In my arbitrary counting system, I’ve tagged Bob Widlar as guru #1. Of the five people in the advertisement, his life is probably the best documented. I recommend two excellent sources: Bo Lojek’s “History of Semiconductor Engineering” who devotes an entire chapter to Bob Widlar’s life, and Wikipedia. The Wikipedia page is quickly digestible (and free) and offers a reasonably technical discussion of Widlar’s early works. And the Wikipedia page references Lojek’s book (not free) extensively.



I won’t attempt to write a better story than either of those two sources. Instead, I will offer my reaction to both of them and put a little spin (or anti-spin, as it were) on them.

“History of Semiconductor Engineering”, Chapter 8: Robert J. Widlar – The Genius, The Legend, The Bohemian. Lojek says, “Bob was a fiercely independent individual, very happy to be by himself, and he did everything in a stunning way, which was absolutely natural to him, but completely weird to so-called “normal people.” I completely agree with that! He also said, “…the personality of Bob Widlar is a clear manifestation of unhealthy changes which occurs in our society.” By that, I believe he meant that today’s “political correctness” stifles what engineers say, do and how they pursue their ideas; and therefore, we won’t see someone like Bob Widlar again. I don’t completely agree with that – partially, but not completely. Then again, I never met Widlar. But I look around and I see a lot of character, a lot of people being true to who they are and an acceptance of eccentricity. And I think that Widlar deserves a great deal of credit for that acceptance surviving till today. Simply because he existed and accomplished what he did in the manner that he did, we integrate that into what is acceptable – at least in analog.

Wikipedia concisely covers Widlar’s childhood, education, stint in the Air Force, work at Ball Brothers, segue to Fairchild and then National and Linear Technology where his face appeared on this ad. I gave my own version of these events in my post attempting to account for 10,000 hours. Wikipedia doesn’t clarify how Widlar got his degree in the same three years that he served in the Air Force – and at the same time he wrote the textbooks for the Air Force’s instruction in semiconductors (when no suitable text existed). Wikipedia discusses the µA702 and µA709. But you don’t get a mental picture of Widlar and Talbert working nights and weekends to make the process and product concurrently, without management approval – they had other jobs. The µA702 was introduced a year after Widlar joined Fairchild – probably only a few months after management found out about it. Widlar locked himself away for 170 hours and came out with the µA709 – that’s a week (without much sleep).

On a technicality, Wikipedia makes mention of a patent dispute between Widlar and Linear Technology regarding parts LT1 through LT20. Those are not correct part numbers. In Lojek’s book, the part numbers are LT1-10, -15, -16, -17, -18 and 20. I suspect those are simple typographic errors. I know Widlar designed the LT1010. So it’s possible that LT1015, LT1016, LT1017, LT1018 and LT1020 were also his designs – I’ll check on that.

My real issue with the Wikipedia page is the personality section. The image of him as a heavy drinker seems to grow with time while the image of him as a hard-working, brilliant, prodigious engineer fades as time goes on. And that has to do with context, in my opinion. In the late 1960s and 1970s, the semiconductor industry was the “wild west.” The local watering holes (Walker’s Wagon Wheel, Marchetti’s, etc.) were packed – during lunch and all night. They were crazy times. True, even in those days Widlar stood out. But against the backdrop of today’s sensibilities, Widlar looks even wilder. His hard-working side, his genius, his many contributions fade against the backdrop of technology today. We are spoiled by the accumulated knowledge of circuits at our fingertips. It’s impossible for us to imagine a time before the bandgap reference, before matched transistor pairs and temperature compensation. Widlar worked with a slide rule. He studied the process and drove the process with Dave Talbert to create what was necessary and he invented the rest out of sheer genius.

Accepting that, I’m astonished by the amazing speed with which it all happened. Remember, Widlar worked in the applications group – not R&D or circuit design. Talbert was a process engineer for the microLogic group. They did everything after hours! Widlar joined Fairchild in 1963, released the µA702 in 1964 and µA709 in 1965. Along with “inventing the IC op amp”, he laid the blueprint for how all analog ICs data sheets are written. He wrote his own applications section. He traveled the world giving seminars – he gave a seminar in Madison Square Garden and packed the house! By the end of 1965 he left Fairchild and went to Molectro. His mere presence there was enough to convince National Semiconductor to abandon their east coast roots and build a company around it. In 1966 Widlar made the LM100 and, in 1967, the still legendary LM101. A career’s worth of products in consecutive years, bang, bang, bang. Amazing.

So forgive my soap-box speech, but I think Wikipedia needs to be changed. I’d like to see our community come up with better language than “…lived the life of an alcoholic loner.” Wikipedia can be changed. Yes, he drank (a lot) and preferred to be alone. But I can’t reconcile the image of an alcoholic loner with the technical accomplishments of Bob Widlar, with the obsessive amount of work he poured into every circuit, with the charm and humor or with the business savvy. Sorry, had to put that out there.

Think about the next analog circuit you work with. So much of that circuit exists because of Bob Widlar.

Guru #1.
Bob Widlar, center, at National Semiconductor in 1967

3

View comments

Silicon Valley Analog Semiconductor Companies Decades ago, when Don Hoefler made the phrase "Silicon Valley" famous, Fairchild Semiconductor was the center of the universe. This was still mostly true in the myopic view of analog semiconductor veterans, as well. Some may argue. And they would be right. Absent from this view are the companies outside of Silicon Valley (and a few select others). I'll get to those, later. Link to external presentation Absent from today's landscape are ALL of these companies. Fairchild Family Tree One could argue that the seminal analog integrated circuit was the µA702, designed by Bob Widlar in 1963 (released in 1964) when he joined Fairchild. It wasn't good enough, in his opinion, so he designed the µA709 which was superior - both in performance and in sales revenue. Frustrated with Fairchild management, he and Dave Talbert (an equally brilliant fab process engineer) joined Molectro. You could argue that Molectro - founded by James Nall from Fairchild - was a spin-off, but it was intended to create custom logic ICs. Talbert and Widlar thought their process would be good for analog. Back then, National Semiconductor was a stodgy transistor company from back east. In 1967, Charlie Sporck, Fred Bialek, and Pierre Lamond - all from Fairchild - essentially re-founded National as a Silicon Valley company by acquiring (or merging with, and then relocating to) Molectro. That brought Widlar to National. Signetics was founded in 1961, Intersil in 1967, and PMI in 1969 - all from Fairchild. In The Beginning Analog has been around forever; it's the way the natural world works. Analog circuits, like op amps, were made with vacuum tubes before integrated circuits or even transistors were invented. The semiconductor transistor was invented at Bell Telephone Laboratories in 1947. The effort had begun years earlier with government funding during World War II. There was a wartime culture of cooperation and communication among research labs, universities, and military-industrial companies. The invention was licensed to many companies in 1952. William Shockley, whose group at Bell Labs demonstrated the first transistor, left in 1955 and started his own company - Shockley Transistor. In 1957, key members of his company left to start a new company - Fairchild Semiconductor. Silicon Valley in 1963 By 1963, when Widlar was designing the µA702, Fairchild was attracting talented engineers but also losing them. A few original Fairchild founders had already started new companies. This wave was significant because they were all emerging IC companies - they probably still made transistors and diodes - but the future of ICs was clear. Ill-fated Rheem Semiconductor is an interesting story. They stole some IP from Fairchild on their way out. But it was for "mesa" transistors right about the time Jean Hoerni invented "planar" processing, which is what made ICs practical. Amelco was renamed Teledyne Semiconductor rather quickly. Ultimately, they were acquired by Microchip Technology. Schismogenesis Schismogenesis literally means "creation of division". Gregory Bateson first used the term in a publication in 1935 to describe certain forms of social behavior between groups. "Spin-off", in my opinion, has a neutral feeling and implies that the parent company agreed to release some part of its organization as a new company. The Fairchild spin-offs were generally not amicable and often involved lawsuits. Typically, a group of engineers believed they had a great idea that needed funding but Fairchild management felt otherwise. That created a schism. Very quickly, that lead to the genesis of a new company founded by those engineers. In "The Dawn of Everything" (2021), anthropologist David Graeber and archaeologist David Wengrow suggest that schismogenesis describes the way groups define themselves in opposition to other groups. "Fairchild management only gave stock options to senior executives, we will give stock options to every employee." Fairchild, National, Intersil, and AMD were all broadbased chip companies, i.e., they made logic, memory, discrete transistors, and analog. Invariably, the schism was that analog went underfunded and the genesis was a group of pure analog companies. Acquisitions Starting on the top row:American Microsystems (1966) was renamed AMI Semiconductor (AMIS), then acquired by ON Semi in 2008.What was left of the original Fairchild was acquired by National Semiconductor in 1987, then spun-out several years later, then acquired by ON Semi in 2015.ON Semi, itself, was a spin-off of Motorola in 1999. What was left of Motorola Semiconductor was spun-off as Freescale in 2003.The blue hats are the analog companies shown on the earlier slides. Several of the companies shown here are analog companies. Second row: Signetics was acquired by Philips in 1975 and later became Philips Semiconductor.Philips spun-off the semiconductor group as NXP in 2006.In 2015, Freescale was acquired by (merged with?) NXP.Third row: In 1965, Analog Devices was founded as a vacuum tube module company. In 1971, it created an IC company which it then acquired.Computer Labs (data acquisition boards, later data acquisition ICs) was acquired in 1978.Precision Monolithics was acquired in 1990.Hittite Microwave was acquired in 2014.Linear Technology was acquired in 2016.Maxim was acquired in 2020.ADI has acquired many other companies along the way. This only highlights the significant analog acquisitions. Fourth row: Texas Instruments was founded in 1951 from a reorganization of Geophysical Services, Inc.Unitrode was acquired in 1999.Burr-Brown was acquired in 2000.National Semiconductor was acquired in 2011.Fifth row: Amelco was acquired in 1960, the first acquisition of the newly formed venture group called Teledyne. It was spun-off as TelCom Semiconductor in 1993 and acquired by Microchip Technology in 2000.Supertex was acquired in 2014.Micrel was acquired in 2015.Sixth (bottom) row: The original Intersil was absorbed into GE then acquired by Harris in 1988, who then spun it off in 1999.Elantec was acquired by Intersil in 2002.Xicor was acquired by Intersil in 2004.Intersil was acquired by Renesas in 2017.The companies aligned on the right of the chart currently exist, but none of their headquarters are in Silicon Valley. They do, by virtue of these acquisitions, have a presence here.

Robert H. Swanson, Jr., built a company — a culture — that was the most profitable in the semiconductor industry, yet flew below the radar. Its employees famously ‘never left’ in Silicon Valley where job-hopping is applauded. Once called the Toughest CEO in the Valley, he attracted and retained many of the most talented and fiercely independent analog engineers. For over three decades, relentlessly vigilant to his founding strategy, Linear Technology Corporation focused on analog while the rest of the industry went digital. There’s no simple equation for how he did it, nor did he do it alone. To understand better, you have to integrate the legendary characters he met and the forces at play throughout his career, including his early years at the star-crossed Transitron Electronic Corporation. The Cold War In the skies outside of Boston after World War II, you couldn’t mistake the screaming whistle of a P-51 Mustang in flight. Bob Swanson could watch military training exercises from his house near Hanscom Air Force Base and dream of flying. Some people look to the sky and see possibilities, Bob looked to the sky and saw dogfights. Like many kids of his generation, he wanted to be a fighter pilot. American nerves were on edge in the 1950s due to the Cold War and the constant threat of a nuclear attack from the Soviet Union. In 1953, news of a Soviet rocket launch proved they were ahead in the arms race, making war more likely and intensifying the need for the nation’s best and brightest to do their part. The most direct path to flight school was the Navy’s 18-month program that required at least two years of college. So Bob enrolled at Wentworth Institute of Technology for a two-year degree in Engineering. No one from his family had ever attended college. While in school, the inter-continental ballistic missile program had a series of spectacular failures with Atlas rockets blowing up within seconds of lift-off. That made the news of the Sputnik satellite all the more shocking. In October of 1957, the Soviet Union succeeded in launching a beachball-sized satellite into orbit — proving nuclear warheads could reach America. The urgency created by Sputnik highlighted the shortage of scientists and engineers; one could sense throughout the community that studying engineering was a patriotic duty. Bob’s focus was math, but he was intrigued by a course in transistor physics. The nascent transistor technology was still cutting-edge and it was quite rare for any school to offer such a course. He would go on to get a degree in Industrial Engineering from Northeastern University. Fresh out of school in May 1960, Bob took the test to go into the Navy. Had there been an immediate slot for pilot training he would have gone down that path. But Bob would have to cool his jets until December if he wanted to be a Navy pilot. Meanwhile, he needed a job to make car payments. Transitron Electronic Corporation The world’s second-largest semiconductor electronics company at the time, Transitron, happened to be less than 10 miles from his home. They had just won a major contract with Lockheed to build transistors for the Polaris missile program and were hiring. Despite having a safer job offer from the more established Polaroid, he decided to apply. Transitron Electronic Corporation company headquarters, used with permission © the Nick DeWolf FoundationHe stood in line with people applying to be operators, engineers, and clerks. There was a line on the application for special skills, so he wrote “advanced statistics”. The personnel manager happened to be milling around the lobby and picked up his application. “Do you know statistics?” he asked. “Yeah,” Bob nodded. “Come with me,” he said as he walked away. They took Bob to a room and brought in several people who started grilling him on statistics. Bob knew statistics, including how to do a chi-square test. Along with the heads of manufacturing and transistor development, was the chief engineer — a tall, lanky guy named Nick DeWolf. Bob had no idea who they were or why such senior folks were so interested in him. Evidently, one of the requirements from Lockheed for the Polaris germanium transistors was statistical control including the chi-square distribution. They offered Bob a job on the spot. Bob explained that he had an offer from Polaroid but he couldn’t start for two weeks because his new boss was on vacation. So Nick said, “I’ll tell you what. You work here for two weeks and if you don’t like what you’re doing then you can go work for them.” Let’s pause for some context. The transistor was invented at Bell Laboratories in 1947. Western Electric licensed the technology in 1951; Western Electric was the commercial arm of AT&T whereas Bell Labs was the research arm. Many people bought licenses and started companies to make transistors. One of them was David Bakalar, a Ph.D who had worked at Bell Labs, and he started Transitron with his brother. He recruited Nick DeWolf from General Electric, along with many folks from Europe. Nick had been focused on testing transistors at GE because two-thirds of every batch were trash. The technology was still in its infancy; no manufacturing or test equipment existed so you had to build your own. The biggest market was the military and there was plenty of money behind it thanks to Sputnik. So any type of engineer was in demand, but one who could implement statistical process control was particularly valuable to Transitron. Bob figured he had nothing to lose and it could at least help with the car payments on his ’57 Ford convertible until he went into the Navy. He took the job. He immediately started analyzing the data; eventually, his work only took a few hours a day, so the operators taught him how to take measurements using a Tektronix 575 curve tracer. Then they invited him out on the manufacturing line to see how the transistors were made. He was doing experiments and learning quickly; learning what they weren’t yet teaching in school. When the Polaris transistor was ready for production, Pierre Lamond offered him the job of production engineer. Instead, he opted for a role they called ‘control engineer’ on their diode line. Beyond just diodes, he went to all the other departments to see if his knowledge of statistics was of value to them in any way. Obviously, he never went back to Polaroid. After three months he gave up on the Navy. The Lesson David Bakalar was 34 when Transitron went public in 1959. Together, he and his brother made over $34M that day. He was a prolific recruiter and attracted some amazing talent to Transitron, but he didn’t share the wealth. They were creating a brand new industry while doing meaningful work for the Cold War effort, but management didn’t value the workers. Within a year after hiring Bob, Nick DeWolf left to start his own company, Teradyne, making semiconductor test equipment. Nick said, “…the people who owned the company made the mistake of not sharing the stock with any of the employees… We were all working for love. Then they went public. And they got on the front cover of Fortune Magazine… So the stock shot up. The people who owned the company [the two Bakalar brothers], were suddenly possessors of a big pile of dough… And they grew to believe that it was all just their own personal efforts that had made the company work.” People have a natural sense of fairness. They know when they aren’t being valued or respected. As Nick said, “We were all working for love,” but they weren’t treated fairly in return. Hardly anyone remembers Transitron today. Those who do remember it, think of it as a place that had a lot of talent but couldn’t manage to retain any of it. They clearly thought of engineering talent as a commodity. Within the electronics industry, many well-known people passed through Transitron: Wilf Corrigan (co-founder/CEO of LSI Logic)Nick DeWolf (founder of Teredyne, Inc.)Dave Fullagar (inventor of the uA741 op-amp, co-founder of Maxim Integrated)Jim Diller (Fairchild engr mgr, National Semiconductor, founder of PMC-Sierra)Hans Camenzind (inventor of the 555 timer at Signetics)Pierre Lamond (Fairchild mfg exec, co-founder* of National Semi., venture capitalist)Tom Longo (founder of Performance Semiconductor)Les Vadasz (part of the founding team of Intel)George Wells (former CEO of Exar)Value and Meaning It’s not all about the money, though. Money is feedback, but meaning propels you forward. Feedback can be positive or, in these cases where the money seemed unfair, feedback can be negative. According to Jim Collins in Good to Great, if you have the right people, they will be self-motivated. That motivation or sense of meaning is personal; you have to have it already. For some, it may be patriotic duty, the creative freedom to invent something, or perfecting your craft. “A lot of things make me angry. I’m obsessed with everybody doing the best job they can. I mean, we’re talking about survival,” Bob said in a 1991 interview. As his career progressed, he found ways of aligning everyone to the strategy and doing the best they could. Finding the right level of pressure wasn’t always easy. Whether it was engineering, customer service, or finance, everyone needs to do their best. Respect the value and meaning in every role; that seed was planted early on. For Bob, it wasn’t the money that caused him to leave Transitron. He could see that the company didn’t value the talented engineers around him. He could also see that the germanium technology was quickly being passed by new silicon technology developed by Fairchild Semiconductor in California. But that’s another story. In 1963, he and his new wife moved to California. I’ve worked for three companies that were at the top of the industry; they all made a fatal mistake. — Bob Swanson, Feb. 2021, referring to Transitron, Fairchild, and National Semiconductor

Reposting an article written by a co-worker for Microwave Journal. If I'm violating some copyright, I'll take it down. Source: microwavejournal.com

February 4, 2020

John Cowles, Senior Director of Engineering & Technology, Analog Devices, Norwood, MA

Barrie Gilbert was unconventional.

The movers trucked my boxes from the former Linear Technology buildings in Milpitas across town to the new Analog Devices campus in Santa Clara. It's not far; it's just a new building. There was no closing ceremony. It's not like this was hallowed ground. It was just another Silicon Valley tilt-up.

The term "tilt-up" refers to the way many buildings are made here. They pour a concrete slab. Then they pour concrete walls horizontally and tilt them up to form the sides of the building.

It was March 15, 2017. I spent a few hours with Bob Swanson on his last day as Executive Chairman of the company he founded with Bob Dobkin over 35 years ago, Linear Technology Corporation. A few days earlier, the legal transaction concluded with Analog Devices acquiring the company. Bob had packed up what he wanted from his office and graciously let me sort through the rest and keep it for historical purposes. As I work through it, I will post more articles here.
4

The quintessential analog component is the operational amplifier. And the overwhelming majority of modern integrated circuits are made on CMOS processes of ever-shrinking geometry. But early on, the “low and slow” use of MOS transistors in an analog circuit was ridiculed. “Complementary” was, well, not complimentary.
1

Dr. Hung Chang Lin is referenced in many analog circuit patents and papers for several ideas: single-diode compensation for temperature stabilization in bipolar transistors, the use of a diode-connected transistor instead of an actual diode to improve matching, and the invention of lateral transistors.
1

I read the following blog post last week. I’ve met David Manners a few times during my marketing days and always looked forward to his visits. Quite often, his posts lead me to do a little more research, as is the case today.
2

Sometimes you get off track temporarily, and sometimes you change course completely. I have many interests and Analog Footsteps is only one of them. I’ve been off track lately. Cycling is another interest and usually there is almost no crossover between the two. Cycling picked up for me in the fall and consumed a bit more of my free time. So did another chip veteran’s story. If I were writing “Digital Footsteps” then the two tracks would have been aligned perfectly.
3

I grew up in Michigan as a "car guy" with virtually no awareness of what what happening in Silicon Valley ...when everything was happening in Silicon Valley.  Detroit was the place to be.  And I went, only to find out that I didn't want to be there because it wasn't what it used to be.  After accepting a generous package to voluntarily quit General Motors, I moved to Silicon Valley.
1
Labels
Blog Archive
About Me
About Me
My Photo
I have always been fascinated by the history of whatever I'm doing: my job, the music I listen to, where I live. I love the old photos, I love finding the missing pieces of the puzzle, I love the personal stories.
Loading
Powered by Blogger.