Tedium - The CPU That Will Never Die 💾

Today in Tedium: Like me, you probably have used a Z80 CPU every day without realizing it. I, on the other hand, was fully cognizant that I was using a Z80 back in the day. Growing up as an Amstrad enthusiast, how could I not? All that said, as the Z80-based Amstrad CPC turns 40, let’s dive into a brief-history of the chip that powered it, and many other machines I’ve owned since. Midlife crisis averted? Just maybe. In today’s Tedium, We take a look at some of the interesting and obscure uses for this little chip. So, Z80, “this is your life (so far).” — Mattl @ Tedium

Today’s GIF comes from a video by Anas Kuzechie featuring his Z80 breadboard programming efforts.

(Seth Morabito/Flickr)

The chip was developed by a bunch of ex-Intel and Fairchild engineers who felt they were being ignored at Intel

Federico Faggin was a key figure in the development of Intel’s soon-to-be-dominant microprocessor business, but it was a brief moment when Intel took its eye off the ball that led to his creation of the Z80.

In the mid 70s, before Intel would become the company we know today their primary interest was selling static RAM and ROM chips. There was also a recession going on and Intel had to lay off a number of people. Faggin was soon looking towards the exit himself. The computer revolution was coming, but Intel wasn’t (yet) interested. Microprocessors were just seen as a way for Intel to sell more memory chips. Faggin would say that the 8080 took nine months of negotiations before it was allowed to happen.

Irritated, Faggin invited colleague Ralph Ungermann out for drinks to see if he was interested in starting a company of their own. He was—and soon enough, both engineers resigned, with Ungermann leaving Intel just a few weeks before Faggin’s Halloween 1974 exit. They would be soon joined by transistor-level designer Masatoshi Shima, but initially told him to stay at Intel until they had a product, or some money.

As Faggin would later say in an interview with the Computer History Museum:

Microprocessor marketing, under the direction of Bill Davidow, was beginning to assert its own rights to develop chips. Now that was absolutely correct. They should’ve been defining chips from Day One, but that was a job that I had done from just about from the beginning after the 4004 and the 8008. And so I felt boxed in at Intel. Also I was working very hard and I felt that I could do better if I started my own company.

The newly formed company (as yet unnamed) began designing a microcontroller chip instead called the 2001. After meeting with Synertek to discuss fabrication of the microcontroller it became clear that they would be unable to compete with Intel with this product. This spurred them to begin creating a real microprocessor. It also gave them a business plan and put them in a good position to talk to investors.

One of those investors turned out to be Exxon. Yes, that Exxon, the one with the gas stations and infamous oil spills.

Ah, Exxon, that oil company that seeded the computer industry. (Mike Mozart/Flickr)

Big-oil seeds the 8-bit revolution

Early on the industry newsletter Electronic News published a story on the newly formed company, which led to an investment of $500,000 in June 1975 from Exxon Enterprises, Exxon's high-tech investment arm.

With money and a new plan for a microprocessor, Shima finally joined in February 1975. He immediately set about producing a high-level design, while Ungerman began working on the myriad peripheral chips that would complement Shima's work, with an early version of the design ready a few months later. By the end of 1975 things were almost ready.

Meanwhile, Faggin was looking for someone to fabricate the new chip. Approaching Synertek again, but plans fell through when the company asked for a second source license, which would allow Synertek to sell their own Z80 chips. Faggin then approached Mostek, who agreed to an exclusivity deal and later received a second source license of their own. Synertek would later get their own second source license, amongst several other companies. More on that in a bit.

In July 1976 the Z80 went on sale. One of Zilog's very first customers was a buyer who, unbeknownst to them, worked for NEC.

The Zilog team had worried about the trend of Japanese companies making and selling unauthorized clones of American components and Faggin had come up with the idea of adding subtle "trap" transistors that would fool a visual inspection. Shima's previous experience with NEC minicomputers had led him to add the concept of two sets of processor registers to the Z80 and agreed with Faggin's concerns and added six traps to the design. As revealed in the excellent oral history of Zilog, Shima was told by an NEC engineer that the traps had delayed NEC's copying efforts by around six months.

The NEC UPD780C, which is compatible with the Z80. (Wikimedia Commons)

As Ralph Ungermann tells it:

When we were packaging the product up for shipment, somebody knocked on the door of the office and it was an NEC guy with four $100 bills for two chips. And then right after that a very successful Japanese computer company came in with four $200 bills. That was just the start of the copying.

With a successful launch, Faggin would approach their funders at Exxon with a view to opening their fabrication facility. Exxon agreed, which allowed Zilog to capture approximately 70% of the Z80 market while giving the go-ahead for Mostek to begin producing their own Z80-compatible chip known as the MK3880. This also provided stability for the overall Z80 market as an established company was also producing the chips should the plucky startup fail, despite the hesitation of Exxon.

As Faggin recalled:

We had a hard time finding acceptance at Exxon Enterprises of the idea of second sourcing. Second-sourcing was a necessity for us, because being a startup company, we couldn’t see how we could get into large accounts where people had to bank on a small company, invest in software development, and risk that the company disappear. So it was essential for us to have a second source. And it took a while to figure out the strategy how to do that. And that was weaved in with the design of the Z80 that started in detail when Shima joined the company.

Second sourcing wasn’t uncommon during this time—famously, AMD started as a second source of Intel, only for it to later blow up in Intel’s face. But in many ways, it shaped the long-term fortunes of the Z80 more fundamentally than Intel’s second-sourcing ever did.

Pac-Man, a notable exampe of a Z80-based game. (sei/Unsplash)

From arcade cabinets to graphing calculators: The Z80’s gradual evolution away from the mainstream

Admittedly, most of us don’t use Z80s in the computers on our desks in the modern day—but there were still plenty of places where the Z80 would continue to see use well into the modern day.

For one, the Z80 was a common choice for creators of video games with a Z80 powering Pac-Man, dual Z80s in Scramble, and three in each Galaga machine. Admittedly, I'd love to tell you about some mythical modern game with 4096 Z80s inside it, but sadly, this trend of adding more CPUs to an arcade game didn't continue.

In Europe, meanwhile, the home computer market fell in love with the Z80, with machines such as the failure of the Tatung Einstein (although it did see use as a development tool for other Z80 machines), Sinclair's ZX80, ZX81 and ZX Spectrum and my beloved Amstrad CPC would see various Z80 machines released between 1980 and 1990 with Amstrad even producing the ill-fated dedicated Z80 based games console the Amstrad GX4000 and a range of Z80-based dedicated CP/M-based business machines in the form of the Amstrad PCW.

And over in Japan, where Z80 cloning had become commonplace, saw its own Z80-based machines in the form of the MSX-standard, a collaboration between Microsoft and ASCII Corporation. Many MSX machines were built, with some (especially Toshiba models) around a selection of custom system-on-a-chip offerings called the MSX-Engine which often integrated a Z80-clone alongside other needed chips. After all, the MSX was more like the VCR, with many companies producing machines so cheap that readily available hardware greatly helped.

(Console marketing like a VCR? It was a trick later attempted by the 3DO company in the 90s, as featured in a recent Tedium.)

In the US things were a little different, while the Z80 would have initial success with Digital Research's CP/M operating system, some early business machines from the era were also Z80-based including Heathkit H89, the luggable Osborne 1 and the Kaypro series.

The original model of Radio Shack's TRS-80 would use a Z80, while DEC's Rainbow 100 and the Seequa Chameleon would feature a Z80 alongside a custom version of MS-DOS and compatible Intel 8088 chip (slower variant of the later infamous x86 founding model, Intel 8086) allowing them to run both the wealth of Z80 CP/M software as well as somewhat compatible software.

In terms of dedicated games hardware, the Z80 would power much of Sega's early forays into home video game consoles as the central processor Sega's SG-1000 (1983), Master System (1986) and Game Gear (1990). Even Sega's first successful system in the US, the Sega Genesis (1989) released a year earlier in Japan as the Sega Mega Drive would have a Z80 onboard to handle sound (and by virtue of a cartridge pin-adapter, backward compatibility with earlier systems, a hallmark of Sega's 8 and 16-bit systems). Such was the commonality amongst Z80 based systems that after the 1983 video game crash a system would be produced that could play both Coleco's 1982 Colecovision and Sega SG-1000 games, the Telegames Dina. The ill-fated Colecovision add-on to turn it into a home computer, the Coleco Adam would not see the same love.

Portable Z80 machines were produced from RadioShack's TRS-800 Model 100 to the later Amstrad NC-range (rebadged as the Dreamwriter and used in schools in the US) and Clive Sinclair's Cambridge Z88. Both machines would offer some limited built-in programming ability and a rare conceit by Sir Clive to his own history by way of a built-in version of BBC BASIC. In recent times, David Given ported a CP/M-like operating system (but not CP/M itself due to licensing restrictions) to the Amstrad NC-200 after discovering an undocumented keypress that would let it boot software from the built-in floppy drive.

(Aaron Lefler/Unsplash)

As technology progressed and home computer hegemony arrived, the Z80 found a second life in various personal devices, calculators and pocket computers. Since the early 90s, Texas Instruments has been producing the very popular TI-8x range of graphing calculators most of which contain a Z80-based CPU, with models released as recently as 2015 containing a modern, faster variant of the chip.

(Infamously, TI held a de facto monopoly on the calculator market in the U.S., due to their required use in standardized testing, which likely helped sell a lot of Z80s.)

Meanwhile, Sharp, a second-source manufacturer of the Z80, has been producing Z80-based small computers for decades, starting with the Sharp MZ-series and X1 machines from the early 80s, through to later models in the Wizard line of personal organizers (as gifted to Jerry's father in a 1998 episode of Seinfeld).

Remember the Apple Newton? Amstrad released their own similar device six months earlier, and yep, it used a Z80. Good luck finding a usable one now, the soft touch material used on the PDA 600 as it was officially known has turned into a sticky mess. Amstrad’s swan song for the Z80 would come when it attempted and failed to entice people into buying the doomed emailer, (sorry e-m@iler) line of landline phones, with later models featuring downloadable ZX Spectrum titles as one of the myriad reasons to buy one.

A good reason to not buy one: it would dial a premium rate number every day to collect your email. A device that sold so poorly that the CEO—the guy who came up with the idea for the product, to his later regret as Alan Sugar became obsessed with the idea—quit Amstrad after 25 years.