Relay Computer Update Part 1

It’s no secret the past few years (5 now, wow) I’ve been scheming up building a relay-based computer of my own design. Such a thing would have been too far out of the realm of possibility had it not been for finding the webpage of a CS instructor by the name Harry Porter who, with very limited electronics knowledge, but quite bright in computer architecture, built his very own machine. His machine inspired others of various success, often taking the shape of breadboarded science projects, or table-filling trays of relays and lots of blue LED’s.  His project also inspired me, but it was clear there was more that could be done.

Relay computers have been around for a long time. Konrad Zuse built arguably the first ‘stored program’ machine, back in Germany around the outbreak of WW2, because he was sick and tired of doing civil engineering calculations for school by hand. Employing the assistance of friends, family, and discarded electronic components, he built his monster. At the time he went to the German high command for funding, but thankfully for the rest of us, they turned him down and so it was the Allies that took a gamble on electronic computers as a tool to fight the war (ENIAC for firing tables and the British Colossus for German codebreaking). While Zuse’s machines were electromechanical (and thus speed-restricted by mass and other physical properties) the future was clearly with vacuum tubes. But we’re straying here...

Now, undertaking such a project is not trivial- not in time, complexity, nor money. So ‘getting it right’ the first time is important. And unlike building a stereo or a spaceship, a computer is a limitless tool. It manipulates data based on strings of other data.  So how do you know when you’ve got it? Clearly, a set of ground rules must be established. What’s your input and how does it get put in? How much memory and of what type? Will there be output other than status indicators? And what are we to monitor the status of? What will the system clock be comprised of and how fast will it be? What are the widths of the data bus and address bus? And so on…

A big advantage is that I’m a fan of 1^st and 2^nd gen computers so I have spent the last few years reading everything from history books on Amazon to obscure technical papers scanned into PDF form by archivists. I’ve read bios of the pioneers, studied period trade magazines (courtesy of the Rod) and poured over pictures and layouts. Watched videos, read manuals, collected artifacts, etc. etc. In fact if my memory was worth anything, I’d be the equivalent of that guy that could rattle off the players and stats from some baseball game back in 1932.

Having that historical knowledge on one side, and the technical desire on the other, has pretty much shaped the design of this machine.   Roughly the size of an IBM 1620, but thousands and thousands of times slower. These other guys are flipping switches to input data but I’d like to see a Flexowriter interfaced, for both input and printed output. PLUS punched paper tape. Storage? They’re using paper and pencil. I’d like to design and build a magnetic tape unit to store and retrieve programs. And for RAM, I’d like to see a relay based RAM system in addition to something solid state.

There weren’t really relay computers in the 1950’s (the idea being obsolete 20 years prior), but if there were, I’d like to think that this design would be right at home.

More later……