h a l f b a k e r yWhat's a nice idea like yours doing in a place like this?
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Computers are largely made of
transistors which switch from one
state to another, rather like points
on a railway, and thus it would be
possible to simulate a computer in a
(very large) Hornby 'OO' train set.
Trains, representing data, would pass
down tracks and be diverted by points
and
would, as they pass down the track,
flip trackside levers which would
further alter the state of other sets
of points, and so on. A medium-sized
domestic train set could be used to
simulate a NAND gate and many of these
would be used to build up more complex
components like adders, etc.
"The Enemy" by Desmond Bagley
https://astrofella....emy-desmond-bagley/
Paragraph 9 [8th of 7, Jan 26 2017]
MIT railroading clun comes up with "hacker" and information wants to be free
https://en.wikipedi...Model_Railroad_Club
[beanangel, Jan 27 2017]
http://www.thingiverse.com/thing:1720219
[xenzag, Jan 27 2017]
Another take on the general idea
Mechanical_20Internet
[doctorremulac3, Jan 29 2017]
Shunting-yard algorithm
https://en.wikipedi...ting-yard_algorithm
WKTE [8th of 7, Jan 31 2017]
City And/Or Photo As CPU
City_20And_2fOr_20Photo_20As_20CPU
Traffic FETs [8th of 7, Jan 31 2017]
MONIAC
https://en.wikipedia.org/wiki/MONIAC
Very clever stuff [hippo, Feb 03 2017]
[link]
|
| |
Yes, definitely! I'm trying to imagine the logic gates now.
Would an inverter be like a yard full of carriages which
were added to a train if and only if there were no trains
coming? Could you represent zeroes by empty trucks and
ones by full ones? |
|
| |
//Would an inverter be like a yard full of carriages which were added to a train if and only if there were no trains coming?// |
|
| |
You'd need some kind of clock signal to make sure you inserted the correct number of trains. |
|
| |
It might be better to represent a logic signal as a _pair_ of tracks. A 1 is represented by a train on one track, and a 0 is represented by a train on the other track. |
|
| |
[8th] is going to get all moist over this one. |
|
| |
<whimpering and twitching> |
|
| |
There's a spy thriller that uses one of the protagonist's model railway story as a plot device. |
|
| |
MIT model railroading club ...around 1946!-50's [link] |
|
| |
although it does not look like they made linked logic gates from their trains, only very much sort of. |
|
| |
Also, did you know the .5b gets about 30,000 - 40,000 views a month from search engines. electric blanket ideas are popular. |
|
| |
That's a clocked flip-flop, not an AND gate. Ideal logic gates present the instantaneous result of their inputs and function at their output. |
|
| |
We consider it likely that you don't understand. The precise extent of your lack of understanding has not been defined, but is estimated to be very, very large. |
|
| |
An artist friend of mine made a mechanical gate based
logic computer at Manchester Art college in the 60s. Last
link doesn't show it, but it's a really interesting example.
Trains should work equally well. + |
|
| |
// There are propagation delays // |
|
| |
Yes, we know. Hence the use of the adjective "ideal", referring to a theoretical model of a "Perfect" gate, which can not actually exist. |
|
| |
This would be a great way to suppress Skynet's progress. It'd have to figure out a way to change the points. |
|
| |
While integrating a train into the machinery of a
mechanical computer system is possible, I don't
know if the components currently exist in a
Hornby 'OO' train set. Specifically, while there are
switches that can select which track a train goes
on, I am not aware of a mechanism by which a
passing train can change the state of another
switch. I'd like to be shown otherwise since
building a computer from standard parts in a toy
that has been available since 1938 would be very
interesting. |
|
| |
Yes, I don't know exactly what the mechanism
would be for a passing train to trigger a set of
points to change but it seems like it shouldn't be
too complicated |
|
| |
If the points aren't locked (or driven) then passing through them in the "wrong" direction will change their state. |
|
| |
In the early days of railways, "falling" or "slip" points were extensively used at the exit from passing loops. Points coud be weighted to lie normally for the through line, but a train from a branch could simply push them open. |
|
| |
However, when such points were "facing" to traffic, numerous accidents resulted, leading to the introduction of "facing-point locks". This is a two-lever system requiring the signalman to follow a defined sequence of lever movements, unlocking, moving and re-locking the points, to allow the relevant signal to be cleared. |
|
| |
The Borg's knowledge on these matters is clearly extensive. Out of curiosity, did you keep the anorak and dandruff after the assimilation? |
|
| |
<waits for Greater Anglia rail to announce that the 8:45 to Liverpool Street has been delayed due to Boolean failure> |
|
| |
// did you keep the anorak // |
|
| |
// and dandruff after the assimilation? // |
|
| |
We have never experienced that particular affliction ... |
|
| |
// 8:45 to Liverpool Street has been delayed due to Boolean failure // |
|
| |
"This is a customer announcement. It appears that the 14:42 from Oxford Clarendon has entered a state of Quantum Indeteminacy in the vicinity of Didcot. We are unable to give an expected arrival time for this service as we cannot simultaneously determine the velocity and location of the train; further, it is also impossible to resolve if the train will leave Didcot, or Didcot will be moved by the train. Our emergency Physics crew is preparing to address the problem, just as soon as they've decided on which algebraic notation to use... " |
|
| |
Sounds like an animation would be easier to make but maybe
less interesting to look at. |
|
| |
Then again, this could be made as a school or university
project with engineering and computer science departments
assisting each other in a national competition. [+] |
|
| |
// maybe less interesting // |
|
| |
Definitely less interesting. |
|
| |
From an academic viewpoint, see <link> |
|
| |
Simple NAND gates are asynchronous, so it would have to be the presence of two or more trains triggering an output state change. The trains would have to be running on parallel tracks, and stop at a signal. This would release a third train to be the input to another logic element. |
|
| |
Real-world gates have fan-out, so there would have to be a way to route multiple trains. |
|
| |
Using points as "memory" is interesting, as they are a quantum state; they relate that a train has passed through previously, not that a train is immediately present. It becomes impossible to predict the arrival of a train, thus confirming the macroscopic observations of the behavior of Virgin Rail services. |
|
| |
This also leads to the concept of the Schrödinger train, which is fixed by the act of observation; no point in looking at a timetable, the only way to see if there's a train is to go to the station and wait until one appears. |
|
| |
Now, that's strangely familiar ... |
|
| |
// even whimsical ideas need to be vaguely workable. // |
|
| |
So that means that the logic needs to be bit-serial using the wagons as the 0's and 1's - full waggons and empty ones, the loco being the start bit and the guards's van (caboose for the uneducated) being the stop bit. |
|
| |
4-bit operation should be easy to implement. |
|
| |
Now ... are the trains a representation of the data (words, opcodes) or the actual data items ? |
|
| |
A 4-bit parallel bus would require four tracks; a loco "present" represents a logic "1". Could be done ... but probably not synchronously. |
|
| |
The way I first thought about this idea was that you
wouldn't try and replicate logic gates directly but instead
would use trains and points to model transistors, and then
build up a gate, such as a NAND gate, out of multiple
transistors. So if a train (i.e a 1) went down the line
representing the 'base' terminal of the transistor, this would
switch a point to allow any trains to pass along the track
representing the collector-emitter line. If this point wasn't
switched, these trains would have to be sent to a train
recycling branch line... |
|
| |
The problem there is the train needs to pass continuously to sustain the logic "1" if the transistor analogy is to work, representing the flow of electrons (or holes) into or out of the base. |
|
| |
[IT] 's point about FETs - particularly IGFETs (or indeed thyristors, or UJTs) is an interesting one. <link> |
|
| |
//the train needs to pass continuously to sustain the logic "1"// That depends, shirley, on which architecture you're modelling? Some architectures approximate to an "ideal" system in which current only flows while the system is being charged to the new voltage - leakage is then low. So, a siding holding a train is sort of a perfect memory store. |
|
| |
"tristate" was a somewhat poor choice for a name, because the less-perceptive sometimes misinterpret it as meaning "trinary". |
|
| |
The three output states are 0, 1, and "invisible". There's got to be a name more catchy and precise than "tristate" ... |
|
| |
For a good time read a bit of steampunk scifi called "The Difference Engine" that postulates a mechanical computer big enough to develop self awareness. :-) |
|
| |
Thinking about this more, it may be that a slightly
different type of logic is required of a
model-train-based computer. This is analagous to
water-flow analogue computers such as MONIAC (see
link) which have a logic that is not quite the same
as traditional transistor-based computers. |
|
| |
MONIAC, and similar machines, are surprisingly
sophisticated when you look at the maths they're
doing. They can, for example, solve multiple
simultaneous differential equations (obviously far
beyond the abilities of electronic computers at the
time MONIAC was made) so some of the additional
parameters you mention might not be beyond its
abilities. Also, a good point about the information
flow in the circuitry also being the visualisation -
you might get this with trains too. |
|
| |
This entire idea is wrongly named. Either that, or the name is right and the entire idea is wrong. |
|
| |
Shirley it should be "Train set computer"? A "Computer train set" would be some sort of railway simulacrum built from computers. |
|
| |