We all know you can build a computer out of relays, and if you’re a regular reader of Hackaday, you’ve probably seen a few. Actually designing and fabricating a computer built around relays is another thing entirely, and an accomplishment that will put you right up there with the hardware greats.
The newest inductee of the DIY microcomputer hall of fame is [Jhallen]. He’s built a microcomputer ‘trainer’ out of relays. It’s got more click and clack than the Tappet family, and is a work of art rendered in DPDT relays.
The biggest consideration in designing a relay computer is the memory. You can implement a CPU in a few dozen relays, but even a small amount of memory is still hundreds of additional components. In this computer, [Jhallen] is sort of cheating. The memory is implemented as 256 32-bit words on a microcontroller alongside a controller for the front panel. The CPU is still all relays, with support for self-modifying code, a bunch of instructions for conditional jumps, and an ‘increment and jump if not equal to zero’ instruction.
Below, you can check out a very in-depth video of the relay computer in action, starting off with some satisfying click and clack of Euclid’s algorithm and a demonstration of the variable clock rate. The video goes on to demonstrate the assembly language of the relay computer itself and a bit of the overall architecture. This is really one of the most educational demo videos for vintage computing we’ve ever seen.
[Jhallen] assembled a few of these boards and he’s selling some of the extras. If you have $600, you can pick one up over on Tindie (standard Hackaday / Tindie disclosure statement). Considering the amount of soldering required to assemble this board, we’re going to guess that’s a very fair price.
Smart guy ! He uses DPDT, not SPDT relays :-D
At this moment, I finish a relays/diodes only hexadecimal display, using only Russian parts (because I can).
Because a relay computer using integrated circuits for display is not quite really a relay computer, right ? :-D
IME DPDT simplifies things greatly when doing logic. Combining the relays with semiconductor diodes helps even more by making wide* fan-in AND and OR gates less costly. It’s kind of historically accurate as semiconductor diodes were available at the same time relay computers were designed/considered.
(* wide is relative of course, wider than is affordable with pure relay logic at least IME, depends on a lot of things)
From what I gathered this year, through documentation and experiments…
Yes diodes were known for a long time (end of 1800s) but were small-signal, detector-type. It has taken a lot of time, and the slow progress of silicon manufacture in the 60s, to get *reliable* switching diodes, that could handle more current than a few mA, and at sufficiently low price. In the 50s apparently it was still cheaper to use a tube/valve/vacuum thing than a “solid state rectifier”. Germanium was low-threshold but Schottky can go lower and silicon has a much better stability, leakage and temperature range.
Better diodes changed the practicality of computers, reduced the size and cost, helped develop the field and industry…
We can buy 3K diodes for $10 today, yet we have no idea how critical it was for early computers in the past.
Without these cheap diodes, I wouldn’t even consider doing my own relay processors. Displaying a digit uses a large diode ROM, and they are critical for the DRAM array and the instruction memory. No diodes, no practical computer, only a ALU…
The LGP-30 computer from 1958 was a desk sized vacuum tube and diode based “general purpose” computer. It used 4096 32 bit words of drum memory coupled with about 120 tubes and a single logic board with several hundred diodes. It had 16 instructions, including multiply. The clock was derived from a fixed track on the drum, and the register display was a 3″ oscilloscope tube that showed the accumulator and program counter as square waves, since they were stored on the drum which was constantly rotating and being read out. The logic was all serial, it took quite a few clock cycles for example, to perform the multiply.
Wise words :) I’ve decided to make a complete, practical desktop relay calculator (with trig functions etc) with zero semiconductors of any sort, and it takes absurd amounts of relays, and lots of eBay sniffing for good deals – it’s basically a design driven by what was available on eBay, since I’d never wish to spend the list price for the parts I’m using (that’d be US$25k at the moment, I’m keeping track). So it’ll be a project that will be only be replicable if someone finds similar deals for similar relays, although I’m making it flexible voltage-wise: logic levels are 12V, but I’m using 2.0, 2.4, 5.0 and 12V relays – since that’s what was available for cheap.
Thought it looked familiar: https://hackaday.com/2017/09/11/relay-computer-you-can-hear-it-think/
The project so nice, they posted it twice!
https://www.youtube.com/watch?v=_GLzXVKlzUg
– Again?
– Again.
:o)
Claclaclac-click-clack-clack, the brainchoff is computing your request.
Ooooh nooo!
Buffer overflow.
Awesome channel, thanks!
Yes, it’s a crazy channel, glad you liked it!
Masterful level of troll.
Holy shit, how many times are you people going to post about this damn thing?
I seem to remember that one project has been posted 3 times already… Remember the Raspberry Pi Project ? :-D
wow nice project, like old computers in movies
Relays are satisfyingly clicky clacky, but do hobbyists ever make a CPU from discrete transistors these days? I’ve seen some CPUs made from 74 series logic chips, but that’s a level less connected.
I’m working on one, 4-bit, all transistor-resistor. Including the registers and a few bits of ram, the rom is dip-switch though. I’m about halfway done, but time is always slim, so it will still take a couple of months, I’m afraid.
Also, this guy:
https://hackaday.com/2017/09/10/hackaday-prize-entry-io-the-cardboard-computer/
We don’t know, you tell us.
http://bfy.tw/E0Za
Today is your lucky day, I felt like googling for you:
https://hackaday.com/2015/06/23/discrete-transistor-computer-is-not-discreet/
Hi Claude :-)
the short answer is “of course, because we can !”
more links : https://hackaday.io/project/6668 https://hackaday.io/project/7669 and I plan to make a processor with germanium, 80s generation and 2000’s generation transistors.
I know this was already posted recently. But I love this thing so much I don’t care.
Post it again!
Maybe we could have multi-author opinions on some articles.
MCU for mechanic engineers :)
I still love it. Is it cheaper than SIX HUNDRED BUCKS yet?
Hey I want to make some money on these things :-) I only made 10, so the contract manufacture cost is pretty high (through-hole assembly is not cheap).
What do you guys think about a kickstarter for fully assembled computers? I was thinking of making a wooden box for them (like a flat art box with a cover). This might get the cost down, but I’m not sure if there is enough interest.
Or do you think bare boards or kits would be better? A lot of soldering would certainly involved, but I’ve had a few people ask about this.
Oh, also I have six of the rev.2 boards left that I’m selling in eBay auctions. These are all broken, but could be made to work with debugging. Someone asked for one of these, so it gave me the idea to try to auction them off.
I completely understand the time, effort and cost that goes in to a small run like this. $600 isn’t massively expensive considering, although it does place the board in the ‘exuberant luxury tinker toy’ bracket, which, is way outside of most peoples’ budget. I guess it’s for the hard core. £100 – $130 is what I dare pay for something like this. The volume manufacture to achieve that sort of price would probably require so many units to be made that it would way over-saturate the demand for such a niche device.
Please link your ebay auctions (If that’s allowed on here) Keep up the good work and best of luck!