The World’s First DIY Minicomputer Was Almost Australian

23 Comments

The EDUC-8, a DIY minicomputer design that came out in “Electronics Australia” magazine, was almost the world’s first in August 1974. And it would have been tied for the world’s first if inventor [Jamieson “Jim” Rowe] hadn’t held back from publishing to rework the design to expand the memory to a full 256 bytes. The price of perfectionism?

Flash forward 50 years, and [Gwyllym Suter] has taken on the job of recreating the EDUC-8 using modern PCBs, but otherwise staying true to the all-TTL design. He has all of his schematics up on the project’s GitHub, but has also sent us a number of beauty shots that we’re including below. Other than the progress of PCB tech and the very nice 3D-printed housing, they look identical. We have to admit that we love those wavy hand-drawn traces on the original, but we wouldn’t be sad about not having to solder in all those jumpers.

If you want to dig more into the EDUC-8, or the 8008-based Mark-8 that just nosed it out across the finish line, you can find the original articles online, or even check one out in person at the Computer History Museum. We wonder how many of these were ever made, and how many survive in the wild?

23 thoughts on “The World’s First DIY Minicomputer Was Almost Australian”

  1. ATOM-8 was a minimal 8 bit CPU, designed by Mr. Tomisaki.Arata(富崎新) and published in a Japanese magazine; “トランジスタ技術”(Transistor Technology) in issue #5 of 1973. Issue #6 was general software and issue #7 was on peripheral and supplementary, and was re-published in “つくるコンピュータ” (Homebrew Computer) 1976 issue.

    It was similar to the Kenbak-1 in that it was a bit serial design, but with a much smaller program space of only 32 words.
    Each 8 bit word had 3 bits for the instruction, and thus only 8 instructions, and 5 bits for an address.
    It used 25 TTL MSI,SSI chips and an Intel 1101A 256 bit memory chip.

    http://anycpu.org/forum/viewtopic.php?f=15&t=677

    Although the Elektuur 74 came before the Mark-8 and Educ-8 by a few months:-
    https://anycpu.org/forum/viewtopic.php?f=15&t=53
    https://en.m.wikipedia.org/wiki/Mark-8
    https://en.m.wikipedia.org/wiki/EDUC-8 (Another bit serial machine.)

    The Atom-8 predates the Elektuur 74 article by ten months, and may be able to claim the title of “the first DIY computer project from a magazine”.

    Report comment
    Reply

  3. “The EDUC-8, a DIY minicomputer design that came out in “Electronics Australia” magazine, was almost the world’s first in August 1974. And it would have been tied for the world’s first if inventor [Jamieson “Jim” Rowe] hadn’t held back from publishing to rework the design to expand the memory to a full 256 bytes. The price of perfectionism? ”

    It’s a rhetoric question, I suppose, but I think he did the right thing.
    Without the memory expansion, it would have been less useful to use.
    Also it’s questionable if it had changed anything in terms of popularity.

    There are many early educational computers built like KIM-1 that became a footnote in history merely and longer arw being used.

    Also, let’s think of the bigger brothers of this unit here, based on 8080.
    With a few exceptions, they’re all became irrelevant. The units featuring an Z80, – and an 8085 to some degree -, had largely superseded them.

    All in all, I think that this unit here is special enough alone for trying to put an 8008 processor to use.
    The 8008 was a calculator processor, essentially. Very basic. Using it in such a desktop “PC” was a very sophisticated attempt.

    Report comment
    Reply

      1. Hi there, thanks for your comment, I didn’t know this. 🙂

        But it makes sense in retrospect, the 8008 was about the bottom line for a programmable general purpose computer.

        I mean, 256 Bytes of RAM aren’t much for practical use (about 256 letters in ASCII) but difficult to manufacture in the TTL era.

        It’s the equivalent to about 2048 flip-flops. From perspective of a logic circuit, this would be quite notable.

        Report comment
        Reply

  5. There was nothing wrong with the 8080 as a CPU, except that compared to those that came later it was expensive and hard to use, requiring a complicated power supply and constellation of glue chips. The Z80 basically was an 8080 made cheaper with simplified interfacing, but then the designers looked at the amount of blank mask available and almost literally asked “What else can we cram in there?” (This is very apparent when you examine the opcode layout of the Z80.) It would also be fair to label the Z80 an 8080 with benefits. But when designers and programmers got wind of those benefits, few of them wanted to look back. The 8085 was Intel’s attempt to claw back a bit on the price and design points, and it found a lot of use in non general purpose industrial control equipment where string handling and DRAM refresh weren’t such sexy upgrades but the single voltage power supply was. Intel went on to dominate that space with the 804x and 805x families which came to define the role of “microcontrollers” as opposed to “CPU’s.”

    Report comment
    Reply

    1. In my opinion, the i8080 support chips were a good thing. Maybe more relevant than the i8080 itself, all in all.

      Like the i8253 timer or the i8250 UART.
      Ok, maybe not the i8250 so much. It was buggy and not as good as Z80 SIO or Z80 DART.

      What set the Z80 apart from i8080 the most was the 5v power supply, the special registers (IX, IY; ability to process 16-Bit data) and the DRAM refesher (wise devs used SRAM, though, anyway).

      To be fair, the i8085 did go into right direction already, but it wasn’t Z80 compatible.

      Not even partially, which could have had increased its adoption.
      Most Z80 programmers weren’t so downlooking to plain i8080 systems, at all.

      They not seldomly tried to stick to plain i8080 code for sake of portability (also with CP/M in mind) and also had used plain i8080 compilers.

      Unfortunately, some of the advanced Z80 instructions made certain projects possible to begin with.
      Performance and memory requirements would have been worse by opting to stick with pure i8080 code.

      The i8080 was like an AdLib Gold in the days of an SB Pro 2.
      It was the original, the former pioneer.
      But it was no longer leading. Also, in contrast, AdLib Gold was technically better, the i8080 was not.

      If i8085 was just a bit more like, say, the NSC800 instead.
      Not a complete drop-in replacement to Z80, but with the most popular Z80 instructions being supported.

      I mean, intel (and IBM) didn’t really learn from this.
      The 8086/8088 had same “problem”, essentially.
      It was stuck with an oudated instruction set forever, like the i8080 was before in the days of Z80.

      Whereas the competition (NEC) had provided a better alternative (V20/V30; 8086-2 instruction set as used by 80186/80286).

      This is especially sad if we have a look at creations like the IBM PS/2 Model 30 with its i8086 processor.
      The machine was released in the late 80s, by the time the i8086 was outdated for more than 10 years!

      The old i8086 instruction set was somewhat limiting here.
      A NEC V30 with the newer 8086-2 instruction set would have maintained a partial 80286 compatibility (no PM), at least.
      No idea what IBM was thinking here to go for an i8086, it was sort of unlogical.

      That’s all related in sofar, because the rest of the PS/2 series consisted of mostly 80286 PCs at the time (plus a few i386 systems), being all capable of running 16-Bit OS/2.

      Providing at least 80286 Real-Mode compatibility. This compatibility could have been relevant when compiling Family API applications running on both DOS and OS/2 machines.

      Alas, the choice for the outdated i8086 prevented this compatibility and required such applications to use plain i8086 instructions only, if compatibility was bring important.

      Gratefully, most Turbo XT owners did refit their Turbo XT with a NEC chip on their own.

      That being said, I’m just a layman here. This is merely the point of view from a hobbyist.

      Report comment
      Reply

  6. I’m australian and would dearly love to back up this claim, but AFAIK both the australian EDUC-8 and american Mark-8 were beaten to the punch to be the first home-built computer to be published as a series of articles in a magazine. That honour goes to the dutch ’74 Computer-1′ that appeared in Elector February 1974:
    https://anycpu.org/forum/viewtopic.php?f=15&t=53

    Also, here’s the cover photo from Electronics Australia for the first EDUC-8 article:
    https://www.surfacezero.com/g503/showfull.php?photo=262955
    I have all the EDUC-8 issues in a bound EA volume somewhere.

    Report comment
    Reply

    1. That’s true, but no construction plans were published, sadly. With those dirt-cheap miniature blue relays available these days a reconstruction of this electromechanical brain would fit in a small(er) suitcase :)

      Report comment
      Reply

  9. In my first year at Swinburne College Of Technology (one of several names for what is now Swinburne University) in 1979, I got to do pracs on an EDUC 8. Computer magazines in Oz used to refer to it, tongue in cheek, as an E DUCK-8.

    Report comment
    Reply

    2. Yes however, these weren’t published kits. They were schematics only and that’s where the bar is set. When it’s something anyone can buy and assemble. The 74 and this one were not and required a huge skill set to assemble where as the mark 8 and educ8 were available at electronics hobbyist stores.
      Prior to this there are even more schematics and ideas published, right back to the 50s but until they reached the pcb stage they aren’t classified as kits for home

      Report comment
      Reply

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.