The Last Word in Integrated Logic
Federico Faggin, maker of the microprocessor
Federico Faggin was born on the 1st of December in 1941 in Vicenza Italy. He was the second of four children, and his mother stayed at home to look after the them. In 1943, Federico’s parents moved their family to Isola Vicentina where they’d both been born and raised, and there they lived with Federico’s paternal grandparents. On his grandparents’ property there was a garden, a pub, a carpentry shop that built furniture, and a large shed in which were two threshers leased to local farmers during harvest seasons. While Federico’s family did have radio, running water, and other then-modern luxuries, much of the surrounding area did not, and folks continued as they always had. There was a lot of animal power put to use on the farms, wood fire for heat and cooking, and oil lamps and candles for light.
As a child, Federico liked to chase fireflies and play with wooden swords with neighborhood kids, and he enjoyed the black and white movies on Sundays at the parish hall. One day much like any other, Federico was playing outside and saw a young man holding a model airplane and winding the rubber-band powered propeller. The plane flew and Federico took off running beneath it. He was fascinated and trying to figure out what made it work. Later that same day, he gather what materials he could find, and he tried to replicate what he had seen. He failed in this, but he wasn’t put off the hobby. His third attempt at building a model plane was successful, and the hobby taught young Faggin quite a bit about physics, electronics, design, and troubleshooting.
In Faggin’s last year of high school, he became interested in computers and transistors. These weren’t yet part of any school curriculum as they were quite new, but he read everything he could find that discussed them. Digital microelectronics created a new passion in Faggin, and his interest model airplanes waned.
In the autumn of 1960, Faggin accepted a position as assistant engineer at Olivetti’s Electronics Research and Development Laboratory in Borgolombardo. This is where Olivetti’s early electronic computers were being built. He rented a room in Milan on Vicenza Street which he shared with a former schoolmate who’d also taken a job at Olivetti. The two rode a company bus every day to and from work, and they traveled back to Vicenza by rail on Friday evenings to spend time with their families.
As mentioned, transistors and computers were not taught in schools yet. This is something Olivetti well knew, and thus Faggin had two months of training in those topics after which he was assigned to the Circuits Department. His first project was to assist in the design and the building of a small transistorized electronic computer. Faggin’s also needed to design the interface between the magnetic core memory and the machine’s processor. Faggin’s first boss transferred control of the project to a physicist, Dr. Sibani, when it came time to design the main registers, instruction set, and control unit. Shortly after this change, Dr. Sibani was injured in a car accident. In a meeting, Sibani gave Faggin books on computer architecture and tasked him with studying them. A month later, Faggin designed the rest of the system and began the work of implementing it. This machine used 1000 logic gates made out of germanium transistors, and it used 200 PCBs. I/O was done with a teletype. The computer was completed in 1961.
After this project, Faggin chose to return to school. He wanted a university degree not only for career advancement, but also to gain a more thorough understanding of physics and mathematics. Faggin began his studies at the University of Padua on the 8th of January in 1962. Faggin’s favorite subjects were solid state physics, circuit theory, cybernetics, and mathematics which he studied more than was required of him. In his 3rd year, Faggin met his future wife Elvia, and graduated summa cum laude in 1965.
Following graduation, Faggin was an assistant professor at the university for about a year. He then left for CERES. Dr. Sibani had left Olivetti to start his own company, and Faggin was hoping to marry Elvia and start a family, and he required a bit more money for those goals. The company also suited his interests. CERES was developing thin-film circuits in partnership with General Microelectronics (the first commercial MOS company). Joining CERES, Faggin was sent to Sunnyvale to get a week of training on MOS technology. Faggin loved the trip, and loved the energy of the Bay Area. CERES was, however, quite short lived. It sold to RCA.
Faggin was then recruited by SGS-Fairchild in Agrate Brianza, where he worked in Research and Development. He was tasked with MOS process technology. After developing their MOS manufacturing process from scratch, he went on to develop the company’s first two commercial MOS ICs.
Elvia and Federico were married on the 2nd of September in 1967. In February of 1968, the young couple moved into a small furnished apartment in Mountain View California as Federico had been assigned to the Fairchild R&D laboratory in Palo Alto. Fairchild is notable as being the company at which Jean Horn invented the planar process. This process allows for the simultaneous manufacture of many transistors on the surface of a mono-crystalline silicon wafer. This invention paved the way for the integrated circuit. Prior to this, transistors had to be individually packaged. ICs, like any new technology, were expensive upon introduction, but the price of an individual IC fell rapidly. When Federico arrived at the Fairchild R&D laboratory, he was set to work on MOS (metal-oxide semiconductor) process development. He was still reporting to SGS in Italy, but also to his boss at Fairchild. His first task was to develop silicon gate technology (SGT). This meant that he first needed to invent a new process to make isolated self-aligned silicon gate transistors, a method for etching deposited silicon patterns, and then document and test those processes. After that, he had to develop an integrated circuit using this technology to prove its worth. The new technology was ready by the end of April in 1968, and the first product produced with it was the Fairchild 3708 which was released later that same year.
In June of 1968, Fairchild under the directorship of Bob Noyce was ending its relationship with SGS, and Gordon Moore was asking Faggin to give a speech at IEDM about silicon gate technology. Faggin’s direct supervisor, Les Vadasz, offered him a role at Fairchild where he’d be able to finish his project, and Faggin accepted. He and Elvia wanted to stay in California, and Federico was enjoying his work. On the 1st of July, the first official date of Federico’s employment at Fairchild, Gordon Moore and Bob Noyce left Fairchild to found Intel (INTEgrated Logic). Andy Grove and Les Vadasz (among others) left shortly thereafter.
Federico was acutely aware that this new company was going to base its existence upon his work with SGT. All of the founders had shown keen interest in his work. Meanwhile, much of Fairchild resisted adopting the technology despite the company’s success with the 3708. According to Faggin, Fairchild had become highly bureaucratic and different departments were unwilling to cooperate and share ideas. In response, Faggin went ahead and innovated some more. He doubled the transistor density he could achieve with his process while quintupling the speed and maintaining the same power dissipation.
Unlike Fairchild, Intel immediately began employing these new technologies as Federico had expected. They released the Intel 1101 in 1969. This was a 256 bit static RAM built using SGT. Intel was on a mission to rid the world of magnetic core memory; a mission that the company would later achieve. By the end of the year, many more employees at Fairchild had left the company. Faggin was frustrated, and he really wanted to get back into the kind of work he had done at Olivetti. He contact Les Vadasz to see if there was any chip design work he could do for Intel.
The first interview was set, but Federico had to cancel as Elvia went into labor. While Vadasz was annoyed with the postponement, it’s quite difficult to argue with a new human being born. Marzia came into the world on the 6th of March in 1970.
A month after the birth of his daughter, Faggin was working in the MOS design department at Intel under his old boss and leading the Busicom Project. On his first day, he found that months prior Intel had begun work with Busicom on developing a 7 chip calculator. Intel had reduced this to 4 chips. Specifications had been made, and essentially no other work had been done since November the prior year. As a result, Faggin had just shy of six months to design all four chips and there was no one available to assist him. Beyond that, Intel at that time was a memory company so Vadasz and Grove didn’t care much about the project.
A few days after Faggin started at Intel, Masatoshi Shima arrived at Intel in California from Busicom in Tokyo to check on the status of the project. It was only then that anyone at Busicom was made aware that no one at Intel had done anything further with the project. Shima was rightly was furious.
Faggin just kept working. He kept working hard, 70 to 80 hours each week. Shima then stayed to help out. The specifications had already been created, there was no time to change anything, and as a result Federico was now working very hard on a product with which he saw some serious design problems. On the few occasions that he had questions, he was told that it was his project and that he should just figure it out. He verified that the architecture was free of errors, but otherwise didn’t change it. He had too much to do. There was still logic design to be done, circuit design to be done, layout, cutting, masking, initial fabrication, testing, fixing any problems found in testing, and then the tasks related to production.
While Federico’s work continued, Elvia was being helped at home by her sister Irene. Irene needed to return to Italy after about 3 months, and Elvia went with her as Federico had no time to offer his family. This wasn’t too bad since Elvia wanted to introduce Marzia to grandparents, uncles, and aunts. Elvia requested to be informed on his work progress, and they would chat on weekends. Elvia’s background was in humanities, but she was interested in electronics to some degree. She read magazines on the topics, and discussed them with Federico with some regularity.
The 4001 and 4003 came back without error. The 4002 required a minor fix. The 4004 came back at the end of December in 1970. Faggin was testing the chips and none of them worked. He started checking them under a microscope and he found that one of the masks hadn’t been applied. Three weeks later, another run of chips came back. He stayed in the lab through the night testing, and everything was working. In January of 1971, Federico Faggin completed testing of the first microprocessor ever made. It was his second self-designed computer, but this one was built with multiple technologies he had pioneered. He was just 29 years old.
The Intel 4004 was a 4 bit microprocessor built of 2300 transistors. It had a 12 bit address width, and it was clocked at 750 kHz. The chip was built on a 10 micron process and packaged in a 16 pin DIP.
Following the successful delivery of the 4004, Faggin was tasked with the Intel 8008 which was made available commercially in April of 1972. The 8008 was an 8 bit CPU with a 14 bit address width. It was manufactured on a 10 micron process and built of 3500 transistors. It was packaged in an 18 pin DIP, which resulted in 30 TTL chips being required to interface memory and I/O. This packaging was dictated by Grove and Vadasz despite protest against it and meant that the chip did not perform as well as it could have otherwise.
Federico wasn’t satisfied with the 8008. He knew a better product could be made and he pushed for such a project within Intel. The company had developed a new N-channel MOS process for 4K bit DRAM, and this is what Federico wanted to use. He also wanted to use a new bus architecture, a new interrupt structure, more instructions, and a 40 pin package; all while keeping the chip machine-code compatible with the 8008. He wanted the instruction cycle time to be 2 microseconds which would put the product within the realm occupied by minicomputers. Approval took time but once he had it, Faggin hired Masatoshi Shima to help with logic design and firmware. The processor was released in March of 1974. This was the Intel 8080 and it was quite successful.
In the early 1970s, a series of calamities hit the US economy quite hard. There was collapse of the Bretton Woods system, the Vietnam War, the ‘73 oil crisis, a steel crisis, and ultimately a market crash. The US was in recession in a big way. Intel laid of 10% of its employees. A reorganization was underway within the company, and Federico wasn’t too pleased. He had already had to fight for everything he wanted to do, and he felt as though his technologies and expertise had been stolen, and that others had been credited for his work. On the 31st of October in 1974, Federico had his last day at Intel.
The 1st of November in 1974 was Faggin’s first day at his startup. He was the CEO, Ralph Ungermann was the executive VP. Funding was secured relatively quickly from Exxon Enterprises for $500k, and Mr. Faggin’s own savings went in as well. One of the first hires was Masatoshi Shima. The plan for the company was to develop a “super 80.” This would be a CPU compatible with the Intel 8080 but would use depletion-load MOS SGT as well as several refinements to the design. Intel was short for integrated logic, and Intel would be the main competitor. So, ilog was also short for integrated logic, but as Federico said it, Ralph said “Zilog,” to which Federico replied “Yeah! Zed, the last word in integrated logic!”
The name, the plan, the funding, and the key people were all in place. Zilog was one of very few companies to be VC funded in 1975, and one of the downsides of this was reluctance to engage in much speculation. The result was that Exxon owned 51% of the company.
The employee count quickly grew to 11, and the team was dedicated to designing their super 80 CPU. Federico was back to his eighty hour weeks, doing layout, drafting, design, and so on. The last 10 days were the worst. According to Federico, some errors had been made earlier and he realized very late that he was going to have to rework a seriously large portion of the design to make everything fit. In the end, he made it all work. The Z80 CPU, PIO, CTC, SIO, and DMA came into being.
The first fully working chips came back on the 6th of March, on time and on budget. This only accelerated activities within the new company. The product launch was planned for May of 1976 with development systems expected to ship in June of 1976. They needed more money, and they wanted to build a fab. This time, the money was easier to get. The company also launched its first ad campaign.
The Z80 is an 8 bit CPU with a 16 bit address width, made of 8500 transistors on a 4 micron process, and running at a clock of 2.5 to 20 MHz. The first samples were sold in March of 1976. The chip was unbelievably successful. It was used as the CPU in the HeathKit H89, the Osborne 1, several Kaypro machines, the Sinclair machines, the various TRS-80 machines, the MSX machines, and many more. The Z80 is still manufactured and sold today.
With the CPU working beyond expectations, Faggin needed a second source. He reached out to Mostek, Synertek, and Rockwell. He then hired Len Perham from AMD as VP of Manufacturing. The two then commissioned a 25k sqft facility from Carl Berg. More office space was needed and rented in Los Altos. They also leased a building Sunnyvale.
Things were expanding and growing very quickly, and Federico was stressed, not eating well, and having some intense stomach pains. He ended up in the hospital due to a perforation of his pyloric valve. With the business roaring, people showed up at his hospital room the day after his surgery to discuss matters of urgency to Zilog.
The first chips came out of Zilog’s own fab in January of 1977. This was important for Zilog as relationships with Synertek (the company whose foundry services they’d been using) soured. They had had several runs with zero yield, and they’d been making unreasonable demands at the same time. Luckily, Zilog’s negotiations with Mostek were also proceeding well. Mostek provided second source for the Z80. At the time, this was important for any chip manufacturer as fabs weren’t the most reliable manufacturing facilities, and yields were often low.
Zilog was taking off like a rocket. From 11 employees and 2000 sqft of office space on the 6th of March in ‘76 to 1100 employees, two wafer fabs, a system manufacturing plant in Sunnyvale, a chip assembly plant in Manila, and four more buildings in Cupertino in March of ‘79.
For Federico Faggin, the Z80 was the last engineering project he’d ever manage. Exxon wanted to compete directly with IBM via Exxon Office Systems. When Zilog was at its peak that business organization (Exxon Office Solutions) was comprised of three divisions: Vydec, Quip, and Qyx. As time marched on, Federico saw very clearly what was coming. That 51% stake that Exxon had demanded meant that Exxon was going to be able to completely absorb Zilog. At the end of 1980, Faggin resigned as CEO, took an executive position in Exxon, and became the Chairman of the Board of Zilog. He then relocated to NYC. He chose to leave completely 6 months later, and Exxon bought out his shares of Zilog.
A few months later, the IBM PC was released using the Intel 8088. While Zilog had some success with other chips and with microcontrollers, it would never again see the great heights and success of the Z80. The PC and its clones would come to completely dominate the market. Zilog would meanwhile be bought and sold several times. Today, Zilog is a subsidiary of LittleFuse.