Clover Computer: A Modern Z8000 CP/M Machine

Seeing some old Zilog 16-bit chips on eBay recently, [Scott Baker] was curious enough to snap them up and build himself a Z8000 computer. It started as a two-board solution, then he added a display module. Instead of layering the boards vertically à la a PC/104 stack, [Scott] decided to build them flat. His first backplane was triangular, but he opted for a square to accommodate one more expansion board in the future. The assembled contraption resembles a clover, hence the name Clover Computer.

The Z8000 was Zilog’s first 16-bit microprocessor, introduced in 1979. It was not hugely popular for a variety of reasons (the Z8000 Wikipedia article has some interesting details). The Z8000 was eclipsed in the marketplace by Intel’s 8088 and Motorola’s 32-bit 68000. One interesting point is that the Z8000 did not use microcode, and as a result, its transistor count was significantly less than its contemporaries. The Z8000 was used in some military applications, and despite its limited commercial success, it continued to be available from Zilog and licensed second sources up until 2012.

[Scott]’s design splits the system into a CPU board, a memory and serial board, and a display board. Along the way, he learns 1980’s era tricks from the Olivetti M20, one of the few computer systems designed around the Z8000. He also manages to find a recent Z8000 implementation of CP/M by GitHub user [], which [Scott] forked and adapted to his project (see project repo here). He succeeds in getting everything working, and ports a monitor, Tiny Basic, and Zork.

Check out his project write-up introductory link, and see it in action in the video below the break. Did you ever use or encounter the Z8000? Let us know in the comments!

20 thoughts on “Clover Computer: A Modern Z8000 CP/M Machine

        1. Ah, I missed that because it was a UK eBay link and I’m UK and didn’t check if it was abroad….

          Though most of the vintage computing stuff is only £100-200 shipping from the US to UK….

  1. Steve Ciarcia had a Z8000 board for the “IBM PC”. Called the “Trump card”. I think early eighties.

    But limited software. There was a BASIC for it, obviously faster than the built in. It had more ram, so maybe a software disk.

    Part of me thinks there was math software for it, but that may have been the original use for the dtack grounded 68000 board for the Apple II.

      1. As Irecall, it was an interface to slow things down for peripheral ICs. You needed extra circuitry. If you ground dtack, itruns at full steam. The guy promoting it, didn’t need the interface,so it made things simpler. The dtack grounded newsletter is online, maybe at

      2. >What’s the significance of grounding dtack?

        DTACK basically meant that the address was ready to transfer data on the current clock cycle. You add wait states by delaying it cycle or two for slow I/O or slow memory. Grounding DTACK caused memory accesses to run at full clock speed with no wait states. With good enough chips you can usually get away with this at 8MHz.

        If neither DTACK nor BERR is asserted (such as with undecoded address space), the 68000 will just sit and wait forever. A watchdog counter that detects lack of DTACK and asserts BERR after 64 cycles or so should avoid that.

    1. Have been looking for the “Trump Card” software for about a month, but all leads have come up empty. I think it’s been lost to time. The articles on the hardware are out there, so it would be feasible to build one in 2022, but lacking the software mitigates the value. If anyone does have this software lurking somewhere, it would be a great asset to the community…

  2. Back in the early 90’s I worked for a company that had some Z8K Onyx systems that ran Mark Williams Co’s Coherent (a V7+ unix clone). I made a bid to buy one of them from the company, but nothing came of it sadly

  3. My first job was writing assembler for a dual Z8002 embedded system autopilot for the Nulka hovering rocket. I remember it was big-endian, which worked out nice for all the fixed-point: you could drop from 32 to 16 to 8 bit accuracy without changing the address. It also had an interesting interrupt vectoring support: the interrupting device could put an 8bit interrupt number onto the data-bus which would index into a vector table to get the address to jump to. We used a periodic interrupt with a hardware counter to give us 256 time-slices for our hand-rolled RTOS. Fun times.

  4. The first working computer I built from chips up (in the early ’80s) is a very simple Z8001 one (though not minimal: it has a Z8010 MMU). I wrote an assembler, but it’s languishing on 3.0″ [sic] floppies. I didn’t write much for it, mostly tests.
    (I had previous experience with Z80 assembly language, so I thought the similarities would help, but they were confusing instead, e.g., I’d forget to say which register I wanted to DJNZ cos on a Z80 it’s implicitly B.)
    PS I’d love to get my hands on a 32-bit ‘Z8000’ (Z80000 or Z320)!

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.