirate radio station as a student in Greece, and you are known for including musical components in your labs. What are some favorite musicians of yours?

Let me just mention, in no particular order, some of the musicians I used to play back then on the radio because, well, they were my favorites: The Animals; James Brown; Dave Brubeck; Aretha Franklin; Booker T. & the M.G.’s; The Beatles; Otis Redding; Blood, Sweat and Tears; Creedence Clearwater Revival; Stevie Wonder; The Mamas and the Papas; Paul Revere & the Raiders; Jimi Hendrix; Iron Butterfly.

Stay up-to-date with the Columbia Engineering newsletter

* indicates required