Mechanical logic gates

[Swatopluk]

Does anybody here have an idea how to construct a purely mechanical XOR gate (preferably without springs involved).
It should function in the way of the sketch seen below. Alternatively the input pins could also rotate into position* instead of moving forward and back and the output pin could be at right angles to the input pins.
Magnets or anything electric is out. Pure lever/push actions would be preferred.

*pin wheel, cogwheel

OR, NOT, NOR are easy.
AND and NAND I am not sure about. Any ideas?

[ivor]

Here's one made out of LEGOs.

http://goldfish.ikaruga.co.uk/xor.html

[Sibling Zono (anon1mat0)]

That doesn't look easy!
--
It reminded me school days playing with gates to make an specific circuit (and adding machine). My respects to the builders of those huge adding machines of the turning of the century.

[Swatopluk]

Too complicated for my purpose. Especially too many necessary parts.
And as one can see from the video, it would be close to impossible to make a sequence.

But thank you for the interesting link.

[Bob in a quantum-state-of-faith]

Does anybody here have an idea how to construct a purely mechanical XOR gate (preferably without springs involved).
It should function in the way of the sketch seen below. Alternatively the input pins could also rotate into position* instead of moving forward and back and the output pin could be at right angles to the input pins.
Magnets or anything electric is out. Pure lever/push actions would be preferred.

*pin wheel, cogwheel

OR, NOT, NOR are easy.
AND and NAND I am not sure about. Any ideas?

Now that you got me thinking about it, I'm intrigued.

May I ask, what it's for?  That is, I need some idea of how robust it needs be (what sort of forces are we talking of, ounces?  Grams?  Pounds? Many-pounds?).  And why the prohibition with springs...any particular reason?  (curious, is all).

More questions:  is gravity an acceptable orientation force, or will the thing be rotated/moved about?  (I was thinking of dropping levers and cams, here... )

((now, I'm thinking of a hydraulic system, with valves.... naaaah. ) 

[Swatopluk]

Gravity as orientation is OK (especially, if no springs are allowed).
Output pin should be able to exert a force enough to either operate a mechanism behind it or to release another power source. I think 1-2 pound should be sufficient.
The mechanism should be crude and simple enough that it could have been built by a medieval mechanic. Springs were not that common at the time and would be difficult to get.
I am working on just another of these fake Byzantine cipher machines, in this case a mechnical one-time pad.
If the letters are encoded binary then the XOR is the easiest reverse operation
If a XOR b = c then b XOR c = a and a XOR c = b

[Bob in a quantum-state-of-faith]

Okay, let me cognate on it a bit.

Levers with pins and cams ought to work.

May have to resort to a reset lever, though:  that is, the input pins can be set with an operation, then the XOR outputs on one pin.

Then the output pin is pushed back, to reset everything.... or a side lever could reset.

Or, may we count on continuous force on the input pins?  Then, when released, a weight could push them both to the zero or out position, also withdrawing the output pin.

I'm thinking outloud, here, but two pins on one side, and a third one opposite.

Push in either pin alone, the output pin is out-- so long as force is maintained on either input pin.

But, push in both input pins at the same time, and the output pin stays in.  Must release both pins to reset; releasing either pin alone would do nothing, in fact, it would not release until the other pin is also released.

I think this is possible, with simple levers and weights.  Maybe a cam.  I'll need to locate my sketch pad....

....I'll start with one pin push in, with a lever/weight, and the output is out.  No direct linkage, just one item pushing against the other, a needed requirement, I'm thinking.

Then, add a similar second pin/lever. 

Finally, a blocking or releasing mechanism that inhibits output, if two pins are in at the same time.

--------

Let me know if you want latching, with a fourth reset lever or force-in maintaining style.  It will matter.

[beagle]

I' d use   a^b = (a+b)&!(a&b)

(or use deMorgan to rearrange the last part to (!a)+(!b)

With the ones you've already got you'd need the and gate.
My and gate  would be a box with a vertical slot cut on the front and back sides.
a plate with a dowel across it's middle would be free to move up and down this slot.  The inputs would be rods coming in from the bottom and aligned to touch each end of the plate. The output would be a rod dropping from middle of the top face of the box and connected to the middle of the plate by a hinge (or tightly confined to just sit on it).

When only one rod was pushed up the plate would be at 45 degrees and only just over half the output rod would emerge. With both the plate would be horizontal and more of the output rod would emerge.
With neither input  gravity would make the plate sink to the bottom of its vertical slot.
I would use additional leverage outside the box (a phrase I hopefully will never use again) as necessary with the combination of gates to stop the outputs dropping below the 1 threshold for a series of gates and to avoid your computer needing to be ten storeys tall. I'm sort of assuming that we can have a simple additional gate that avoids creep by resetting anything below the 1 threshold to no or insufficent rod movement to count as a 1 going forwards.

Will think about doing xor simply (but I was rubbish at woodwork/metalwork at school...).

[Swatopluk]

I now have an AND gate (wil put up sketch when finished). In combination with an earlier idea it could work (although I'd prefer a single gate).

[Swatopluk]

AND gate:
Input on the left side
If only one of the input blocks is moved forward, the intermediate output rod moves only half the distance of that. For full forward both blocks have to move.
The proportional lever on the right doubles the movement of the intermediate rod.
The output tunnel is twice as long as that first distance, so the output rod only sticks out of the box at full length (=1),when both input blocks move (=AND).
If only one block moves, it reaches just edge of the box, if none moves it is fully retracted (both cases count as Output= 0).
For reset a weight at the lower end of the proportional lever should suffice (not depicted in the sketch).

Edit: The upper end of the proportional lever is a bit short here. The distance between the joints should of course be equal.

[Swatopluk]

First part of XOR:

If none or both input pins are 'in' (=1) then a pattern

10
01

is on the output side.

if only one input pin is 'in' then the pattern is

11
00

or

00
11

Putting an AND behind both those output pairs (and if necessary an OR* behind those ANDs) gives a net XOR.

The OR (lower sketch) should be self-explaining.

This is all too complicated. There should be an easier way.

[Aggie]

Output pin should exert moderate force at least enough to trigger a stronger force exerter.
It should also work, if the parts are rather crude. If the whole thing would be 2x2 inches then 1 mm tolerance should still be sufficient. Input and output should not be on the same side, otherwise it doesn't matter much. No requirement for it to work in more than one position, i.e. it can use the direction of gravity as a part of its working mechanism (e.g. a weight that keeps Output=0 when the input pins are not engaged).

Very very crude mechanism could be done with a balance and removable 'pins', but would require a vertical orientation (see attachment; sketch is cruder than the ).  I suspect this is much less elegant than you are looking for - will keep pondering. Recommend using heavy elements for weights such as barium, gold, lead or uranium.


Hmm, noticed that this design would require a spring to retract the output pin.  How about alternatives to the coil spring (leaf spring, etc)?  I originally conceived it as being oriented with the output pin up - no spring necessary, but you'd have to hang the weights off hook.

[Swatopluk]

Ouch! As usual I overlook the simple solution for the oversophisticated.
Your design does not require a spring. A weighted lever in the right postion will take care of that (or a weight on a string).
Thanks!

[Aggie]

You're welcome.  

If you wanted weights that didn't need to be removed completely, you could put some pins out either side of each weight which would line up with matching slots.  A 1/4 turn when the weight was mostly retracted would support in the 0 position, and a quick turn would drop it back to 1.

[ivor]

I'm so glad you're not building a b o m b Swato!

You guys are so smart it's scary.