Relay Computer Consumes Six Years And 4.5 Suitcases

If you thought your home-brew project was taking a long time, [Jeroen Brinkman]’s MERCIA Relay Computer project probably has you beat. He began working on this impressive computer back in 2014, and has been at it ever since.  In fact, the ongoing nature of the project is embedded into the name itself — the English translation of the acronym MERCIA is “My Simple Relay Computer Under Construction”.  Being interested in old analog and relay computers from an early age, [Jeroen] took on this project to educate students about how computers work.  The entire computer is build only using relays, diodes, and capacitors, not to mention color-coded wire based on signal functions. Using relays as the primary switching elements is at the core of his educational goal — anyone can understand how a relay works.

Understandably, this thing is big.  But he has cleverly packaged it to visually show the major building blocks of a computer.  While the exact size isn’t stated, we can estimate based on the photo of [Jeroen] standing next to the modules that these panels are about 1.5 m tall and perhaps 60 cm wide.  The whole computer is nine panels wide, making it about 5 meters long.  Except for the ROM assembly, pairs of panels are hinged together and they fold like a book and carried like a suitcases when being moved.  If you enjoy the clickety-clack sound of relays, be sure to watch the relay longevity test in the video below and check out our article on the 1958 FACOM from last year.

This is a fascinating project, but unless you have a couple thousand relays laying around and a decade of free time, it’s probably better to just enjoy [Jeroen]’s work rather than build your own.  We hope he releases schematics and other documentation once the project is finished.  You can follow his Facebook build log if you want to keep track of the progress. Thanks to [David Gustafik] for the tip.

38 thoughts on “Relay Computer Consumes Six Years And 4.5 Suitcases

  1. I guess it would be cheating to use an arduino to set the banks ROM and microcode?
    But it would be so enabling especially for demonstration purposes to switch configurations quickly.

    1. Yes, yes it would be cheating.
      That’s like taking a school bus instead of walking barefoot in in twenty below weather 5 miles uphill each way through 5 feet of snow EVERY DAY!
      Setting dipswitches builds CHARACTER!

        1. First, “Olympic swimming pool” defines the length, but not the number of lanes, nor the depth. But second, even if you use the nominal size, 50m x 25m x 2m, this gives you 2500m^3. A very large suitcase, 1m x .5m x .5m, would be .25m&3, so there would be 10,000 very large suitcases in a nominal Olympic swimming pool, so you’re off by at least an order of magnitude.

  2. nice project, but the web site brings back memories of the 90s. Complete with “web rings” buttons, mostly text, low quality pictures, etc.

    All it needed was a quick time video, some java crapplets, and we would have a living 90s time machine.

    1. That webring is probably one of the last in existence but it is still relevant, even more so recently, because 20 years ago it was a weird novelty, today it is an open club that dignifies merging art & science.

  3. 75 years latter, someone has finally built a transportable version of the so efficient, reliable, cheap and fast relay computer.
    You cannot stop the inexorable march forward of science and technology.
    Next step : make it portable !
    Nice working skill though.

      1. Also, you can save a lot of money and space that way. I don’t know how much the relays in this project cost, but I’m seeing reed switches in the neighborhood of $0.50 on Amazon, and haven’t even looked at AliExpress. It’s almost trivial to make relays out of reed switches, and you can make them with as many poles as you need, by bundling reed switches together. The main down-side is that you only get normally-open contacts, but you can add bias magnets that are opposed by the coil to get normally-closed. You can even make self-latching relays, using a bias magnet that isn’t strong enough to close the contacts, but will keep them closed once closed by a coil. To turn it back off, you either run current through the coil the other direction, or use a second coil, wired in reverse, to oppose the bias magnet’s field. This makes a set/reset flip-flop in one relay. I’ve done this – it works.

  4. Sometimes I wonder how efficient and or fast a relay computer can get if using modern parts.
    Considering that the mechanical properties of relays surely must have gotten better during the last 50-70 years since relay computers were “common”.

    I know for one thing that relays have gotten considerably smaller if anything. Considering that there is relays that are about a cm long on their longest side and still manage to have 2 switches inside…

    Question is though about needed holding currents, and general power consumption, as well as switching times.
    Non the less, it is likely to not be all that impressive in the end…

    If one were to use MEMS fabrication, one could likely make something more impressive. But at some point, it might be hard to get people to believe that it is indeed mechanical switches, and not just semiconductors on a chip… (And yes, there is likely many failure points with such a chip, but relay computers aren’t all that practical to begin with.)

  5. It’s so easy to criticize and second-guess something that someone has created. It took an enormous amount of work to develop this system and I for one appreciate it. And let’s not forget the education this project provided.

    Thank you, Jeroen!

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.