Wizard of Watts
James D. Meindl caught the low-power semiconductor wave when it was barely a ripple and brought generations of graduate students along for an exciting ride
Photo: David Stuart
James D. Meindl, professor of microelectronics at the Georgia Institute of Technology, says the most important part of his job is making graduate school fun and exciting. Lots of professors make the same claim, of course, but Meindl, the winner of the 2006 IEEE Medal of Honor, has an explosive story to prove it.
It was the mid-1970s, and Meindl was a professor at Stanford University, in California. His group had just paid more than US $1 million for a shiny new epitaxial reactor, in which atoms are deposited layer by layer to produce semiconductor devices, usually experimental ones. It was the latest and greatest tool of the day, and Meindl assigned one of his newest and brightest students to see what it could do.
The department's safety rules forbade students from working alone, but that new student wasn't much for following rules. One night, working by himself, he opened a valve to let silane gas flow into the reactor. Alas, he'd forgotten to purge the air out of the chamber, and silane explodes on contact with oxygen. The resulting blast ripped the reactor out of the wall. The student was lucky to escape serious injury.
Clearly, he had to be punished. Meindl couldn't bring himself to do it, so he prevailed upon a colleague, who banned the young man from the laboratory for two weeks.
Even today, Meindl beams when discussing that brash young researcher. And well he might: that student, T.J. Rodgers, went on to found Cypress Semiconductor Corp., in San Jose, Calif. Last year Cypress had $886 million in revenues. ”Those were the good old days, when well-meaning accidents were just punished by a slap on the wrist,” Rodgers says today of the incident.
Before he met Meindl, Rodgers recalls, he had never worked on big problems involving the coordination of many individual research efforts. He had never heard of Silicon Valley and had never known an engineer who had started a company. Meindl brought him into this incredibly exciting world, he recalls, and ”it was thrilling.”
Meindl says of his students, ”My reward is to see them succeed.” He's been very well rewarded. Rodgers was one of some 80 engineers who did their graduate work under Meindl's tutelage. Among the others are William Brody, president of Johns Hopkins University, in Baltimore; Levy Gerzberg, president of Zoran Corp., the Silicon Valley company whose signal-processing hardware is in just about every digital camera today; and Richard Swanson, who founded SunPower, a pioneer in high-efficiency solar cells, which was purchased by Cypress in 2002.
Another Meindl disciple, L. Rafael Reif, provost at the Massachusetts Institute of Technology, in Cambridge, says, ”I try to emulate everything about him: I listen to everyone. I try to find the kernels of truth in what people are saying. I try always to find the glass half full.”
Gerzberg credits the success of Zoran to the lessons he learned from Meindl. And Jim Plummer, dean of engineering at Stanford, declares, ”There is no other individual who has had more of an impact on my career than Jim Meindl.”
For many professors, it would be legacy enough to have sent so many students into the hallowed halls of academia and the boardrooms of Silicon Valley. But for Meindl, it's just part of the story. He also did pathbreaking research in the design of low-power circuits and in the interconnects that link blocks of logic in a chip. It was for those achievements—specifically, his ”pioneering contributions to microelectronics, including low-power, biomedical, physical limits, and on-chip interconnect networks”—that he was awarded this year's IEEE Medal of Honor.
And yet, when Meindl started out in technology, half a century ago, semiconductors were little more than laboratory curiosities. Enrolling in 1951 at Carnegie Institute of Technology (now Carnegie Mellon University), in Pittsburgh, he planned to get a degree in power engineering and then design heavy electrical equipment at Westinghouse Electric Corp., where his father worked. But in 1955, Carnegie's graduate power engineering department abruptly vanished: one professor quit and another changed fields.
Meindl, then just starting graduate school, was open to suggestions about what to do next, and Professor Edward Schatz had one. The U.S. military was trying to improve its communications systems and needed an energetic graduate student to analyze the loss of radio-frequency signals transmitted through coaxial cables. The goal was to come up with equations that would describe the signal loss and allow engineers to build better cables. Meindl solved the problem in 24 months, in the process becoming quite well versed in Maxwell's equations, the set of four equations describing the behavior of electric and magnetic fields and their relationships to each other and to electric-charge and electric-current density. Facility with these equations, which are considered to be the foundation of electrical engineering, gave Meindl the insights he needed to understand that latest electronic marvel, the semiconductor.
Meindl got his Ph.D. in electrical engineering in 1958. He took a job at Westinghouse, as he'd planned all along, but not as a power engineer. Instead, he became one of the company's first semiconductor engineers. His initial assignment was to use silicon-controlled rectifiers—diodes that must be triggered by a voltage pulse in order to conduct current—as part of an electronic system that would manage the control rods of a nuclear reactor. He loved the job. ”I got to buy and burn out transistors that cost about a thousand dollars each,” he recalls. ”And I learned that electrical engineering can be a lot of fun.”
Unfortunately, the fun only lasted about a year. In 1959 he received a summons from the U.S. Army. It was payback time. Meindl, a member of the Reserve Officers' Training Corps program during college, went on active duty.
”Here I was, working,” he recalls. ”I had a great job, I had bought my first new car, and I just felt that the world is a really good place. And suddenly, here comes Uncle Sam.”
He ended up at the U.S. Army Signal Research and Development Laboratories, in Fort Monmouth, N.J. [see photos, ” One Haircut, Three Decades”]. It turned out to be ”the most fortunate unwanted experience” of his life. For one thing, he met his future wife, Frederica. She was an administrative assistant for Meindl's supervising officer. They had their first date that October, and the rest, Meindl says, is history.
Encountering his life partner wasn't the only happy surprise at Fort Monmouth. There was also his work assignment, which turned out to be a lot more interesting than he was expecting: he worked with integrated circuits—a field then barely six months old.
Just after Meindl arrived at the R&D labs, the Army awarded a research contract to Dallas-based Texas Instruments Inc., where Jack Kilby had built one of the first ICs. Meindl became the technical liaison for the project. He met Kilby in November; a few months later, early in 1959, he visited Gordon Moore and Robert Noyce at Fairchild Semiconductor Corp., in Palo Alto, Calif. Those three pioneers taught Meindl about the nascent field, and he began his own research, trying to figure out how to make an IC operate at a power level so low that it could be used inside a helmet as part of a radio receiver. Meindl stayed at Fort Monmouth for eight years, two as an Army officer and six more as a civilian.
In the early 1960s, hardly any engineers outside the military were interested in minimizing the power used by electronics. Metal-oxide semiconductors (MOSs), which consumed significantly less power than their bipolar predecessors, were in their infancy and had stability problems. But as the decade went on, the growing popularity of the quartz watch and the in-ear hearing aid, both of which could accommodate only tiny batteries, brought attention to the need for low-power circuits. Still, when Meindl published his first and only book, Micropower Circuits (Wiley), in 1969, you could count the number of copies sold on two hands.
But Meindl was onto something. ”Even at a time when few people worried about power consumption, he thought it was an interesting area to explore, because it would eventually become a big issue,” Plummer says. And so it has: laptop computers, cellphones, and iPods illustrate that product design today is all about reducing power consumption and extending battery life.
By 1966, several professors at Stanford were encouraging Meindl to leave New Jersey and join them in California. In 1967, John Linvill, then chair of the electrical engineering department at Stanford, made Meindl an offer he couldn't refuse. Linvill had come up with an idea for a system that would let blind people—including Linvill's own young daughter, Candace—read. It would use a camera to take a picture of the letters on a page and then translate that picture to a tiny pad of vibrating pins. With training, Linvill reasoned, a blind person would be able to place a finger on the pad and decipher the text. But making such a device portable and useful required two custom-designed, low-power chips. One chip would act as the image sensor—a solid-state camera, basically, at a time when they were experimental. The other chip would operate at a high voltage to vibrate the tactile array, consuming as little power as possible to prolong battery life.
Meindl worked on the project for about a year, along with several graduate students, including Plummer.
”We had significant problems,” Plummer recalls. They were using MOS devices in a high-voltage application—which no one had done before. After a lot of trial and error with the voltage levels, they finally found one that was high enough for the vibration to be felt by the user and yet low enough to keep the devices from burning out.
The group dubbed the device the Optacon, for optical-to-tactile converter, and demonstrated it for the first time at the 1969 International Solid-State Circuits Conference, in Philadelphia. Linvill's daughter Candace demonstrated the converter, and she got a standing ovation. ”That,” Meindl says, ”was the most thrilling moment in engineering work that I have ever had.” He later named his own daughter Candace to honor Linvill's daughter and the moment.
In 1970 Linvill, Meindl, and their team rolled the technology out into a company, Telesensory Systems Inc., now a division of the Singapore company Insiphil. Telesensory produced tens of thousands of the devices and sold them around the world. Today, text-to-speech converters have supplanted the Optacon, but it was an important aid in its time.
Telesensory never made its founders a fortune, but that didn't bother Meindl. Throughout his career, he says, he and his co-workers have always selected ”areas that could have the most impact.”
Inspired by that moment at the 1969 conference, Meindl asked a group of his Stanford students to develop novel low-power sensors and circuits for use in medical research. Gerzberg, who was part of the group, recalls that challenges were everywhere: in signal-processing algorithms, in circuit design, in chip fabrication, in systems integration, and in coordinating with medical researchers. Gerzberg also remembers that Meindl's enthusiasm never wavered. Gerzberg says, ”Once I gave him a demo of a prototype I had built, the first time I had it working, and he actually stood up and clapped his hands for several minutes. He always made me feel so good.”
The students built sensor packages that could be implanted in research animals, transmitting physiological data while the animal went about its normal activities. A medical researcher inserted one such sensor package in a monkey fetus still in the mother's uterus. The monkey mother eventually went into natural labor; the researcher then wirelessly activated the sensor package to provide the first detailed information about the baby's physiological experience during birth. Meindl and several graduate students designed another device to measure the velocity of blood cells as they flowed through different parts of the body. Meindl attended a procedure in which a surgeon used this device to monitor the progress of a heart valve replacement; after the artificial valve had been sewn into place, the flowmeter flagged a problem with the new valve, and the surgeon quickly replaced it, possibly saving the patient's life.
Another achievement at Stanford hasn't saved any lives, but it has saved many designers countless hours of R&D time. In 1971, Meindl posed a simple question to graduate student Swanson: theoretically, what is the lowest possible voltage at which an arbitrary complementary metal-oxide semiconductor (CMOS) circuit could operate? Knowing this value would prevent circuit designers from exploring dead ends—techniques that would drive the voltage so low the designs wouldn't work. Swanson determined that the minimum voltage is a multiple of the thermal energy of the material, with that multiple changing in a predictable way depending on the temperature of the material. Circuit designers the world over still use that fundamental limit.
By the mid-1980s, after nearly two decades at Stanford, Meindl was ready for a new challenge. ”I'd had a number of rewarding experiences,” he says. ”The question was whether I was going to do this for another two decades or try something drastically different.”
”Drastically different” won. In 1986 he moved to Troy, N.Y., to become the provost of Rensselaer Polytechnic Institute (RPI). The job appealed to Meindl for two reasons. First, he saw being a provost as an intellectual smorgasbord—no longer would he be simply interacting with his engineering colleagues: he would learn about new ideas in every department. Second, he was excited about the idea of having a discretionary budget to fund ideas and reap their returns.
Life as a provost, however, was not as rosy as Meindl had anticipated. Colleges across the United States had begun feeling the waning of the post–World War II baby boom, as the so-called baby bust took its toll on their applicant pools. And RPI, more than most U.S. universities, lived or died on tuition income. So for seven years Meindl cut departmental budgets and cut them again. For a guy who liked to make people happy, it was not a happy time. Still, he kept his cool and deftly handled the faculty, recalls Robert Loewy, then a professor at RPI.
Meindl kept a hand in electrical engineering by advising two graduate students—Vivek De, now manager of low-power circuit technology at Intel Corp. in Hillsboro, Ore., and Bhavna Agrawal, now team leader for analog computer-aided design at IBM Corp. in Zurich, Switzerland. It was a bright patch in an otherwise dark stretch. ”Last time I heard,” Meindl beams proudly, ”one of those students [De] had the second-largest number of patents of any Intel employee.”
In 1993, enough was finally enough. Meindl went back to doing what truly makes him happy: teaching and research, this time at the Georgia Institute of Technology, in Atlanta.
Today, at 73, Meindl is a happy man again. He is pursuing one of his long-standing interests: optimizing the arrangement of interconnect wires that connect blocks of logic circuitry on a chip. The Interconnect Focus Center, in Atlanta, a major R&D effort he organized eight years ago with 13 U.S. universities, recently demonstrated a new high-speed optical modulator mechanism for silicon chips, a key component in future optical interconnects.
Says Plummer, ”Back when most people had yet to recognize that interconnects would be a limiter in silicon chips, Jim launched a national program to tackle the issue.”
Indeed, the semiconductor industry has come around to Meindl's way of thinking. Engineers today recognize interconnects as a major impediment to the performance trajectory that microprocessors have been on for the past 35 years. It is the wires, not the transistors themselves, that are sucking up power, threatening chip performance, and dragging out design cycles. In today's billion-plus transistor chips, which have multiple layers of wires connecting transistors and many kilometers of interconnects per square centimeter, the wires cost more than the transistors (see ”Chips Go Vertical,” IEEE Spectrum, March 2004).
So far, Meindl and his Georgia Tech team, most prominently Jeff Davis, a professor there, have come up with a mathematical method to predict the distribution of interconnect lengths within a chip. That is, given the size of a network of logic blocks that must be wired and a set of possible lengths of each wire, a designer can predict how many wires are likely to have each length. The designer can use that information to select the optimal widths for the wires of each length for maximum performance at the lowest cost before beginning to actually lay out the chip. Meindl's group is also trying to find ways to use tubelike interconnects to remove heat from a chip; such wires would move cooling water in and out of the chip using a nascent technology known as microfluidics.
In between his research and teaching, Meindl finds time to run the Focus Center and also act as Georgia Tech's site director for the National Nanotechnology Infrastructure Network. And he's on the boards of three companies: SanDisk, Zoran, and Stratex Networks.
Eight years past the traditional retirement age, Meindl is not thinking of slowing down anytime soon. Nanotechnology and its potential for pushing the envelope of chip design beckons. In one project, he and his students are trying to wire chips with carbon nanotubes instead of copper traces.
But still, fundamentally, it's about the students. Serving on boards, Meindl says, makes him a better graduate advisor, because it helps him understand how companies operate, what research is needed, and what opportunities await new graduates. Involvement with large research cooperatives means he can keep track of activities at a host of universities and spot opportunities for his students to become innovators. And he really needs to understand nanotechnology, he says, because ”it is going to infest every branch of engineering. Students will have to know about it to use the latest developments, no matter what area of technology they pursue.”
Continuing to challenge himself daily in the field he loves, Meindl forges ahead. His busy schedule keeps him on the go, from classroom to boardroom to research lab. As he passes through a hotel lobby on a chilly February day, he has a personal greeting and compliment for every one of the hotel staff members, leaving a trail of smiles in his wake. For Jim Meindl, it's still about making people happy.