I’m reading about the history of computing, and came across the Wiki page on electro-mechanical relays.
This page says (under “applications”):
I have never heard of that, but it sounds plausible. Is this true? Does this mean that in nuclear facilities, there are actually computers which use relays instead of integrated circuits for logical operations? If yes, how large are these machines, what is their computing power compared to ordinary systems, which software do they run on? Or do these devices consist of just of few relays which perform simple functions, “computer” being too big a word for them?
What fascinates me about this thought is the fact that both Boolean logic and the technology to build relays were available in the middle of the 19th century. These machines the Wiki page refers to could be a model of what a Victorian-era computer could have looked like. Great stuff for alternate history writers.
This is about right; they aren’t terribly complex, comprising a few to a few dozen relays, typically. Really, they’re just simple collections of a few basic gates.
I remember some of my Navy training manuals discussing the field of fluidics and I was totally intrigued and captivated by the idea that one could make a digital computer using nothing more than compressed air and specially shaped tubes and chambers.
My Navy book seemed to indicate more purposes for the technology than the Wikipedia page does, specifically applications in high temperature environments. The Wikipedia page is downright pessimistic about usage of fluidics.
I’d love to have a basic four-function calculator made using fluidics. Probably would be the size of a refrigerator and would need a good sized compressor to keep it going.
>the idea that one could make a digital computer using nothing more than compressed air and specially shaped tubes and chambers
One of my father’s patents was for a pneumatic switch having no moving parts. It became popular in some kind of breathing apparatus - I always thought it was respiratory therapy equipment, but maybe it was underwater rebreathers or something. I have the model he built out of layers of colored and transparent plastic - acrylic, I guess - and sent to his mother, who gave it to me years later.
it sounds like that article is referring to the basic control logic, the kind of stuff that requires basic switches to operate which in most cases is done with transistors, but in a high radiation or EMP environment would not function correctly. Not an entire number-crunching computer.
A relay uses an electromagnet to to move physical contacts, instead of semiconductor which is more likely to fail with loose energy partials flying around.
>both Boolean logic and the technology to build relays were available in the middle of the 19th century
This is pretty remarkable in hindsight. Apparently, for more than a century computers (as machines) were envisioned and acknowledged as a possibility but their utility was not taken seriously.
BTW it’s not the application the OP considers, but often in data acquisition small “reed relays” are used to switch signals. Solid state multiplexers in integrated circuits, and other purely electronic methods, are used more often. But none of them can approach the 15 or 18 or so decades of dynamic range in conductivity offered by relays.
>Of course somebody of the name of Napier must know really well about the history of computing
You bet. I invented the two basic elements of the slide rule and never thought to get them to work together. This may have been the first example of vaporware. But have you seen my work demonstrating that the Pope is the Antichrist?
Ladder logic (aka relay logic) is pretty simple stuff, and was designed for this sort of application. And, I’d much rather have my safety in the hands of a box of relays than anything running Windows.
Instead of inputs to an electronic computer, it’s a string of hard contacts that all have to be lined up for something to happen - as an example, you want to control whether or not the control rods can be withdrawn to allow the reactor to run.
Is the cooling pump running?
Is the coolant cool?
Is the steam plant online?
Is the door closed?
All of these must be true. In ladder logic, each of them is represented by a physical switch. A pressure switch to tell if the pump is running, a temperature sensor to tell if the coolant is cool, and so on. Because it’s all physical, it’s almost impossible for a bad bit of programming to louse things up. Bad design or bad installation, on the other hand, can cause problems, but these systems have been around for decades, so they’re well-understood, and such problems will most likely be found in the initial testing.
How do you know there isn’t a glitch in the programming of a computer that it won’t decide it’s OK to start the reactor even though the pump is not running, but it is 12:10 PM on September 14? Or someone wrote a logic bomb? Or someone walked across the carpet, touched the keyboard and introduced a static shock that momentarily kicked all inputs to true? The only way to set a logic bomb in relay logic is to manually get at the wiring and move wires, and other than the electromagnetic pulse from a nuclear detonation, relays are immune to pretty much any sort of power surge or interference.
gotpasswords, on a slight tangent from relay logic, for every electrical system of this nature there must be a few totally mechanical devices that accomplish a complex task via pressures, temperatures, springs, etc.
In one of the plants where I worked, one of the cabinets that controlled water flow worked on totally mechanical principles: there were little bellows attached to various points along a series of see-saw levers. Each bellows received a different pressure or temperature input, and the interaction of a few of these levers, with a handful of different bellows at specially chosen points along the length would result in positioning a valve that would produce an output signal.
It really looked like a Rube Goldberg contraption, with a glass front on the cabinet. But it worked quite nicely, providing a smoothly damped feedback loop that maintained the desired differential pressure.
An engineering friend mine was in a meeting. Somebody was proprosing some really overly complicated, expensive, and needlessly precise way of doing something that should have been simple.
At some point, he voiced his complaint (probably in a very polite way, thats just the way he is), and noted that this was a “Rube Goldberg” solution.
The big boss jumped all over his case for being anti-semitic…and no amount of splaining could convince the big boss otherwise.
Good thing the presentation wasnt brilliant, for if he had proclaimed “Bob, your an Einstein”, there is not telling how much trouble he might have got in.
