Syncopated Systems
Seriously Sound Science

Common Threads: Media, Science, Technology, and Other Magic

May 2020: The COVID-19 Chronicles

The First Microprocessors

After the Apollo Guidance Computer (AGC), the next major advance in computer integration would come with the introduction of microprocessors.

The first commercially-available microprocessor is widely regarded to be the 1971 Intel (i) 4004, although other notable contenders include the Four-Phase Systems AL1 and Texas Instruments TMS 1802NC (apparently a predecessor of the first CMOS microprocessor, the 1974 RCA COSMAC 1802). These were followed in 1974 by the Intel (i) 8008, Intel (i) 8080, and Motorola (MC) 6800. (The last two were initially priced at $360.) Fairchild Semiconductor announced its F8 microprocessor in 1974 and released it in 1975.

September 1975 advertisement for MOS Technology 6501 and 6502

September 1975 advertisement for MOS Technology 6501 and 6502
(click for full image)

The next year brought price competition. In 1975, Motorola dropped the price of its 6800 to $175. However, MOS Technology—which had gained nearly half of Motorola’s 6800 development team in the previous year—released microprocessors that were much cheaper: for $20 the “pin compatible”—though not software compatible—(MCS) 6501 could be installed in circuits designed for the Motorola 6800 and for $25 the (MCS) 6502 offered economy by integrating expensive clock driver circuits. (At one store, likely about five years later, I recall seeing an inexpensive adapter that would allow a 6502 to be installed in a socket designed for a 6800.)

In 1976, Zilog released its Z80 as an inexpensive and more-capable version of the Intel 8080.

MOS Technology and the Western Design Center

In 1969, Allen-Bradley had started MOS Technology in Pennsylvania to compete with Texas Instruments calculator chips. (I recall a presentation a few years ago by then-CEO of Mentor Graphics and former Texas Instruments executive Walden C. “Wally” Rhines—born 1946 in Pennsylvania—remarking that calculator sales became so embarrassingly profitable for Texas Instruments that it stopped disclosing the amount on its annual reports.)

I (left), Chuck Peddle (wearing hat), Ron Nicholson (behind Peddle) in Santa Clara, CA on September 8, 2016

I (left), Chuck Peddle (wearing hat), Ron Nicholson (behind Peddle) in Santa Clara, CA on September 8, 2016

The designs of the MOS Technology 6501 and 6502 were led by Chuck Peddle (1937-2019), who had joined the Motorola 6800 team in 1973 and in 1974 left to join MOS Technology, around when Motorola moved the group from Arizona to Austin, Texas.

(I was lucky enough to meet Peddle in 2016, coincidentally in the same meeting room where I had on other occasions met Marcian Edward “Ted” Hoff Jr.—born 1937, creator of the Intel 4004, widely recognized as the first commercially-available microprocessor—and Apple co-founder Steve Wozniak—born 1950—and his wife Janet.)

In 1975, MOS Technology was bought by Commodore International, mainly to secure its own supply chain for making calculators. This also gave Commodore formidable vertical integration, allowing it to undercut competitors such as Atari, to which MOS Technology sold ROMs directly and from which it collected design license fees for its microprocessors through second-source manufacturers including Rockwell International.

In 1978, MOS Technology 6502 co-designer Bill Mensch (born 1945 in Pennsylvania) founded the Western Design Center (WDC) in Mesa, Arizona, which created the 65816 microprocessor core, an extended version of the MOS Technology 6502 used in the 1986 Apple IIGS computer and the 1990 Super Nintendo Entertainment System (SNES).

What Happened to Motorola?

Motorola made some fine products, and its 6800 microprocessor was certainly one of them. Many people loved it; for example, I was introduced to the Apollo Guidance Computer by a group called the 6800 Club, which was named for users of the Motorola microprocessor within the Homebrew Computer Club.

Motorola’s Origin

Motorola was founded in 1928 in Chicago, Illinois as the Galvin Manufacturing Corporation by Paul Galvin. After leading the development of an inexpensive car radio that could be mass produced, in 1930 he gave the company a new name that reflected its new focus on products for motorists.

Motorola’s Semiconductor Products Division started in 1948 as a research-and-development (R&D) laboratory in Phoenix, Arizona.

In 1955, Motorola adopted its now-familiar logo and its new division released the first commercial high-power germanium-based transistor.

By 1974, the division had established a new semiconductor fabrication facility in Austin, Texas that “was having difficulty producing MOS chips” and Motorola intended to move its 6800 development team from Mesa, Arizona (a suburb about 20 miles east of Phoenix) to Austin.

6800: Good Part, High Price

While at Motorola and based on feedback he received from prospective customers, Chuck Peddle had proposed to Motorola a low-cost version of its 6800 microprocessor.

After being instructed in writing to abandon the project and facing relocation from Mesa, Arizona to Austin, Texas, Peddle and six others from his 17-person team left Motorola to create the MOS Technology 6501 and 6502.

Galvin’s son Bob Galvin started working for Motorola in 1940, became its president in 1956, and became CEO in 1958; as its chairman in 1975, the younger Galvin acknowledged that Peddle was right, reorganized the semiconductor division, and replaced its management.

68000: Zeroing-in on Target

In 1979, Motorola introduced its (MC) 68000 microprocessor, which appeared to dominate the second-generation of microcomputers including Apple Lisa (1983) and Macintosh (1984), Atari 520ST and 1040ST (1985), and Commodore Amiga (1985), plus video game systems including Sega Genesis (1988), Atari Jaguar (1993), and Sega Saturn (1994).

I recall, while working at IBM’s Almaden Research Center in 1988, my boss Don Feihmann remarking that IBM had considered using the Motorola 68000 in the 1981 IBM Personal Computer (PC), but it was considered too powerful; to avoid risking competition with its workstation computers, IBM selected the inferior Intel (i) 8088.

The Intel 8088 was introduced on June 1, 1979, but the Motorola 68000 was not formally introduced until September 1979, with initial samples released in February 1980 and production devices available in November 1980. Rhines suggests that this delay was how “Motorola, with its superior technology, lost the single most important design contest of the last 50 years”.

PowerPC: AIM for POWER

In 1991, Apple, IBM, and Motorola formed the AIM alliance, which in 1992 began releasing a series of microprocessors called PowerPC, which were based on IBM’s POWER Instruction Set Architecture. (Upon this architecture, IBM Austin in 1986 started developing the company’s RS/6000 series of computers, and in the early 1990s IBM licensed me as a developer for those computers. While working in the Advanced Computing Environment at IBM’s Almaden Research Center in 1988, I worked primarily with the company’s prior generation of workstation, the IBM RT PC, which used IBM’s ROMP microprocessor.)

Barely more than 10 years after releasing its first Macintosh (based on Motorola’s 68000 microprocessor), Apple released its first Power Macintosh and introduced its customers to the PowerPC microprocessor.

Freescale Freefall

In 1999, Motorola spun off its remaining Phoenix-based semiconductor operations (making discrete, standard analog, and standard logic devices) to create ON Semiconductor.

Motorola adopted ARM cores to launch its i.MX microprocessor series around 2000-2001, and for its Kinetis microcontrollers introduced in 2010. (While working for Sony in 2010, I participated in the ARM Partner Meeting in Cambridge, England.)

In 2004, Motorola spun off what remained of its Semiconductor Products Division to create Austin-based Freescale Semiconductor.

The following year (2005), Apple CEO Steve Jobs announced Apple’s transition to Intel microprocessors. In 2006, Apple released its first computer with an Intel microprocessor and soon stopped selling computers with PowerPC microprocessors.

Around 2007, I interviewed with Freescale Semiconductor for the role of program manager for its ColdFire line, which was launched in 1994 to offer embedded microcontroller versions of the 68000 series microprocessors the company launched in 1979. (Based on the 6800 microprocessor released in 1974, Motorola had similarly released microcontrollers around 1984 including the 68HC11 I began experimenting with around 1989.) Although I wasn’t offered the job, I feel honored to have been considered to take the reins of the technology that was—at least in my opinion—the company’s greatest hit.

In 2008, Freescale Semiconductor acquired Austin-based SigmaTel, where (from 2000 to 2001) I had been part of the team that launched the STMP3 line of system-on-chip (SoC) devices for portable media players, which by 2004 had been used in over 70% of flash-based MP3 players, including the first Apple iPod Shuffle. (The first two devices included a Motorola/Freescale Semiconductor DSP56000 core and the third device instead used an ARM core.) Barely more than a year later, Freescale Semiconductor laid of 70% of its former SigmaTel staff.

In 2015, Freescale Semiconductor was acquired by NXP Semiconductors, which itself spun off from Philips in 2006, long after its 1975 acquisition of Sunnyvale-based Signetics, known for its 1972 (NE) 555 timer integrated circuit and the 1972 “final specification” for its 25120 write-only memory (WOM).

The merger of Freescale Semiconductor and NXP Semiconductors formed an American-Dutch company with dual headquarters in Austin, Texas and Eindhoven, Netherlands.

Six Sigma vs. Total Quality Management

Through the Six Sigma (6σ) process-improvement program it created in 1986, Motorola built a reputation for making products that were reliable and could be depended upon to do what they were designed to do.

Unfortunately, the Six Sigma program focused the company’s attention so narrowly on reducing manufacturing defects that Motorola (and its spin-offs) ignored other areas of total quality management (TQM), such as defining quality by customers’ requirements.

Ignoring Communication from Customers

The Motorola 6800 microprocessor is an example of an excellent product that simply offered more than what customers needed for more than customers wanted to pay. In contrast, the 6501 appears to embody a product strategy that has since 2001 been described as developing a minimum viable product (MVP).

With the Motorola 6800 priced initially 18 times—and the following year still 8.75 times—that of the MOS Technology 6501, how could it be that many times as good (or as good as that many) of the cheaper product?

Ignoring Communication to Customers

As an example from my own career, in probably 2006 or 2007 and for a rather wonderful major seller of Apple computers and related products (Other World Computing), I improved the design of a high-performance PowerPC microprocessor upgrade product to better facilitate overclocking. (I had gotten pretty good at designing variable-voltage power regulation circuits, especially in 2001 for the first computer with 64-bit AMD microprocessors, the Newisys 2100.)

When my client started manufacturing its existing product with the latest microprocessor from Freescale Semiconductor (which had been part of Motorola until 2004), it saw its manufacturing yields suddenly and mysteriously drop to near zero.

I was able to resolve the problem fairly quickly, but only after Freescale Semiconductor finally released its documentation for the device—only well after it was already on the market, purchased through a supply channel, and assembled into a produdct.

It turned out that Freescale Semiconductor had introduced a subtle but very relevant deviation from its previous designs that prevented the new microprocessor from working reliably until after the documentation was available for comparison. (The height of the device’s package had been reduced, creating unreliable—if any—contact between the new device and the heat spreader needed to prevent its self-protective thermal shutdown.) Such a critical change should have been documented much more clearly, and much sooner. Within the industry, it is common (and somewhat expected) for manufacturers to communicate such a significant and subtle change to designers in the form of an application note or product bulletin.

Picking the Customers’ Pockets

As a more personal example, after being somewhat dissatisfied with my second Nokia cellular telephone in or around 2008 and recalling the good quality of the Motorola DynaTAC 8000G cellular telephone I had bought in 1989 (with accessories for about $1300), I bought (if I recall correctly) a Motorola ROKR Z6m.

So many things about my new Motorola cellular telephone dissatisfied me that I soon stopped using it and instead returned to using my poor second Nokia for a few more years, until it finally fell apart to the point that I could no longer use it reliably.

One problem I recall with the ROKR appeared to be an intentional design defect with its speaker: although it was perfectly capable of playing music at ample volume, as a speakerphone its volume was limited to the point that it was unusable as a hands-free option while driving, making apparent an attempt to sell an accessory device that I had no interest in buying—and I certainly was not eager to purchase another Motorola product after making such a realization. (There was also a small piece below the screen that would facilitate sliding the unit open and closed, until it fell off, revealing that it had been clumsily designed to be glued on.)

(The next cellular telephone I bought was a Samsung Galaxy S II international version, which I also used until it fell apart to the point that it was no longer usable. This lasted for about 7.5 years, longer than other cellular telephones I had used. So, in 2019, I replaced it with another Samsung, an international Galaxy XCover 4s.)

Lying to and Abandoning Customers

Another client I had starting working for in 2013 had designed a product based on an evaluation board or development board for a Freescale Semiconductor i.MX microprocessor that had pretty clearly been misrepresented as a reference design. (Though I knew what to look for because I had created many actual reference designs about a dozen years earlier for a company that had since been acquired by Freescale Semiconductor, comments in the schematic diagrams such as “not to be used in customer designs” were dead giveaways.)

This client had done many foolish things, including manufacturing thousands of units before the design had been completed through verification and validation. As the company should reasonably have been predicted, the product didn’t work and it obviously couldn’t be shipped to earn the revenue the company had leveraged to manufacture it.

Although I reached out to Freescale Semiconductor, I received effectively no product information nor design support; I had to rely upon my own prior knowledge and learned skills to deduce the cause of the problem, create a practical workaround, and save the 10-year-old company.

Next article in this series: Champions of the Space Age

Previous article in this series: Early Computers

Back to series index: Common Threads index