feature article
Subscribe Now

Z80, Adieu

I’ve published several obituaries in EEJournal over the past few years, but this is my first such article for a microprocessor. After 48 years, Zilog has issued an end-of-life notice for the Z80 microprocessor, introduced in early 1976. I have a long history with the Z80 that stretches back all the way to that introduction year. I consider the Z80 to be the best 8-bit microprocessor ever designed, and I am sorry to see it fade into history, but all good things come to an end. Today, the Z80’s limited 64Kbyte address space and slow clock rate make it a less-than-ideal choice for embedded projects. It cannot handle today’s large data types including images, audio files, or video streams. It’s not especially good for implementing network stacks. Watching a Z80 tackle IPv6’s 16-byte Ethernet addresses is just downright ugly. These are limitations brought on by the Z80’s inherent “8-bitness.” So, let’s say a fond farewell to this old friend by taking a nostalgic trip back to its beginnings.

There would have been no Zilog and no Z80 microprocessor if Intel had not treated Federico Faggin in such a shabby manner. Faggin was instrumental in designing the first commercial microprocessor, Intel’s 4-bit 4004. When Faggin left Fairchild Semiconductor and arrived at Intel in 1970, Stan Mazor handed him the general specifications for the four chips in the 4000 series microprocessor family and informed him that Masatoshi Shima from Busicom, the calculator company that had contracted with Intel to build this chipset, would be arriving in two days to review the project’s progress. Shima was expecting to see a completed logic design for the 4004 microprocessor and partial layouts for the other three chips in the family. None of this work had been done. When Shima arrived and discovered the lack of progress, he was furious. Faggin and Shima dug in and delivered the four chips in nine months, at the beginning of 1971. That was the first time that Faggin pulled Intel’s fat out of the fire.

As soon as the 4004 project was completed, Faggin took over the 8-bit 8008 microprocessor project, which was similarly stalled. Faggin used all the design tricks he’d developed for the 4000 series and managed to complete the 8008 microprocessor’s design by the end of 1971. That was the second time that Faggin pulled Intel’s fat out of the fire.

Faggin then lobbied for an improved version of the 8008. The new microprocessor would become known as the Intel 8080. However, Intel was fully committed to DRAMs at the time. Microprocessors were seen as a sideline, so it took until November, 1971 for Intel to green-light the new microprocessor project. Faggin again went to work and first silicon for the 8080 arrived at the end of 1972. The 8080 became a huge success for Intel. In just three years, Faggin had played a central role in the development of Intel’s first three microprocessors.

When Andy Grove became an executive vice president at Intel in 1974, there was immediate friction between Grove and Faggin. In those early days, Grove viewed microprocessors as an unwanted distraction from Intel’s real business, which was manufacturing DRAMs. He’d change his tune a few years later when Japanese DRAM vendors were killing Intel’s memory business. However, Faggin was not willing to wait around and he was not willing to work for “Attila the Hun,” which was the nickname Grove had earned thanks to his aggressive-bordering-on-abrasive-bordering-on-abusive management style. At the time, Intel did not want to fully commit to the microprocessor business, so Faggin decided he’d create his own microprocessor company. Faggin left Intel on October 31, 1974, on Halloween. He started Zilog the following day.

At first, Faggin planned on creating a microcontroller as Zilog’s first product. He soon realized that he’d also need to create all the software development tools and application literature needed by any microprocessor or microcontroller. He put the microcontroller idea on the back burner and came up with a concept that required less support – the Super80 microprocessor – which would implement a superset of the Intel 8080 microprocessor’s ISA. That way, Zilog could ride on Intel’s coattails. The Super80 would be built with the latest 5-volt NMOS process, so it would be much easier to design into a system than the Intel 8080, which required three power supplies (5 volts, -5 volts, and 12 volts).

Faggin laid out most of the Z80 himself, by hand. In his autobiography, “Silicon,” Faggin wrote:

“When it was completed, the Z80-CPU layout was my masterpiece! To this day, I am very proud of it and still have never seen a random-logic handmade layout denser than the Z80.”

Here’s a die photo of the Z80 microprocessor taken by Stephen Emery of ChipScapes, which would tend to confirm Faggin’s statement.

Die photo of a Zilog Z80 microprocessor. Image credit: Stephen Emery, ChipScapes

Zilog got fully working samples of the Z80 microprocessor back from its fab partner on March 6, 1976 and announced the product in May. I remember sitting in a conference room at HP’s Calculator Products Division in Loveland, Colorado just after the Z80’s introduction and listening to Zilog sales representatives explain the virtues of their new microprocessor. At the time, I remember noting the single-supply requirement and the built-in DRAM refresh. This microprocessor had obviously been designed from a system designer’s viewpoint.

I came to appreciate the Z80 microprocessor’s other unique features after buying my first personal computer, a NorthStar Horizon, which was based on the Z80 and the S100 bus. The Horizon was designed to use an RS-232 terminal for a human interface, but I wanted more performance and installed a Vector Graphic Flashwriter II memory-mapped video card. While writing a terminal emulator program for the Flashwriter II card, I discovered the Z80 microprocessor’s EX and EXX instructions and its auxiliary register file. Those two instructions cemented my affection for the Z80.

The Z80 replicates the 8080 microprocessor’s register set, which consists of four pairs of 8-bit registers. The accumulator and the eight flags constitute the AF (accumulator and flag) pair. The other three pairs are the BC, DE, and HL pairs. However, the Z80 improves upon the 8080’s register set by adding an auxiliary register set that duplicates the existing register set. Register pairs in the auxiliary set are designated AF’, BC’, DE’, and HL’. The Z80’s EX instruction swaps the contents of the AF and AF’ register pair. The EXX instruction swaps the contents of the BC, DE, and HL pairs with BC’, DE’, and HL’. (I have always suspected that these instructions merely toggle a register pointer rather than moving any actual data because they each execute in one machine cycle.)

The beauty of this auxiliary register scheme is that code written for the 8080, which included the CP/M operating system and all of the software applications written to be compatible with the 8080 microprocessor, were blissfully ignorant of the Z80’s auxiliary registers. Consequently, the auxiliary registers were ideal for fast, hidden context switches, which is precisely what I needed for my terminal emulation code. Again, the Z80 seemed to have been architected with system design in mind, and I am a system designer. My affection for the Z80 is reflected in the original manuals I’ve kept from the days that I wrote Z80 code, more than 40 years ago.

 

My Z80 programming library dates from the 1970s, but the books are still on my shelves as a testament to this microprocessor. Faggin’s autobiography “Silicon” appears at the bottom of the image. Image credit: Steve Leibson

According to a Zilog announcement dated April 15, 2024, last-time-buy (LTB) orders for Z84C00 ICs will be accepted until June 14. Depending on how many orders Zilog receives by that final date, quantities for LTB orders may be limited. Of course, last-time buys like this one from Zilog are for production volumes. If you only need one, or a few Z80s, you can always check eBay, but I recommend checking with Anchor Electronics in Santa Clara, California. Z80 parts still appear on their price list and I’ve always enjoyed doing business with them.

References,

Federico Faggin, “Silicon,” Waterside Productions, 2021

5 thoughts on “Z80, Adieu”

  1. Steve,
    Whenever I would look at a custom laid out (not automated) chip, I would infer whether they had issues (functional/timing) with the design (or where last minute changes found available ‘free’ space). The Z80 does look beautiful and logically arranged. Faggin did a wonderful job on the architecture and the design.

    Thank you for including the Mostek Z80 book that held the entire family of Z80 peripherals that Mostek manufactured and supported.

    I wonder where Intel would be today if they had jumped (rather than be pushed off the cliff) to microprocessor earlier AND kept Faggin on Intel’s payroll.

    A sad day but one that we should celebrate given how long the Z80 survived against the competition.

    Bill

  2. > I consider the Z80 to be the best 8-bit microprocessor ever designed

    A good one, certainly, but I think not the best.

    Even at the same time, the 6502 brilliantly made use of the fact that RAM of the time was as fast as logic (later also exploited by ARM) to make a CPU that was just as fast but used half as many transistors. Faster if you had a lot of variables and ran out of registers on the Z80.

    A little later on, the 6809 was a much better 8 bit microprocessor than either.

    In the 1990s the AVR became what I would consider to be the best 8 bit ISA, though it is designed as a microcontroller with Harvard architecture running a fixed program, not as a general-purpose computer. Recent versions have switched to Von Neumann architecture. Also it uses a lot more transistors.

  3. There’s no denying the 6502’s success, brucehoult. It was the foundation for the Apple II, the Atari VCS, the Atari 400 and 800 personal computers, and many other truly successful products. Alas, the 6809 was a bit late to the party, but a fine machine that unfortunately had to compete with the 16-bitters that came out at the same time. The AVR was able to take advantage of two decades of progress in semiconductor processing, giving it a grossly unfair advantage. Many microprocessors have their advocates and they’re all welcome to their preferences, in my opinion.

    1. What really pushed the AVR and the PIC to the forefront of the day was, oddly, Motorola.
      Motorola had a lot of design wins with the 6805, then one day said “we sold them all to Detroit”.

      The AVR and the PIC were the only available near equivalents then and we had to design out any Motorola parts due to lack of availability, moving to one or the other.

      So Motorola created their own competition, in Atmel and General Instruments/Microchip, that would later consume large parts of Motorola’s market share.

Leave a Reply

featured blogs
Jun 9, 2024
I've seen several do-it-yourself (DIY) miniature versions of the Las Vegas Sphere, but this is the best contender thus far....

featured chalk talk

IoT Data Analysis at the Edge
No longer is machine learning a niche application for electronic engineering. Machine learning is leading a transformative revolution in a variety of electronic designs but implementing machine learning can be a tricky task to complete. In this episode of Chalk Talk, Amelia Dalton and Louis Gobin from STMicroelectronics investigate how STMicroelectronics is helping embedded developers design edge AI solutions. They take a closer look at the benefits of STMicroelectronics NanoEdge-AI® Studio and  STM32Cube.AI and how you can take advantage of them in your next design. 
Jun 28, 2023
39,162 views