Mechanical Computing

FTL writes "Tim Robinson has built a computer capable of solving polynomial equations -- using Meccano. His difference engine (mirror) uses a similar approach to Babbage's design. He's also created a differential analyzer (mirror) complete with a GUI. Both could be scaled up indefinitely to handle larger problems. 'Computing by steam' is possible."

Comments

Tinkertoys

[Tablizer]

Don't forget about the Tinkertoy computer

Rod Logic

[3Suns]

Reminds me of the nano-scale "rod logic" used for computation in Neal Stephenson's "The Diamond Age". Those were rods with bumps on them arranged in a 3d grid, and as the were moved back and forth the bumps somehow performed computation.

Re:Rod Logic

[MarcQuadra]

That's actually probably how nanoprocessors will operate, getting electricity to stay where you want it on the scale is a lot harder than using nanotubes with rods suspended in them. Of course, there will be actuators at the 'leads' that are electrical or light-actuated.

Next project?

[cjellibebi]

I recon this guy should try to build a machine that uses computational logic gates (NAND, NOR, etc). From that, he can build up things like binary adders and simple flip-flops. Then, add an instruction-decoder, and an arithmetic-logic unit - and viola - a Meccano CPU.

My first computer

[Registered Coward v2]

was a mechanical rig that used 1 inch soda straws for 1's and blank holes fro zeros. You pulled a crank and it added two numbers. I wish I could remember its name - it was some sort of "science kit." from the 60's.

Re:My first computer

[kfg]

I've been trying to remember the name of that puppy too. I had one and it was fascinating (if you're the sort of kid who also takes apart those old, plastic, push button, gear driven adding machines they used to sell in the grocery stores to see how they worked. Mechanical computers were actually nearly ubiquitous in the 60s).

Anyone with 60s comic books should be able to find an ad for one in the back, right next to the 6 foot long fiberboard submarine.

I never had one of those. I still blame my mommy.

KFG

Re:My first computer

[kfg]

It was card programed, but not punchcard programed. You had a plastic card with "teeth" on one side. Sticking a bit of straw on one of the teeth was a one. A tooth without a straw on it (a "hole" between the straws) was a zero. You ended up with a gap toothed "comb." The straws were just a way to make cheap pegs.

The answer came out in binary formed from stickers of little white and black squares stuck to tab ends of cards.

KFG

Re:Beautiful!

[linzeal]

I have an abacus that hangs on my wall and it is always fun to see people use it, or at least try to. I learned how to use an abacus in 2nd or 3rd grade I think and the tactile sensation I think helped establish the immediciacy of mathematics in my own physical space for me.

Obviously it is.

[Valar]

Of course mechanical computation is possible. The easiest example I can think of is division/multiplication. Two gears, the ratio of which is the multiplier. Turn the first gear a number of turns equal to the multiplicand and count the rotations of the second gear.

Re:Next thing to do..

[GridPoint]

Linux probably would a bit too large to fit, but a port of Contiki might be worth waiting for, given their track record... Now if only someone would care to make a mechanical Ethernet NIC and we could build a fully mechanical webserver. You wouldn't be able to stand the noise a slashdotting of that would make!

Reminds me off the great novel by Bruce Sterling

[amarodeeps]

...and William Gibson, called The Difference Engine. I recommend it, it's a fascinating idea, which is basically: what if computer became available much earlier, in the form of mechanical computers--they would take up entire factory buildings, and people would essentially become experts at creating these ornate ivory punchcards (if I remember correctly...). Actually, I should really pick it up and read it again.

Re:Next thing to do..

[cjellibebi]

I think a more realistic next step would be to build a CPU out of logic-gates. See my other post for how this could be accomplished. After that, build masses and masses of flip-flops, and multiplexers so you can access (2^A)*D of them using A address-lines and D data-lines. Now you have storage, so you can run stored programs. Using an existing computer which has GCC ported to it, write a GCC back-end for your new CPU, and then compile Linux on it.

Of course, you would have to modify this port of Linux to take into account how the Meccano compouter handles IO, etc. Add a means of networking, and you can turn it into a Linux server. Add user Input/Output, and you have a workstation.

Mechanical PDAs

[scattol]

That's OK, but for the man of the 50's on the go what do you do? You get a CURTA calculator. They were said to be popular with rally drivers for instance.

They are relatively valuable and pretty nifty calculator. You can try to get a feel for it with the simulator. Enjoy!

Re:Next thing to do..

[cjellibebi]

I'm not really sure if Contiki really is smaller than a Linux kernel can ever become. Contiki has been built mainly for 6502-based systems. What I've heard is that the reason there is no back-end for GCC that produces 6502 code is because the 6502 only has a 256-byte stack, and for reasons unknown to myself, GCC has a problem with this (I'm not sure if Linux also has this problem). Contiki has been built with the CC65 C compiler for 6502's compiler instead. So if the Meccano computer does not have this limitation, then Linux could run just as well as Contiki.

Contiki/Linux just needs to be compiled on a real compiler with a back-end that produces code that the Meccano CPU can run. As for which OS to try out, try and compare the size of a Contiki kernel to thet of a Linux kernel, and go with the smallest (question: Is there a size comparison of the two kernels on a machine that is capable of running both OS's?). As all the flip-flops for memory/storage will have to be built by hand, it would make sense to try the smallest OS on the machine. I suspect that Contiki would be the smallest, but I am not sure if Contiki can run as a server OS, so it would be useless, unless you also made user Input/Output devices out of Meccano as well.

square roots mechanically

[Alien Being]

My dad was selling those newfangled electronic Friden calculators back in the mid sixties when I was a little kid. Once in a while, he'd bring home one of the old mechanical machines that had been traded in, like this one. I'd love to have one of those babies now.

The Antikathera Mechanism

[bsDaemon]

big deal. This Thing is over 2,000 years old! It's an astronaumical computer and clock thingie. With gears and everything. "boo ya," as the kids say.

Re:Next project? The Analytical Engine!

[WegianWarrior]

Charles Babbage gave up on the differiensial engine because he thought he could build a general purpose mechanical computer - the Analytical Engine! To recreate that device would really rock... if I had the time, money and (last but not least) the knowhow, I might try it myself...

Off course, an Analytical Engine would be way larger than a Difference Engine , since it would have to include a CPU (the 'mill'), a input device (Babbage himself suggested punch cards - an idea which the early electromechanical computers picked up), an output device (Babbage wanted to built a complete, automated printingpress, curveplotter and a bell to alert the operator of errors),and last but not least a 'store' (memory - the one envisoned by Babbage would store 1000 numbers, each 50 digits long). The Analytical Enginge was to be programable - which was it great strenght compared to the Differensial Engine - in a language resembling todays assembler languages. Such a machine would be slow and lowpowered by our standards, but would have been a gigantic leap forward back in the 1830's... shame he never got around to build it.