Feb 23, 2017

A CMOS Op Amp Story - Part 1

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. So, how did we get to the CMOS op amp? And how would the Vegas strip and TV test patterns play a part in the story? Read on!
Wikipedia says that Frank Wanlass patented CMOS in 1963 while at Fairchild. It doesn’t say he invented it. Much like Alexander Graham Bell in 1876, he got to the patent office first. Coincidentally, also in 1876, W. G. Adams and R. E. Day published a paper in the Proceedings of the Royal Society entitled, “The action of light on Selenium.” It was the first observation of surface effects and described what we know of as the field effect transistor.
The footsteps for this story go back and forth between RCA on the east coast, Silicon Valley on the west coast and Utah, of all places.
Why Utah? I’ll save that for “part 2” of the story. As well as the Silicon Valley folks.
Why RCA?
Briefly, during World War I, the U.S. government commandeered all the radio companies and their patents for the war effort. After the war, they cut a deal with General Electric to take the assets from the American Marconi company and create the Radio Corporation of America (RCA) as well as the National Broadcast Company (NBC). Also part of the patent pool were GE, AT&T, Westinghouse and the United Fruit Company (yes, bananas). RCA bought the company that made the Victrola and moved into television in the 1930s. The TV tube research and development complemented their vacuum tube development in support of their own end-to-end television business. RCA Laboratories (similar, but not as famous as Bell Laboratories) was set up in Princeton, NJ. RCA was second only to Bell Labs in the number of transistor related patents during the 1940s and 1950s, indicating a strong commitment to basic semiconductor research.
Sarnoff set an ambitious vision for RCA and the lab, part of which included transistors. At RCA, Jerome Kurshan replicated the Bell Labs point contact transistor less than a month after their announcement. (Second place.) Prior to the consent decree, RCA tried to design around the Bell patents and by 1952 had solved the production issues for point-contact transistors. Under Kurshan’s guidance, George Sziklai developed “complementary” circuits in the early 1950s with H. C. Lin, long before MOS or even integrated circuits - but well after the 1876 (Adams and Day) or 1925 (Lilianfeld) “field effect” disclosures. J. Torkel Wallmark started a project assembling several amplifiers, switches and passives to make a prototype digital logic device in 1957 - but Kilby had his “integrated circuit” working first. Nobody remembers who came in second. No matter how you look at it, RCA made a blunder by not putting “complementary” and “MOS” together. Wanlass got the patent first, but Fairchild did nothing with it for several years. For the record, in 1960 Karl Zaininger and Charles Meuller fabricated a MOS transistor at RCA and then Fred Heiman and Steven Hofstein followed in 1962 with an experimental 16-transistor CMOS device at RCA. (Arguably first, but forgotten.) But it would be many years before all of the stability and production problems were resolved. So, at this point, RCA was in the lead - but CMOS was not yet practical.
As with the whole industry at that time, government contracts for the military and space programs funded and directed much of the development. Almost everyone felt that CMOS was too difficult and had too small of a market. Bipolar logic ruled. Straight NMOS dominated elsewhere. Even RCA wanted to focus elsewhere. But, Air Force contracts forced RCA to continue efforts on CMOS despite their problems and internal hostility toward it.
George Rotsky wrote a nice piece for EETimes about the struggle with making CMOS, most of which is probably true. I’ll paraphrase. In 1959, Steven Hofstein and Frederic Heiman joined RCA's David Sarnoff Research Center as part of a work/study program with Princeton. The group leader of the Electron Devices Research Laboratory, Tom Stanley, offered a choice.
"Follow everyone who are working on the junction field-effect transistor that Shockley invented in 1952. If you accomplish anything, you will end up working as a technician for some senior engineer. Or, work on the MOS transistor which everyone says will never work. There are problems with surface states, a problem Bardeen uncovered at Bell Labs. If you succeed, you'll be a hero. If you fail, so what?"
The initial focus was to find a better way to clean the silicon before growing the oxide. There seemed to be a million cleaning recipes, Hofstein recalls, none of which did any good. He then remembered that, to improve surfaces, metallurgists annealed metals in a hydrogen atmosphere. He tried it. Success! It was an exhilarating moment. Instead of cleaning the silicon before oxidizing, the trick was to anneal after oxidizing. Hofstein and Heiman were in inventor's heaven. Unfortunately, the MOS transistors weren't stable.
Meanwhile, Fairchild had discovered that sodium makes silicon unstable, but didn't know how to overcome surface states.
Along about 1965, Hofstein, with the blessings of the IEEE, organized the Silicon Interface Specialist Conference. He chose Las Vegas for the location, unknowingly sowing the seed for Las Vegas becoming a major site for technical conferences.
Rotsky paints this scene: Hofstein and Andy Grove are in the Stardust swimming pool, possibly chatting about the weather and the Las Vegas ladies. The conversation may have shifted to MOS transistors. Grove may have said that Fairchild had spent more than a million dollars tackling the surface-state problem. And Hofstein may have said that RCA had spent a similar sum on the instability problem. No one knows if or what was said in the pool, but it's possible that Hofstein said, "hydrogen," and Grove said, "sodium." Two million dollars and two words. Within a few weeks, the industry benefited from two good sources of MOS field-effect transistors and, not much later, from a growing industry that made larger and larger MOS ICs.
The Air Force computer system used RCA CMOS logic and memory circuits in 1965. In 1968, RCA introduced their CD4000 series logic family. Finally, they were first! Although they were a commercial success, it did not endure. In addition to government contracts, RCA’s solid state developments were disproportionately focused on television. As early as 1952, RCA had a prototype all-electronic television. And it was only in the late 1950s that RCA entered the merchant semiconductor market, prior to that it was all internal consumption for television and military contracts.
My research into CMOS pointed in several directions: CMOS logic, CMOS memories and CMOS image sensors. And these paths crossed from time to time. RCA, for example, invented the TFT (Paul Weimer, 1961) and did a lot of development of CMOS image sensors. They built the first solid-state camera in 1967 (prior to Bell Labs’ CCD Picturephone, remember that?). Image quality has always been important to RCA even way back in the tube era. I got sidetracked into researching an engineer named Otto Schade. Schade was born in Germany and came to the United States in 1926. He joined the Tube Division of RCA in New Jersey in 1931. His first widely recognized achievement was in 1936 on the electron optics behind the 6L6 beam power tube. He was a brilliant and prolific engineer, most noted for his work on the Modulation Transfer Function (MTF) - a well-known and widely used metric to characterize the resolution of almost any optical or display component as well as the resolution of overall imaging systems. This metric is used to characterize the entire display system as well as the performance of each individual system component from the camera lens to the human eye. This manifested itself in Schade’s high resolution test pattern.  In the course of his career, Otto Schade presented or published more than 30 papers and received 85 patents. He spent the rest of his career at RCA.

Unlike modern Silicon Valley, not only did engineers of that era spend their entire careers at one company but often their children joined the same company. And this was the case for his son. According to an interview with the Society of Information Display, sometimes Otto, Jr., had to cancel dates to serve as the test audience for his father’s lectures, and the planned building of a model-train layout in the cellar was scrapped in favor of a home darkroom to further study image quality. Few sons in history have been told by their father: “Sit down, I want to get the sine-wave response of your eye.”
Turns out, Otto, Jr., also had an impressive career at RCA with many, many patents. And some of his work involved CMOS analog!
I had the pleasure of talking to Otto Schade, Jr., on the phone. He said he joined the Solid State division (semiconductor division) in 1966, coming over from the “tube” division where he had been working on broadcast TV tubes. At the time, the digital group (developing the CD4000 family of digital logic devices) and the analog group were separated. He said, “the analog guys talked to the logic guys as seldomly as possible.” Otto, Jr., was constantly compared to his father and it made things difficult at times. Within the Solid State division, he struggled because his background was in mechanical and tube technologies, not semiconductors. He said, “It wasn’t much fun; there was so damn much learning to do.”
Otto gives much credit to Murray Polinsky for developing the analog version of the CMOS process. The process used by the CD4000 logic family was prone to ESD damage. The analog process developed by Polinsky included the necessary clamps and protection circuits to help these devices survive. RCA is not very well known as a semiconductor company, not even in those days. And the “RCA Review” probably didn’t reach as many people as the Bell Labs Technical Journal but there is a lot of good reading in there. But in that spirit, they tried to spread the information as much as possible. Otto had attended many ISSCC conferences and had attended some of the after-hours sessions where competitors would freely talk about the latest technology. They would say, “Turn off your recording devices and let’s just talk.” Even though he was well aware of Bob Widlar, Bob Dobkin, Dave Fullagar and many others, he didn’t really know them. They still felt isolated back in New Jersey.
Otto was working for Dick Sanquini (group leader, linear integrated circuit development) and the Vice President of the division was Bernie Vonderschmitt. In those days, management was able to talk to the engineers; they had their respect because they had also been engineers. Bypassing the normal organization because he knew Otto, Sr., Bernie asked Jr., “Can’t we make a circuit that has both bipolar and CMOS on one chip?” Schade’s team built a FET-input, 10V op amp with a CMOS output stage. And with that, we have the birth of the CA3130, arguably the first “CMOS op amp”.
That first op amp development was like the old saying, “a camel is a horse designed by a committee”. Technically, there are bipolar transistors in the chip, but perhaps that’s splitting hairs. It was introduced in 1974 and marketed as a CMOS op amp for use in predominantly CMOS systems. It would be several years before anything like it hit the market. I have yet to find anything particularly interesting about the CA3130 or its design but you can still buy them. It was reasonably popular in its day but not particularly precise.
I asked Bob Dobkin about the CA3130. He said it was really useful in “Ideas for Design” - as a noise generator! Coincidentally, I’ll remind you that Dobkin once considered joining General Electric to start up their integrated circuit group. He would have inherited this part.
Bob Pease made a vague claim to a CMOS op amp design in this same timeframe, albeit as part of a larger IC. From April 1999, “What’s All This CD4007 Stuff, Anyhow?”:
“I am slow, dumb, STUPIDO. However, the sudden cessation of stupidity is worthy of some mention. I knew about CMOS circuits in the 1970s. I even designed a CMOS operational amplifier back in '74. It used about 15 transistors to run on ±5-V supplies and provide a gain of 10,000. At the time, several engineers looked at my two-stage, cascoded differential amplifier and asked, ‘Why do you make life so complicated? Why don't you just use TWO transistors of a CMOS inverter for an op amp, like everybody else does?’ … This op amp went into the Teledyne Semiconductor 9400 analog to-digital converter with 8-, 10-, or 12-bit resolution.”And Pease also suggested: “If you want to read about an analog CMOS circuit, you should obviously read AN-88 in NSC's CMOS Databook. (Hey, don't waste your time. It mostly tells you how you can use a 74C04 as an op amp.)”

In 1986, RCA Solid State was acquired by General Electric and Jack Welch. The combined RCA/GE Solid State division was sold off to Harris Semiconductor in 1988 and combined with the original Intersil which they had acquired at about the same time. Otto noted that the lab in Princeton, NJ, (RCA Laboratories) was transferred to Stanford Research International, so they became “SRI East”.
Name Dropping
Dr. H. C. “Jimmy” Lin worked at RCA in the late 1940s and 1050s, working on linear applications with discrete bipolar transistors. He would be instrumental in the development of integrated circuits at RCA and, later, Westinghouse. He is referenced many times by Bob Widlar and most other great analog designers of the 1960s… on the shoulders of giants.
Dick Sanquini worked at National Semiconductor, initially as vice president of microprocessors, microcontrollers, and peripherals with subsequent roles as senior vice president of the consumer and commercial division and as the company's chief technology officer. Mr. Sanquini's career began in 1959 at RCA's Solid State Electronics Division, where he designed the company's first commercially available integrated circuits, and eventually served as general manager and director of memories and microprocessors. Mr. Sanquini holds a BSEE degree from the Milwaukee School of Engineering, Wisconsin.
Bernie Vonderschmitt worked at RCA for 34 years and was involved in an early industry rivalry over the design of color television. In 1953, Mr. Vonderschmitt was picked to head the development of a color television project at the company. Mr. Vonderschmitt headed the RCA solid-state division from 1972 to 1979. At that point, he moved to Silicon Valley to lead the component division of Zilog. At the age of 60, Vonderschmitt and two partners founded Xilinx.
Robert S. Pepper, Ph.D., served as a Vice President and General Manager of the Solid State division of RCA Corporation from 1979 to 1984. He had spent over 15 years in the semiconductor industry, notably as a Vice President and General Manager of the Semiconductor division at Analog Devices, Inc. Dr. Pepper joined Level One Communications in 1986 as the Chief Executive Officer and President. Dr. Pepper holds a BSEE, MSEE, and Ph.D. in Electrical Engineering from the University of California at Berkeley.

NOTES
Wikipedia
Phone call with Otto Schade, Jr, November 11, 2016.
“Otto Schade – A Pioneer with an Insight into Image Quality” Society of Information Society, staff article, http://informationdisplay.org/IDArchive/2006/January/OttoSchadeAPioneerwithanInsightintoImage.aspx

“Current Sources and Voltage References: A Design Reference for Electronics” By Linden T. Harrison

“What's All This CD4007 Stuff, Anyhow?” by Bob Pease, Electronic Design, Apr 5, 1999 http://electronicdesign.com/analog/whats-all-cd4007-stuff-anyhow