RC-3 Relay Computer

RC-3 Relay Computer

September 28, 2012 8:45 pm 6 comments

September 22, 2012, marked the one-year anniversary of an event that is near and dear to my heart: dedication of the RC-3 Relay Computer. I spent 21 months and more than 1,000 hours working on the design and construction of RC-3 while a volunteer at Goodwill Computer Museum in Austin.

The project began with the idea of building some kind of electro-mechanical relay logic demonstrator that would show folks how simple switching circuits could be used to, say, add two numbers. I browsed through catalog listings of thousands of models and selected a relay for the project, ordered a few, and started experimenting with some circuits. This was late in 2009.

RC-3 Relay Rail

RC-3 Relay Rail

About that time we saw an article in MAKE magazine describing how Dr. Harry Porter, a computer science professor at Portland State University in Oregon, had built an entire 8-bit computer out of relays. By some chance alignment of the cosmos, the relay I had picked for our project was exactly the same relay that Harry had used for his electro-mechanical computer.

Excited by the prospect of doing more with relays than a simple logic circuit, I contacted Harry and asked if we could build a computer based on his design. He replied that he would be delighted and agreed providing we gave him due credit. We kept his basic architecture, but made some changes to specific op-codes and instructions. We also used a completely different physical layout, building RC-3 in a single six-foot high equipment rack.

RC-3 Left Side

RC-3 Relay Rails Installed

We never would have gotten the project off the ground had it not been for discovering a skill that I didn’t know I had: begging. I contacted the manufacturer of the relay, and asked if they had any kind of charitable contributions program that might donate product to a worthy cause. To our great surprise, they said yes, and shipped us 450 relays (more than $3000 worth at then current prices). Other sponsors donated switches, cross-assembler software, a cabinet door, and other components.

I spent the first nine months creating drawings. Harry had built his machine entirely by himself, based on pencil sketches of the various circuits involved. We planned on using a team of volunteers to construct RC-3, so I opted for complete schematics for the entire machine. This would also aid in maintenance further down the road.

RC-3 Front Panel

RC-3 Front Panel

After building the first subsection (clock & sequencer) myself, I found it difficult to meet the quality standards I had set for construction, so abandoned the idea of having a large group of beginners do the soldering. Ultimately only three of us did the actual wiring, although we had volunteer help in other areas: printed circuit board design, machining mechanical components, front-panel layout, and software development.

The finished product is an 8-bit data, 16-bit address, machine using 418 relays for all functions except memory. It has eight general purpose registers, plus a couple special purpose registers, and runs at the blinding speed of 6 Hz. For memory we used a 32K RAM chip and a 32K EEPROM. Test and demo programs are stored in the ROM or can be entered into RAM via front panel switches. There are 421 LED indicators on the front panel, and 134 switches.

A full article on RC-3’s design and construction is now available at RC-3 Project.

You can see (and hear!) the RC-3 in operation below…

For your additional viewing pleasure we’ve also assembled some photos of the RC-3 on display at its dedication event last year at the Goodwill Computer Museum.

6 Comments

  • Dave Humphries

    Magic, Konrad Zuse would be impressed, with only 400 or so realys how good is your up time, I would guess pritty good.

    • Funny you should mention that! I recall we overcame an interesting bug with the relays “fusing” that Phil could speak on in more detail. Also I believe we based a portion of the RC-3’s design on Zuse’s work.

  • Stephen is correct; the adder circuit was designed by Konrad Zuse in 1941 for the German Z3 relay computer. All the rest of the circuits were designed by Harry Porter, with a few tweaks here and there by me.

    We did have a problem with relay contacts sticking in the clock circuit, which was determined to be caused by high in-rush currents through the contacts into the electrolytic timing capacitors. We added 4.7 ohm resistors in series with the capacitors and have had no relay failures since.

  • Hi there.

    I have been messing around with the idea of building one of these ever since I saw Harry P’s R.C.

    However as I am just an amature I have got as far as drawing quite a few of the circuit diagrams & building the clock circuit.

    I was wondering wether you would be willing to share any of your circuit diagrams ( esp the sequencer) as it is probably beyond my abilities to go much further as like I say this is just a back yard project for me.

    Thanks.
    Alex.

  • Second that ( to alex’s posting, above ). The availability of any formal schematics would be a great help, even in just the understanding of the principles behind these projects.

  • Full documentation for the RC-3 is now available on this site via RC-3 Documentation”. A complete article on design and construction of RC-3 is at RC-3 Project.

Leave a reply