Zilog’s classic Z80 chip is soon to be dead, though it might not be gone forever if one open source project succeeds in its goal to clone the legendary processor.
Developed by Unity iOS dev Renaldas Zioma, the Free and Open Source (FOSS) Z80 GitHub project outlines its plans to revive and preserve the Z80 with a physical silicon clone. The project is ostensibly a reaction to Zilog’s end-of-life announcement for the Z80, which was introduced in 1976 and was used in many of Texas Instruments’ calculators; gaming consoles like the Sega Master System; the ZX Spectrum; and many other devices.
The aim of the FOSS Z80 project isn’t to merely replace the Z80, which can be accomplished by programming an FPGA to act like a Z80 – and there are Z80 soft cores out there to drop into FPGAs. Nor is the aim to replace the classic Z80 with a closed-source physical clone or some kind of modern system-on-chip, which already exist.
Rather, Zioma wants to make a physical open source clone of the Z80, based on Guy Hutchison’s TV80 soft core.
“It is time for open source and hardware preservation community to step in with a Free and Open Source Silicon (FOSS) replacement for Zilog Z80,” the project’s GitHub page says. “Goal: To develop a drop-in Z80 replacement in 8-bit home computers such as ZX Spectrum.”
Crafting a chip generally involves designing the component; testing and validating it in simulations; and then taping out the blueprint for an initial production run and further testing ahead of volume fabrication. While the first step should be relatively easy as the Z80 has been reverse engineered, producing even physical samples would normally cost a decent amount of money, more than a single developer would ordinarily be prepared to pay for.
To get around this problem, Zioma says he’ll make use of Tiny Tapeout, a multi-project wafer (MPW) service that uses Skywater’s open source 130nm process design kit. MPW chips cut down on the manufacturing costs by squeezing several die designs onto a single silicon wafer, which allows hobbyists and devs to get their hands on custom silicon without forking over tons of cash to semiconductor fabs. In effect, these smaller teams share the cost of a production run.
Tiny Tapeout uses a near-25-year-old process node, which keeps costs down, and operates a shuttle service with Efabless. People submit their open source projects to be included in the next production run, and if accepted into a shuttle, will have their individual designs consolidated into one relatively large die. See Tiny Tapeout 6 as an example of a shuttle; it has many different small die designs squeezed into one larger die.
Multiple large dies are then fitted onto wafers and manufactured, then put into chips, and distributed to those whose projects are in the large die. Thus, in the end everyone gets a load of chips each of which has everyone’s designs in it, and you select the design you want for the application you need using the built-in multiplexer. That multiplexing does limit the number of IO pins each sub-die gets.
The first FOSS Z80 samples will come with Tiny Tapeout 7, which is apparently set for manufacturing in June, though actual delivery is estimated for December 30. This means if the open source Z80 clone has any bugs in the silicon, we won’t know for many months.
The project plans for two further tapeouts using Efabless’s ChipIgnite program, which also uses Skywater’s 130nm node. Under ChipIgnite your design gets a die to itself, rather than it being multiplexed. The final tapeout will therefore see the Z80 clone put into its classic dual in-line package with 40 pins so it can be seamlessly installed into boards made for the original Z80.
Getting to production won’t be easy
Even if the project is successful, it’s clear that final Z80 clones won’t be arriving until next year at the earliest. Plus, it’s not clear how these chips will hit volume production as ChipIgnite quotes a price of $9,750 for 100 chips, or $97.50 each. An extra thousand clones would cost $20,000, which at $20 each is a better deal but is rather expensive for just a single person.
Unless Zioma has the capital to fund this project to completion, or successful crowdfunding or some other line of financing, it’s hard to see how the FOSS Z80 clone will actually make it to wider production volumes. Since the chip will be open source, it’s possible some company could pick it up and bring it to the fabs, but if Zilog itself doesn’t want to make the original Z80 anymore, it’s uncertain whether another company would either.
We’ve contacted Zioma to get his thoughts on the potential challenges for his open source clone, and we’ll update if we hear back from him.
For those who want something close to the Z80 without emulating it or using an FPGA, Zilog still makes the eZ80, which is more or less a more powerful version of the Z80. Not quite the same, but at least you can expect to buy it brand-new for the time being. ®
