Dec 29, 2013

Analog and Molecular Science

The term “molecular science” conjures up many images but very few that are related to analog semiconductor circuits. The term is most closely connected to nanotechnology in the electronics field. In the 1980s, Forrest Carter promoted the idea of a single-molecule logic gate. Carter was a chemist from Caltech, studying there in the days of Linus Pauling and Richard Feynman. Carter joined Westinghouse in 1957 and then moved to the Naval Research Laboratory in 1964. Decades later at NRL, he would use the term “molecular electronics” – a term that was in use at Westinghouse during his time there but for something somewhat different.

To begin to trace our analog footsteps, we need to go back to the Massachusetts Institute of Technology (MIT) and World War II. In 1936, a German physicist name Arthur von Hippel joined the MIT electrical engineering faculty and worked during World War II as Director of MIT’s Laboratory of Insulation Research. He articulated a vision of highly integrated circuits:

Instead of taking prefabricated materials and trying to devise engineering applications consistent with their macroscopic properties, one builds materials from their atoms and molecules for the purpose at hand ... . He can play chess with elementary particles according to prescribed rules until new engineering solutions become apparent.
In 1956, von Hippel held a 10-day summer course on molecular engineering at MIT; and in 1959 he published a textbook entitled Molecular Science and Molecular Engineering, co-authored with numerous researchers from military and corporate laboratories – including the Air Research and Development Command (ARDC) and Westinghouse.

The concept that evolved was in contrast to the typical semiconductor processing – pulling a single crystal ingot, slicing off a wafer and processing up. Instead, the idea was to ‘pull a pnp or npn crystal strip directly from the melt’, or to achieve the desired characteristics by ‘diffusing suitable impurities upon the newly formed dendrite surfaces’. Westinghouse had started development in this area but in early 1957 the Air Force was resistant. Then came Sputnik in October of 1957. Now the Air Force was interested and the money was there. Soon “molecular science” and “molecular electronics” were in the press and technical papers.

From Popular Electronics, April 1960:
A dendritic ribbon to which tiny multiple-function systems have been automatically attached is [?]. Here a series of multivibrators has been created directly on the dendrite. The individual circuits need only be clipped apart and leads attached. Soon, complete amplifier circuits will be produced the same way. The dendrite ribbon will be snipped into different lengths to give amplifiers of different gains - the longer the strip, the greater the amplification!
Eventually, engineers hope to "grow" complex electronic equipment - complete receivers, for example - automatically and continuously from a pool of semiconductor material. These receivers are still far in the future, but they would be unbelievably cheap and trouble - free by today's standards. Because of the low power consumption of molecular electronic function blocks, a single battery would last for years.
Molecular electronic devices will open up exciting new fields with their combination of high performance, small size, and low cost. For example, the wrist radio - a la Dick Tracy - will become common-place. The personal telephone - a tiny gadget to strap on your wrist or carry in your pocket - will become possible. With it, you will be able to call anybody in the world who is similarly equipped.
A flat-screen TV set that hangs on your wall like a picture will become a reality. Rapid advances in electroluminescence have already come close to making practical a screen only a fraction of an inch thick. Molecular electronics will make it possible to pack the rest of the TV circuitry into a hollow corner of the frame!

In the early days of the Noyce and Kilby versions of integrated circuits, “molecular electronics” was often used interchangeably. With this as a back-drop, a tiny semiconductor start-up changed its name from Electro Radiation, Inc to Molecular Science Corporation. They were early pioneers in the custom logic business with RTL, DTL and DTL-TTL blocks that could be configured into larger functions (an ASIC). The company was founded by ex-Fairchild processing expert James Nall.

Nall hadn’t been at Fairchild very long. In 1957, he and Jay Lathrop patented photolithographic techniques while at the U.S. Army’s Diamond Ordnance Fuse Laboratories. In 1959 Lathrop joined Texas Instruments, working for Jack Kilby, and Jay Last hired Nall at Fairchild Semiconductor. But by 1962 he had moved on. In 1963 they shortened the name to Molectro Corporation.

For reasons I don’t understand, Molectro attracted two more Fairchild defectors in 1965 – Bob Widlar and Dave Talbert. Likely it was due to the precision processing that was developed under Nall’s guidance. Talbert developed an EPI reactor and soon Widlar was developing what would become the LM100 – the first linear voltage regulator. Imagine what could have been!

But Molectro was in horrible financial shape and soon entered Chapter XI. In 1966, Peter Sprague brought National Semiconductor in from Danbury, Connecticut, to rescue them, financially. Molectro at the time had something like 30 employees and occupied a facility on San Ysidro Way in Santa Clara, CA. National had about 600 employees in Danbury and more importantly, was profitable. Peter Sprague brought in Charlie Sporck (and Fred Bialek, Pierre Lamond, Floyd Kvamme and Roger Smullen) from Fairchild to take over management of National. Charlie promptly fired half of the Danbury employees and effectively moved “National” to the Molectro site as its new headquarters. From an analog perspective, Widlar brought out the LM100 and National became and analog powerhouse.

National acquired Widlar and Talbert when it took over Molectro, formerly Molecular Science Corp – named for a term that once meant something quite different than it does today.

NOTES:
2. The Long History of Molecular Electronics: Microelectronics Origins of Nanotechnology, Hyungsub Choi and Cyrus C.M. Mody, 2009, http://sss.sagepub.com 
3. History of Semiconductor Engineering, Bo Lojek
4. Spinoff, Charlie Sporck with Richard Molay, 2001

Dec 27, 2013

Federal Telegraph, the Birthplace of Silicon Valley

A rapid scan of historical highlights might lead a reader to this view of events:

  1. Alexander Graham Bell invents the telephone.
  2. Thomas Edison invents the light bulb, and later notices the “Edison Effect” when you insert an extra element into the bulb.
  3. Lee De Forest takes that idea and invents the triode vacuum tube, which he called the Audion.
  4. Silicon Valley was started in Dave Packard’s garage.
  5. Shockley invents the transistor at Bell Labs, starts his own company in Silicon Valley until the “traitorous eight” defect and start Fairchild.

A very digital (sampled) view of history. But this is Analog Footsteps, so we take an analog view of history – in the sense of a continuum where something interesting might have happened right before or right after these iconic events. That’s why I want to talk about Federal Telegraph. I touched briefly on Western Electric’s role in #1, above. Historians question #2 since De Forest couldn’t even explain how it worked – but at one point he worked for Federal Telegraph! Federal Telegraph was a company that challenges #4. I agree with most of #5, except the first part.

Until the invention of the telephone, the telegraph was the main form of long-range, electrical communication. But suddenly, it was conceivable to skip the manual coding and decoding and hear the voice of someone hundreds of miles away. We just needed the technology. Due to their size, complexity and cost, alternator-transmitters were mostly employed for long-range radiotelegraphy, and rarely used for audio transmissions. But another developing continuous-wave technology also showed promise -- the arc-transmitter, which had been perfected beginning in 1902 by Valdemar Poulsen of Denmark. When I learned this, I was a bit surprised because I expected it to have been invented on the East Coast of the United States. Part I’m part Danish, so this makes me happy.

Poulsen presented a paper on the topic in Saint Louis in 1904. One person attending the conference was Lee De Forest, who would spend many years trying to develop arc-transmitters for audio transmissions. Depending on what you read, De Forest was either a creative genius with bad luck in business or an outright conman. Over and over, he built companies and great fortunes only to lose them and his reputation by over-promising and under-delivering (and possibly taking credit for things he did not invent or even understand). In 1911, De Forest was able to temporarily find employment at the one company that actually was operating a successful arc-transmitter radiotelegraph service, the Federal Telegraph Company.

Backing up a bit for a fascinating footnote in history, in 1903 a 15 year old boy, Francis J. McCarty, demonstrated one of the first audio radio transmissions using his own invention. This was in San Francisco. Yes, he was only 15 years old. Tragically, he was killed in a buggy accident two weeks before his 18th birthday. Cyril F. Elwell was hired by local businessmen to evaluate the commercial potential of the McCarty patents. Elwell was an Australian, a recent graduate of Stanford University and teaching there at the time. He got the system to work for a distance of over one mile but concluded that it was not practical.
Elwell and two associates (but not De Forest) in the early days
Instead, Elwell convinced backers to buy the patent rights for Poulsen’s arc transmitter. He founded the Poulsen Wireless Telephone and Telegraph Company (a California company), then later it reemerged in 1911, as Federal Telegraph Company (an Arizona company for more favorable tax reasons, much like today’s “Delaware companies”), with Elwell as president. Along the way, it was briefly named Wireless Development Company.

Federal Telegraph was not alone. All across the country and also throughout Europe, companies were being founded in the telegraph industry. Some were just transmitters. Some, like Federal Telegraph, owned stations and developed new equipment. It wasn’t long before consolidation occurred. Federal Telegraph was successful and gradually acquired other companies on the west coast. With stations in San Francisco and development being done in Palo Alto with connections to Stanford University, it might rightly claim to be the birthplace of Silicon Valley.
WKZO was the station I grew up with in Kalamazoo, MI

But history taught us that Packard’s garage was the birthplace. Didn’t Fred Terman create an atmosphere at Stanford University which led to the start of H-P and that whole Silicon Valley thing? Yes, Terman was inspired by his time at MIT to foster relations between the University and local industry. Terman’s early interest in radio was due to the proximity of Federal Telegraph – he built a crystal receiver in 1913 (age 14). And Bill Hewlett agreed, "Fred Terman didn't start Silicon Valley; the beginning of Silicon Valley was a supernova." He explained that Lee de Forest, an electronics pioneer in the Palo Alto area, and his work were the supernova. I trust Bill. Although, De Forest was only a visitor in the Palo Alto area; regardless, the work done at Federal Telegraph was pioneering.

De Forest moved to California and worked for Federal Telegraph Company at Palo Alto. According to Rogers and Larsen, in 1912 "De Forest and two fellow researchers for the Federal Telegraph Company…” were able to amplify a housefly's footsteps 120 times. This event was the first time that a vacuum tube had amplified a signal. Also Rogers and Larsen add that, "Lee de Forest had a Stanford University connection; his work was partly financed by Stanford officials and faculty." At Federal Telegraph, De Forest finally made his Audion tube perform as an amplifier and sold it to the telephone company for $50,000. By late 1916, he was gone again – back on the east coast seeking greater fortune. As much as I trust Bill, I suspect that Federal Telegraph’s Stanford-trained engineers contributed more than history recorded. But that’s my own opinion.

So why don’t most people know about Federal Telegraph? What happened to it? Well, history records that in 1917, Federal Telegraph was the junior partner with American Marconi in forming the Pan-American Wireless Telegraph and Telephone Company. In 1927, it was acquired by MacKay which was then acquired by ITT. So it didn’t die, but it didn’t live. In historical terms, it basically disappeared.

NOTES:

1. William Hewlett's quote, Carolyn Tajnai, 1995, author of "Fred Terman, The Father of Silicon Valley"

2. http://www.netvalley.com/frame-right-svhistory-1.htm 

3. Moodys Manual of Railroads and Corporation Securities (Volume 2, Part 2), 1922

4. The Examiner, San Francisco, August 30, 1908

5. United States Early Radio History, Thomas H. White, http://earlyradiohistory.us/sec009.htm

Dec 23, 2013

Einstein's Amplifier

In 1905, Albert Einstein published the groundbreaking paper with the simple relationship: “If a body emits the energy L in the form of radiation, its mass decreases by L/V2.” That rearranges to L = mV2. Eventually we came to call energy E instead of L and the particular velocity of light, V, we now call c. E = mc2. In addition to his full-time job as a patent clerk, he cranked out six more papers in 1906 and ten more in 1907. And in his spare time in 1907, he developed an amplifier with his friends, Conrad and Paul Habicht.

The goal was to develop an instrument to study small electrical fluctuations. He thought it might be useful for experiments with his predictions concerning the equivalence of mass and energy. The concept for the amplifier was electro-mechanical. Two strips of metal acting as capacitor plates would move by each other, one inducing an opposite charge on the other. A series of strips would induce the charge ten times and then transfer to another disc. The process would be repeated until the miniscule charge (realistically, half a millivolt) was large enough to be measurable. Not an amplifier as we know it today, but an electrostatic potential multiplier.

Prototypes were built, tested and refined – mostly with Einstein’s encouragement but not too much involvement. It was called the little machine, Maschinchen. “I am astounded at the lightning speed with which you built the Maschinchen,” Einstein wrote in a letter to the Habichts. By November of 1908 it worked acceptably well. In December of 1911 it was demonstrated at the Berlin Physical Society. Your recollection is correct; the electrostatic potential multiplier was not successful. Paul Habicht built and sold a few – you can see one in the Physical Institute of the University of Tübingen.

It’s quite a stretch to say that Einstein could have been an analog engineer. He wasn’t a builder; he didn’t physically take things apart to see how they worked. He was a much better theorist. His experiments were thought experiments. But for a time, he put some thought towards engineering. The Habichts were the engineers in this story. And with persistence, they made it work. But the market went with a better solution.

NOTES:

“Einstein, His Life and Universe”, Walter Isaacson, 2007

“The Maschinchen”, Hans-Josef Küpper http://www.einstein-website.de/z_biography/maschinchen-e.html
Illustration: Dr. Torsten Hehl, Tübingen

Dec 15, 2013

Forgotten Spin-Offs

Most who are likely readers of this space already know the big names of the Fairchild family tree. But I came across an account of the story written by Don Hoefler for Electronic News in 1968. A lot had happened in the preceding ten years since the traitorous eight defected from Shockley Labs. I’ll quote a few items that were news to me, but the whole article is worth reading.

Bay Area:

“A little-known event of 1954 might have altered the course of semiconductor history, and made Boston’s Bay Area – not San Francisco’s Bay Area – one of the world’s leading producers.
Well-remembered is the triumphal return of Dr. William Shockley, co-inventor of the transistor, to his native Palo Alto in 1955. There with the backing of Beckman Instruments, Inc., he founded Shockley Transistor, which gave rise to the vast Bay Area semiconductor community.

Not so well known is that the year before Shockely was negotiating a similar deal with Raytheon Co. He did join Raytheon as a consultant, but wanted a long-term deal which would guarantee him $1 million over a three-year period. Raytheon was unwilling to make the commitment, and the relationship was terminated after only one month.”
Not Fairchild: 
“The first notable California spin-out which had neither Hughes nor Fairchild antecedents, is Siliconix, Inc., founded in 1962 by Dr. William Hugle and the late Dr. Frances Hugle, from Westinghouse, together with Dr. Richard Lee and Arthur Evans from Texas Instruments. The Hugles left two years later to form Stewart-Warner Microcircuits, Inc.”
Transistor Stories: 
“David Bakalar left Bell Laboratories to form Transitron Electronic Corp. in 1952, which gave rise to four spin-outs of its own. James Hangstefer and upwards of a dozen senior technical men departed in 1960 to form Solid State Products. Crystalonics, now also a part of Teledyne, was formed around a Transitron nucleus headed by W. Frustager. John Royan left in 1957 to form American Power Devices, and Claus Lasch left in May of this year (1968) to form an Italian thyristor company under the sponsorship of Ing. Dino Olivetti.

William Pietenpol left Bell Labs in 1953 to establish a transistor facility at Sylvania Electric Products, where a diode capability already existed. From the Sylvania base, Richard Seed left in 1958 to form Semicon Inc., and later, in 1963, Seed Electronics. Earlier, in 1954, Clair Thornton had left Sylvania to provide the technology required to put Philco Corp. into the business.

General Transistor Corp. was formed in 1954 by Herman Fialkov and a group from Radio Receptor Corp., which had been in the selenium rectifier business since the 1940s. In 1957 a group left GTC, along with some key personnel from RCA, to form Silicon Transistor Corp. General Transistor was merged into General Instrument Corp. in 1960.”
Motorola: 
“Motorola Semiconductor was a minor laboratory effort until it hired Dr. C. Lester Hogan away from the faculty of Harvard University in 1957, the same year Fairchild was formed. Dr. Hogan recruited extensively to build engineering, manufacturing and marketing staff required to make Motorola a factor in the business. Many key people came in from General Electric, with others from Bell Labs, IBM, CBS and TI.

Almost immediately, Motorola began to suffer defections of its own, when a group left in 1957 to form U. S. Semcor, which in turn spun out the short-lived American Micro Devices in 1960.

Donald C. Dickson, Jr. left Motorola in 1960 to form the successful Dickson Electronics. Steven Berck and others left Dickson in 1965 to form General Semiconductor Inc.”
These companies read like a “who’s who” list of “who are they?” Some have no direct connection to Fairchild or Bell Labs. And the rapid-fire spin-off occurred before, during and after the Fairchild spin-off frenzy that always seemed like a unique story. Not only is the story not unique in the semiconductor industry, it is not unique in most industries. But that’s another story.

NOTES:
“Semiconductor Family Tree”, Don C. Hoefler, Electronic News, July 8, 1968 

Dec 11, 2013

Bob Noyce, meet Bob Widlar.

The µA702 is credited as being the first true integrated circuit op amp.  It was developed in secrecy by Bob Widlar and Dave Talbert.  Word got out and eventually Fairchild management decided to support it.  Aware of Texas Instruments' success in analog circuits for the Minuteman II missile program, Bob Noyce wanted to pursue that market.

According to Bo Lojek, Bob Noyce and Tom Bay came to see Widlar about it in May, 1964.  Noyce and Bay, however, had never met Widlar.  Of course, Widlar was aware of who they were.  Bay said Fairchild was going to start a marketing campaign for the part and needed a few things from him.  Widlar said, "what do you mean?"  "We are going to announce the µA702 and we are going to sell it," replied Bay.  Widlar knew it had yield issues and he wasn't done yet.  But this was top management, and he had just met them.  "Fuck You!" Widlar said.

"You are not going to sell these things because you do not know how to sell them.  You guys do not know anything about this part.  Nobody inside this company knows anything about it.  If I give this to you, you will just screw up my work."

Eventually, he relented and the rest is history.

NOTES
History of Semiconductor Engineering, Bo Lojek

Photo, EDN.com and Paul Rako's blog post.