Computers should sound like they're doing something

November 5, 2011 8:45 AM

Relays are simple electrical components that turn on an electromagnet with a small current to trigger a switch for a circuit usually capable of handling a larger current. For example, a relay can be used by a 5 volt DC microcontroller to turn on a 120 volt AC heating element in a rice cooker. Since relays can be used for logic, they can also be the primary components of gloriously clicky computers(see this for details on the last one).

Of note is that Zusie has an online simulator.

They're Turing-complete finite machines, so they can effectively do anything a more modern transistor based computer can, just much slower and with much less memory.

posted by mccarty.tim (36 comments total)

re: the last one -- it has a good beat, you can dance to it.
posted by empath at 8:50 AM on November 5, 2011 [1 favorite]

I like to think the last one is a choo choo train for the bits, based solely on the sound it makes.
posted by mccarty.tim at 8:53 AM on November 5, 2011 [3 favorites]

When I was a tech at Chuck E. Cheese's and working on the internals of some of the machines there I grew very fond of the *kachunk* of the relays. Like when you put in your token to skeeball and *kachunk* then the satisfying sound of nine balls rolling toward you to start your play. Pretty much any mechanical game had a relay in it and that sound was fantastic.
posted by Phantomx at 8:55 AM on November 5, 2011

This is great, thanks!

I wonder when the point will be reached when not only computer scientists but even electrical engineers will not automatically be taught and well understand how these, and particularly Porter's Relay Computer, work. Or will that never happen?
posted by Ivan Fyodorovich at 9:06 AM on November 5, 2011

Or will that never happen?

It's unlikely to happen since it's a really, really basic concept. You could literally build a relay in a few minutes with some bits of metal and wire.
posted by odinsdream at 9:32 AM on November 5, 2011

Clearly, I'm obligated to write something, but I have no idea what ...
posted by Relay at 9:40 AM on November 5, 2011 [5 favorites]

It must be gratifying to build a computer that you can watch (and hear) computing. The projects are so delightfully large, impenetrable, and noisy - they define geek. These guys must have tolerant families!

My dream is to build a touring complete, programmable all-mechanical computer. It can be done, but it's even harder than using relays.
posted by Popular Ethics at 9:41 AM on November 5, 2011

I guess I should note that not all relays are mechanical. For example, solid state relays are widely used, which, if I understand correctly, are basically huge optoisolators.

Basically, they are a light-sealed chamber with an LED that runs on small amounts of DC next to a photoresistor. The photoresistor has very high resistance when it has no light, to the point that it's nearly an open circuit. Light it up with the LED, and it has almost no resistance.

They have benefits, like requiring less current to trigger (a lot of microcontrollers need a transistor to control a larger current to trigger a relay) and not having a nasty effect where the current goes backwards when the relay is opened (which can be fixed with a diode or an optoisolator, but can blow a microcontroller or other sensitive component otherwise). I think SSRs can also be switched on and off faster, and last more cycles (number of times they are switched).

But mechanical relays are significantly cheaper, and they make a satisfying click.

This is based mostly on my quick research for a recent relay-using project, so I could have it all wrong.
posted by mccarty.tim at 9:50 AM on November 5, 2011 [2 favorites]

Analog rules when calculating complex systems with simple hardware. Video of the "Moniac", a hydraulic computer that calculates finacial results of various policies:

The MONIAC was approximately 2 m high, 1.2 m wide and almost 1 m deep, and consisted of a series of transparent plastic tanks and pipes which were fastened to a wooden board. Each tank represented some aspect of the UK national economy and the flow of money around the economy was illustrated by coloured water. At the top of the board was a large tank called the treasury. Water (representing money) flowed from the treasury to other tanks representing the various ways in which a country could spend its money. For example, there were tanks for health and education. To increase spending on health care a tap could be opened to drain water from the treasury to the tank which represented health spending. Water then ran further down the model to other tanks, representing other interactions in the economy. Water could be pumped back to the treasury from some of the tanks to represent taxation. Changes in tax rates were modeled by increasing or decreasing pumping speeds.

Savings reduce the funds available to consumers and investment income increases those funds. The MONIAC showed this by draining water (savings) from the expenditure stream and by injecting water (investment income) into that stream. When the savings flow exceeds the investment flow, the level of water in the savings and investment tank (the surplus-balances tank) would rise to reflect the accumulated balance. When the investment flow exceeds the savings flow for any length of time, the surplus-balances tank would run dry. Import and export were represented by water draining from the model, and by additional water being poured into the model.
posted by 445supermag at 9:57 AM on November 5, 2011 [4 favorites]

"It's unlikely to happen since it's a really, really basic concept. You could literally build a relay in a few minutes with some bits of metal and wire."

Well, yes, I'm aware of that. I meant the very basic components of digital computing and its most essential operations.
posted by Ivan Fyodorovich at 9:58 AM on November 5, 2011

My latest encounter with relays was assembling an older Makerbot kit. They had an issue with the DC motor which would draw a huge amount of current in certain motor states and lock up the PWM controller. The solution was to replace the the PWM controller with a relay which would take no shit from the motor and pummel it into compliance with its superior amperage.
posted by RobotVoodooPower at 10:13 AM on November 5, 2011 [1 favorite]

Well, yes, I'm aware of that. I meant the very basic components of digital computing and its most essential operations.

It's a neat way to teach the fundamentals of logical circuits when you can demonstrate something that can literally be seen in action. Demonstrating transistors requires a fair bit of explanation, whereas you could sit a relay on an overhead projector.

The fact that it's a good teaching tool combined with the mechanical simplicity is why I don't think they're likely to be forgotten.
posted by odinsdream at 10:17 AM on November 5, 2011

This is great, thanks!

I wonder when the point will be reached when not only computer scientists but even electrical engineers will not automatically be taught and well understand how these, and particularly Porter's Relay Computer, work. Or will that never happen?

What are you asking exactly? Many computer scientists already have no idea how a relay works, in the same way that many authors have no idea how a printing press works. It depends on who you count as a "computer scientist", but you can get a PhD in the subject without ever once discussing a mechanical relay.
posted by tylerkaraszewski at 10:18 AM on November 5, 2011 [1 favorite]

mccarty...you beat me to that point exactly. But I would add that some of those solid state relays are basically big power FETs with extra isolation that's not necessarily optical. The one advantage of mechanical relays is that when they are off, they are truely open circuited off. SSRs usually give you some large number of mega or giga ohm effective resistance, but they're not off in the same way. They also usually have some parallel capacitance that can bone you at high frequencies.

And there are also SSRs that will do continuous high duty power modulation based on some analog signal in, basically doing a lot of auxiliary work for you.
posted by Chekhovian at 10:20 AM on November 5, 2011

Zusie has Blinkenlights! So cool.
posted by Wemmick at 10:49 AM on November 5, 2011

Needs more Nixie tubes.
posted by Jumpin Jack Flash at 10:58 AM on November 5, 2011 [2 favorites]

How come this post has no mention of Redstone?
posted by rebent at 11:01 AM on November 5, 2011

See the tags.
posted by mccarty.tim at 11:05 AM on November 5, 2011 [2 favorites]

My god, could he have used larger relays? There are really small board-mount reed relays that are very quiet, but I guess since this is a show piece that wouldn't have made any sense.
posted by cman at 11:13 AM on November 5, 2011

CSB:

One thing that I valued from my Comp Sci education at Washington State was that we had to build digital logic circuits as part of our curriculum. One lab had us build a resistor ladder as a D/A converter(I know, not digital). We took the parallel port off our computer and slaved an oscilloscope's X and Y components to it, turning it into a primitive monitor. I built a wireframe 3D rendering engine a la battlezone as extra credit. Damn that was fun.

FYI sparkfun has some good Arduino kits for some good ol' hardware hackin.
posted by The Power Nap at 11:17 AM on November 5, 2011

WAY back in the early 80s, I worked for a company that manufactured specialized relays, handling data for Materials Control (what they needed to make them) and Inventory Control, with a computer that was NOT interfaced with the Accounting system (I was kind of a human interface, and was laid off when they got their shit and systems together). They were trying, not always successfully while I was there, to move from mechanical relays to solid state relays, but some big honkin' mechanical relays couldn't be made smaller (we laugh at you, Moore's Law). At one time, they made the relays that turned on almost every bank of stadium lights in the NFL. Now THAT is a kerchunk!
posted by oneswellfoop at 11:32 AM on November 5, 2011 [2 favorites]

I work at a large, old manufacturing facility where we have control systems of various ages. Everything from pneumatics (now there's an awesome mechanical analog calculating system) and relay logic, all the way to the latest distributed control systems, with a little bit of everything in between. Ladder logic is a programming environment for programmable logic controllers that duplicates the old electrical prints for relay logic. That way engineers and electricians could program and troubleshoot like they always did, and not have to learn some new odd language. Other systems were "blockware" using logical blocks more similar to the old Norpack systems. I liked troubleshooting relay logic systems even though it meant more footwork. Something about a physical device that you can measure easily and hold in your hand that's satisfying.
posted by Eekacat at 12:40 PM on November 5, 2011 [1 favorite]

My antenna tuner uses latching relays to switch in various coils, resistors and whatnot. Much clickiness ensues if I'm tuning a band I haven't used in a while.
posted by scruss at 1:08 PM on November 5, 2011 [1 favorite]

Relays are big, expensive, slow, power-hungry, and they wear out. We just keep 'em around to make transistors look good.
posted by ryanrs at 1:31 PM on November 5, 2011

...the "Moniac" , a hydraulic computer that calculates finacial results of various policies

Very cool. I'm guessing this was the inspiration for Terry Pratchett's hydraulic computer in "Making Money".
posted by jcreigh at 1:34 PM on November 5, 2011

I think the simplicity of relays covers great depth and complexity, odinsdream, which becomes inescapable on further examination, though that simplicity might be essential for the tremendous range of applicability of relay systems.

Claude Shannon got his start by contemplating relays, not so incidentally:

A paper drawn from his 1937 master's thesis, A Symbolic Analysis of Relay and Switching Circuits,[7] was published in the 1938 issue of the Transactions of the American Institute of Electrical Engineers. It also earned Shannon the Alfred Noble American Institute of American Engineers Award in 1940. Howard Gardner, of Harvard University, called Shannon's thesis "possibly the most important, and also the most famous, master's thesis of the century."

And I think it could be argued that thinking about the smallest possible signal that could actuate a relay with a large macroscopic effect could get you a long way toward A Mathematical Theory of Communication, as well.

As I was reading this thread it occurred to me that genes and their promoters (and their RNAs, and the interactions of those with ribosomes, and etc., etc.) could probably be modeled in an productive way with systems of relays, so it was interesting to read in the Shannon link that:

Vannevar Bush suggested that Shannon, flush with this success, work on his dissertation at Cold Spring Harbor Laboratory, funded by the Carnegie Institution headed by Bush, to develop similar mathematical relationships for Mendelian genetics, which resulted in Shannon's 1940 PhD thesis at MIT, An Algebra for Theoretical Genetics.[8]

A 'road not taken' there, I'd say, and one that someone should take.

It's interesting all this talk about AI and whether a true artificial intelligence would necessarily have to be conscious.

I guess the answer is no, because I think genomes will be seen to be AI-like entities which are not conscious.
posted by jamjam at 1:39 PM on November 5, 2011 [1 favorite]

Yay, relays.

The first part of my career (late 70s) was as a broadcast technician. At the station where I apprenticed, we would often build simple controllers to route signals or remotely switch various and sundry units like transmitters, VTRs, film projectors (telecine) and so on, and the device of choice was a particular Potter and Brumfield relay (KH series I think, 24 vdc coil, 4P2T contacts), that we consumed for custom projects at the rate of a few a month.

Relays are still useful components, and one of the first things I extract when plundering discarded electronics. Nonetheless, as mentioned by tim, I suspect that mechanical relays will ultimately be completely replaced by solid-state relays, simply because of ease of manufacture, but also because of greater efficiency and longer life.

Just last weekend I made a controller for our 4-pump hydronic heating system, using 4 relays salvaged from old equipment. I also ebayed some OMRON-clone time-delay relays for the system.

-click-
posted by Artful Codger at 1:41 PM on November 5, 2011

On my speakers that last one sounds like a steam locomotive.
posted by telstar at 2:30 PM on November 5, 2011

It was the clicking of a relay that allowed me to diagnose a recent fuel pump failure. In early September of this year, I was in Utah riding dual sport motorcycles (on-and-off-road bikes) around some very isolated areas. We were riding pretty fast down a nice gravel road when my bike sputtered once, twice, then died.

Three things can make an infernal combustion engine die: fuel, air and spark. The battery was good (headlight was strong and the starter motor cranked well) and the problem was too abrupt for air (it's not like my air filter suddenly gave up and quit). That left fuel.

When I turned the key, I could hear the clicking of the relay but couldn't hear the whirring of the fuel pump. That told me that the wiring between the relay and the fuel pickup was bad. Piece of cake, pull the fuel pump out of the tank, snip snip crimp crimp and back on my way, as the locals watched.

If relays didn't make that clicking sound, I would have had to troubleshoot the entire ignition and fuel system. So hooray for clicky relays!
posted by workerant at 3:06 PM on November 5, 2011 [2 favorites]

"A motor-driven clock oscillator will later be mounted in the big hole in the front panel"

\m/
posted by vanar sena at 3:16 PM on November 5, 2011 [1 favorite]

At work, I sometimes deal with pneumatic components that are basically the air version of relays and transistors.

Since my training is mostly electrical, it wasn't that surprising to me that people went and made logic circuits out of that stuff.
posted by ArgentCorvid at 3:17 PM on November 5, 2011

Ahem -- made logic circuits
posted by ArgentCorvid at 3:28 PM on November 5, 2011

Oo, I thought of something to say (finally)!

Computers should sound like they're doing something

I agree, this is why I like Curta Calculators.
posted by Relay at 3:30 PM on November 5, 2011

Small fry…

An Ericsson ARF Telephone Exchange in operation.

Each of those grey boxes in the racks? Literally packed with relays.

That clattering noise in the background? Tens of thousands of relays operating all the time, switching telephone calls.

And that was just to set up & clear each call. The actual call paths, the connection from phone A to phone B, were switched by crossbar switches - essentially, complicated relays - which can be seen on the left at around the 40 second mark.

Old telephone techs laugh at your little pinball machines, switching matrices, process controllers, etc…
posted by Pinback at 4:05 PM on November 5, 2011 [2 favorites]

Once when I was in boy scouts we were taken to a Bell System Central Office and given a tour. It was a Gymnasium sized building filled with racks and racks of clicking relays and smelled of machine oil.

I get the feeling now it's all been replaced by a filing cabinet sized computer.
posted by bottlebrushtree at 9:26 PM on November 5, 2011

Wonderful, thanks for posting!
posted by carter at 4:27 AM on November 6, 2011