Federico Faggin | Code Like It's 198x

Late one evening at the end of December 1970, Federico Faggin put the probes of a tester on the first wafers of a chip he had been designing for most of the year, and nothing happened. Die after die, wafer after wafer, there was no signal at all. Under the microscope he found the cause: the factory had left out one masking step, the buried-contact layer that his design relied on, so “maybe 60 percent of the gates were not connected”. Three weeks later, in the second half of January 1971, the next wafers came back. By half past three in the morning every test he had prepared had passed. The chip was the Intel 4004, and Faggin called that night “one of the most elating moments of my life”. Five years later he had left Intel, founded his own company, and was testing the first Z80.

Fast facts

Born: Vicenza, Italy, 1941.
Education: physics degree, University of Padua.
Fairchild, 1968: developed the silicon-gate process for MOS chips.
Intel, 1970–1974: led the design of the 4004 and its family; supervised the 8008; conceived the 8080.
Zilog: founded with Ralph Ungermann after leaving Intel on 31 October 1974; the Z80, designed with Masatoshi Shima, sampled in 1976.
Later: Cygnet Technologies (1982), Synaptics (1986, with Carver Mead), Foveon.
Honours: Kyoto Prize (1997) and the 2009 National Medal of Technology and Innovation, both shared with the 4004’s other designers.

From Olivetti to Fairchild

Faggin grew up in Vicenza, building model aeroplanes from his own drawings: “the complete process, from imagining a product to designing it, building it, and flying it.” After technical school he worked for Olivetti, where he designed and built a small experimental computer. He then read physics at the University of Padua, finishing, he said, in four years instead of the usual six.

He joined SGS-Fairchild in Italy in 1967 and in 1968 went to Fairchild Semiconductor in California on what was meant to be a six-month exchange. He stayed. His task there was the silicon-gate process. In an MOS transistor the gate is the electrode that switches the current, and it had been made of aluminium, laid down after the transistor’s source and drain. Faggin’s process made the gate of polycrystalline silicon instead and used it to define where the source and drain went, so the three lined up by themselves. The result, he said, was transistors “three to five times faster” and smaller. He added a step his supervisor did not want him to try, a “buried contact” that joined the silicon layer directly to the chip without a metal link, and it roughly doubled the density of random logic: “In other words, half the cost.” His first silicon-gate product, the Fairchild 3708, went on sale at the end of 1968. Fairchild sent him to present the process at a conference in Washington in October 1968, which put it in the public domain.

The 4004

Faggin joined Intel on 3 April 1970. The interviewer, Les Vadasz, would not tell him what the project was. It was the “Busicom project”: a set of chips for a calculator made by the Japanese company Busicom. Ted Hoff and Stan Mazor at Intel had proposed replacing Busicom’s many custom chips with four, one of them a general-purpose processor that could be programmed, and Busicom’s Masatoshi Shima worked with them. Nobody had yet designed a chip like it in silicon gate, and Faggin had to invent the design method as he went. From April 1970 to March 1971 he was “almost 100% occupied” with the four chips: the 4001 ROM, the 4002 RAM, the 4003 shift register and the 4004 processor, with about 2,300 transistors. Intel later won the right to sell the chips to other customers and announced them in November 1971.

Intel itself calls the 4004 “the first general-purpose microprocessor” and credits Hoff, Faggin, who “took over design leadership from Hoff”, Mazor and Shima. Rival claims exist for other early chips. Faggin pointed out in his oral history that Texas Instruments announced a single-chip processor earlier in 1971, but said that chip never worked or went into production.

The 8008 and the 8080

Faggin’s group finished the 8-bit 8008, designed by Hal Feeney and originally commissioned by the terminal maker Datapoint, and Intel introduced it in April 1972. Faggin then proposed a better 8-bit processor, the 8080, at the end of 1971, after a trip to Europe showed him what customers wanted. Intel’s management took six or seven months to agree. The architecture was his, with “a couple of instructions, one or two, that Stan Mazor suggested”. To design it he hired Masatoshi Shima from Japan. Work began in mid-1972, the first silicon worked at the end of 1973, and Intel announced the 8080 in the spring of 1974. Each generation, he observed, roughly doubled the transistor count: about 2,300 for the 4004, 4,500 to 5,000 for the 8080, and about 10,000 for the Z80.

Zilog and the Z80

Faggin was unhappy at Intel. He felt that Vadasz had later taken the credit for the buried contact, and he saw little financial reason to stay. One night in 1974 he took Ralph Ungermann, one of his managers, out for a drink and proposed starting a company. His last day at Intel was Halloween, 31 October 1974. “I was not sharing my frustrations with them,” he said of his staff, who were upset that he left.

The new company was Zilog. Faggin’s first idea was a single-chip microcontroller, which later became the Z8, but he set it aside for a processor that could run the 8080’s programs and do more. Exxon Enterprises, the oil company’s venture arm, approached them and in June 1975 invested half a million dollars. Shima joined in 1975 to design what was briefly called the “Super-80”. The accounts of when differ: Faggin’s oral history puts it in April, after Exxon’s verbal commitment, while in a 2007 panel with Shima, Faggin said the architecture was mostly done “before Shima came in February of 1975”, and Shima dated the project’s kick-off to “the middle of February, 1975”. Faggin set design rules that would let the chip be made at Synertek, at Mostek and, if need be, at Intel. The first Z80 silicon arrived in early January 1976 with only a few mistakes. Zilog showed it at a local show in March and made the formal announcement in May 1976, with a double-page advertisement headed “The Battle of the 80’s”, comparing the Z80 with the 8080. Mostek became the second source, which made it safe for manufacturers to design the new company’s chip into their products.

BYTE reported in August 1976 that the Z80 ran all 78 of the 8080’s instructions as part of its own 158, needed only a single 5-volt supply and a single-phase clock, and refreshed dynamic memory by itself. Those savings in support chips are why it went into the TRS-80, Sinclair’s ZX81 and ZX Spectrum, the Amstrad CPC and the MSX machines. A 1980 guide to the TRS-80 introduced Faggin to its readers as “creator of the 8008, 8080, and Z-80 microprocessors” and “founder and President of Zilog”.

Zilog never became the Intel rival Faggin hoped for. He fell out with Ungermann around 1978 and 1979, and later said that having Exxon as the only investor kept customers such as IBM away, because Exxon was planning to compete with them. Worn down, he resigned as chief executive at the end of 1980, taking a post at Exxon that he soon left too.

Cygnet, Synaptics and Foveon

In 1982 Faggin started Cygnet Technologies, which made the Communication CoSystem, a box that linked a personal computer and a telephone line for voice and data. It did not succeed and was sold in 1986. That year he and the Caltech professor Carver Mead founded Synaptics to build chips modelled on the brain’s neural networks. Its first product, shown in 1992, was the I-1000, a neural-network chip that read the printed numbers on bank cheques. Out of that work came a touchpad for laptop computers, which Synaptics introduced at the Comdex show in Las Vegas in November 1994. In 1997 Synaptics and National Semiconductor set up Foveon to develop Mead’s image sensors, and Faggin later ran Foveon himself.

His work on artificial neural networks led him to ask whether a machine could be conscious. In 2011, Wikipedia reports, he and his wife founded the Federico and Elvia Faggin Foundation to support research into consciousness, and in 2024 he set out his own theory in the book Irreducible: Consciousness, Life, Computers, and Human Nature.

In 1997 the Kyoto Prize for Advanced Technology went to the “group of four engineers” who developed the 4004: Faggin, Hoff, Mazor and Shima. In 2010 Faggin, Hoff and Mazor received the 2009 National Medal of Technology and Innovation from the United States, “for the conception, design and application of the first microprocessor”.

Sources

Computer History Museum, “Oral History of Federico Faggin”, interviewed by Gardner Hendrie, 22 September 2004, 13 December 2004 and 3 March 2005, reference X2941.2005 (web): his childhood and model planes, Olivetti, Padua, SGS-Fairchild and Fairchild, the silicon-gate process, the buried contact, the 3708, joining Intel on 3 April 1970, the Busicom project and the 4000 family, the first 4004 wafers, the 8008 and 8080, transistor counts, leaving Intel on 31 October 1974, Ralph Ungermann, Exxon Enterprises, Shima, the Z80’s first silicon and announcement, Mostek, Cygnet, Synaptics and the touchpad, Foveon. https://archive.computerhistory.org/resources/text/Oral_History/Faggin_Federico/Faggin_Federico_1_2_3.oral_history.2004.102658025.pdf
Computer History Museum, “Zilog Oral History Panel on the Founding of the Company and the Development of the Z80 Microprocessor”, with Federico Faggin, Masatoshi Shima and Ralph Ungermann, moderated by Michael Slater, 27 April 2007, reference X4023.2007 (web): p. 4 (Faggin, “before Shima came in February of 1975”), p. 6 (Shima, the kick-off in “the middle of February, 1975”). https://archive.computerhistory.org/resources/text/Oral_History/Zilog_Z80/102658073.05.01.pdf
Intel, “Announcing a New Era of Integrated Electronics” (web, secondary): Busicom, the four-chip design, the roles of Hoff, Faggin, Mazor and Shima, “the first general-purpose microprocessor”. https://www.intel.com/content/www/us/en/history/virtual-vault/articles/the-intel-4004.html
Burt Hashizume, “Microprocessor Update: Zilog Z80”, BYTE, August 1976, pp. 34–38: the 8080’s 78 instructions within the Z80’s 158; single supply and clock; built-in refresh.
Robert M. Richardson, Disassembled Handbook for TRS-80, volume 2 (Richcraft Engineering, 1980), introduction: Faggin as “creator of the 8008, 8080, and Z-80” and president of Zilog.
The Home Computer Advanced Course issue 11 (Orbis), p. 220: the Z80 designed by Faggin and Shima to be compatible with the 8080.
MC Microcomputer (Italy) issue 121, September 1992, pp. 243–245: Synaptics, founded in 1986 by Faggin and Carver Mead; the I-1000 cheque-reading chip; Faggin’s physics degree from Padua.
Kyoto Prize, 1997 laureate in Advanced Technology: Federico Faggin (web, secondary): birth year and place, career dates, the “group of four engineers”. https://www.kyotoprize.org/en/laureates/federico_faggin/
National Science and Technology Medals Foundation, “Federico Faggin” (web, secondary): the 2009 National Medal of Technology and Innovation with Marcian Hoff and Stan Mazor, its citation, Olivetti and the silicon-gate work. https://nationalmedals.org/laureate/federico-faggin/
Wikipedia, “Federico Faggin” (web, secondary): the name of Cygnet’s Communication CoSystem; his 2009 medal presented in 2010; the Federico and Elvia Faggin Foundation (2011); Irreducible (Essentia Books, 2024). https://en.wikipedia.org/wiki/Federico_Faggin