Mechanical circuits: electronics without electricity [video] | Hacker News Reader

Mechanical circuits: electronics without electricity [video]

youtube.com

135 pointszdw33 comments

I wrapped my Spintronics boxes for Christmas and just opened them before new years.

I have lost 4+ hours to playing with Act 1.

Highly recommend purchasing if this even remotely interests you!

There is a free browser based simulator where you can experiment, too: https://upperstory.com/spintronics/simulator/

Maybe we move in this direction in the future. For example, "Rod logic" in The Diamond Age, which is likely inspired by Drexler's nanotech idea that nanoscale mechanical computers may have advantages (energy?) over electric ones.

More in-line with the video: https://en.wikipedia.org/wiki/Mechanical%E2%80%93electrical_...

There's been some work with 3d printing mechanical gates: https://news.ycombinator.com/item?id=19259364

I would really like to get a resin 3d printer to create some dense circuits that would be stupid amounts of extruder printing.

I'm not sure if mechanical copies of electrical circuits will do that though, it will probably be something more babbage engine like.

There are two serious problems with mechanical computing that would need nanoscale solutions: wear and sticking/friction.

In about 2004 my alcoholic mechanical-genius of a roommate looks at me and says: "I think this computer thing is staying; here's a cool fact: a car's differential is an adding machine, but it's analog and not digital." Making an analog device into a digital device requires finding a way to add saturation in finite time.

For differentials, this can be done by letting rotation represent the value of the device.

Here's the plan we sketched out in ~2005... let `D(x,y) = (x+y)/2` represent a differential (we used spur gear differentials because they're more robust & easier to manufacture). Then, if 1=clockwise, and -1=counterclockwise, then we get the following truth-table for `I0 := D(L,R)`:

    I0 := D(L,R) ==
     1  1 =  1
     1 -1 =  0
    -1  1 =  0
    -1 -1 = -1

We then can use another spur-gear differential with a floating input and two palls on the floating (F) side and the output to "clamp" the result (which has to be scaled by 2x): `I1 := D(I0,_)*2`:

    I1 := D(I0,_)*2 ==
     1 F = 1
     0 F = 0
    -1 F = 0

Finally, we scale the output and add -1: `I2 := D(2*I1,-1)*2`, converting back to 1,-1.

This is an "AND" gate. You can build "OR", "XOR", "NOT", etc., fairly easily.

EDIT: A similar device can be made from those Japanese bamboo water gates. If you've got a bucket with a hole in it, then the bucket will drain if the amount of water coming in is less than going out. So, if the input is A, and the output is 3/2A, then if we've got two inputs we've got (A+A) > 3/2A. So, the result is that with 0 or 1 inputs, the bucket drains, with two inputs it overflows.

The bucket is attached to an arm, and it is the far end of the arm that has a channel with water coming in to the channel; notice, that the water in the bucket only controls the flow of the water in the channel; it is not the water in the bucket providing water into the channel.

If the bucket is full, then the channel is up and no water flows; otherwise the bucket allows water through. This forms a NAND gate.

I've never found a way to make this reliable, although I didn't really try.

Funny thing, the maker of Spintronics made Turing Tumble first -- it's a similar game which explores mechanical (marble) logic! https://upperstory.com/turingtumble/

Many systems can be represented as second order ordinary differential equations which results in these beautiful analogies

In this context, Bond graphs were invented to abstract many physical systems and represent them graphically: https://en.wikipedia.org/wiki/Bond_graph

I'm hoping someone with better mechanics knowledge than I can help me understand something that's been bothering me about this: if you have a simple circuit of the "spin battery" driving a "spin resistor", at equilibrium (when the chain is moving with constant speed) isn't the torque applied by the battery necessarily the same as the frictional torque in the resistor? Otherwise you'd still have some kind of acceleration.

For a twice larger "spin resistor", then, the equilibrium torque would be twice as large, and intuitively, the chain speed would be half as fast. So the same amount of work is being done by the "spin battery".

Doesn't this mean, then, that the "spin battery" is constant power, rather than constant voltage? Their FAQ says the unit of "spin voltage" is Newtons, and that the battery is 3 spin volts, but this doesn't agree with my understanding of the mechanics.

I think the chain just wouldn't ever move at constant speed. It would be slowing the whole time as the energy stored in the battery is spent in the resistor.

That's a good point, but I think for an approximation, we can assume its constant. It gets up to speed very fast and then stops equally fast. Presumably this is because of the constant-torque spring they're using? In any case, by observation it does seem that for most of the time, the speed isn't changing.

Yea

I think Spintronics resistors will resist motion with a force proportional to rotational speed, which I think follows from the resistance coming from oil in the bearings. So the torque will be the same for both resistors (since the battery applies constant torque), but the speed at which the resistors apply this torque will be different by a factor of two.

So you're saying the bearing friction is twice as large in the bigger resistor but the speed is half, resulting the same torque? This seems like the simplest explanation, and what my mind went to first, but I couldn't find any resources that describe this behavior mathematically.

I'm also curious what defines the steady state speed.

> For a twice larger "spin resistor", then, the equilibrium torque would be twice as large, and intuitively, the chain speed would be half as fast.

For the twice-larger resistor, the torque is the same, and the speed is half as fast. Can you explain why you think the torque is twice as large?

To move the 2x resistor at the same speed as the 1x resistor, you need twice the torque.

Maybe my misunderstanding is how the speed would be independent of torque.

My thought was that the 2x resistor has 2x the coefficient of kinetic friction, which means 2x force and 2x frictional torque. In order for the system to be at equilibrium, meaning constant speed, the battery torque must match the friction torque.

I got this as a means to held build a better intuition for how electronics works.

It’s one thing to read a theory, or watch an animation. But to be able to play with such a high fidelity analog is by far the best.

It’s like going from print debugging (measuring values in a circuit for example) to a full on debugger where you can stop the world, adjust something and see how _everything_ is affected.

For me the best intuition for electricity is pipes with gas. Weight of gas is charge. Pressure is voltage. Weight of transferred gas per second is current. Pump is power source.

Inductance is momentum

Capacitance is volume

What would a 'gas circuitry' equivalence of an inductor be?

(I'd guess that a capacitor could be made from a rubber sheet blocking the pipe)

A spinning rotor immersed in the flow. Like a water wheel.

I finally could understand how a capacitor can "DC block" anything. I thought that in the moment AC starts to pass and capacitor become conduct, there will be nothing obvious that would block the bias to pass along, since it's in the same "signal".

The answer is clear using this spintronics analogy.

Tension is anything wanting to move. Since the cap charge up with the input tension, eventually the tension become zero (in resistor's point of view), like force vectors pointing to each other. Any change of bias will move the resistor by the difference between input tension and capacitor tension, until equilibrium is reached again. For now on, the resistor will swing back and forth exactly the AC part of the source.

I really wish the Spintronics chain was compatible with Lego Technic chain and gears.

I was one of the backers of this, and have had it for a couple weeks. The product is of extremely high quality and it's really fun. Happy to answer any questions.

Just looking at the first spintronics "circuit" in this video I see a problem. If you take out the "resistor" and just have the "battery" alone, pull the cord and let go, the "battery" will still spin and lose all it's "charge." In actual electric circuits that won't happen, the battery has to be connected to something to discharge. That breaks the analogy right off the bat.

Back in college I was a teacher's assistant for the one electrical engineering class that mechanical engineering students were required to take and I learned just how unhelpful, even detrimental, analogies for electricity are. The professor didn't use any, but the students had heard some water-through-pipe analogies and had questions for me. I quickly figured out that debugging their analogies was not helping them solve the actual circuit problems they needed to solve for the class.

It really bugs engineers to not know how something works, but the thing is, nobody really has all the answers to how or why electricity works, and yet we still do amazing things with it. You don't need to know.

For example, why is it that moving electrons back and forth in one wire causes electrons to move back and forth in other wires that are miles and miles (light-years even) away (talking about antennas and RF here)? Nobody knows! But we know it happens, we have characterized the behavior quite well, and we have mathematical equations that describe it, and we can do amazing things with the knowledge we do have.

Here's what you need to know for basic circuits:

V = IR

P = IV

Power supplies (batteries, wall outlets, etc ) supply power, which you know from the second equation above is voltage and current.

Capacitors resist change in voltage

Inductors resist change in current

Diodes only let current flow in one direction, after a threshold voltage has been exceeded.

Amplifiers take in two input powers and add some of one to the other.

Transistors (if we are sticking to the digital realm) are voltage controlled switches.

In the analog realm transistors are more complicated (but often they can be thought of as amplifiers), but you also don't encounter them in basic circuits unless you start taking apart power supplies and amplifiers.

EDIT: almost forgot my favorite thing (somewhat tongue in cheek): all electrical problems are connectivity problems. You either have too much connectivity (extreme being a short circuit), or not enough (extreme being a broken circuit).

The battery won't spin and lose all charge, it has a breaker mechanism that will stop it similar to how seatbelts work, it gets stuck if it pulls too hard. This is to prevent damaging a circuit if you forget resistors.

Of course this is a model of how electricity should work, but it is still useful and they document the differences from "real" electricity decently in the books.

In any case its a super fun toy and I highly recommend getting it!

i love the fact that you can have two perspectives physical abstractions :)

Olá

Life is like a box of terrible analogies. - Oscar Wilde

This analogy is quite interesting.

Each element and connector are actually equivalents of circuit loops that are spliced together when you are connecting them.

So a spinotronics diode works more like a loop of "diode wire" out of which you can make multiple diodes by splicing other connectors and elements into it.

That's why you can make FULL BRIDGE RECTIFIER with just two spinotronic "diodes".

Electrons are described as fluids when there is superconductivity.

From https://en.wikipedia.org/wiki/Superconductivity :

> Unlike an ordinary metallic conductor, whose resistance decreases gradually as its temperature is lowered even down to near absolute zero, a superconductor has a characteristic critical temperature below which the resistance drops abruptly to zero. [1][2] An electric current through a loop of superconducting wire can persist indefinitely with no power source.

FWIU, an electric current pattern described as EM hertz waves (e.g. as sinusoids) is practically persisted at Lagrangian points and in nonterminal, non-intersecting Lorentz curve paths at least?

IRL electronic components waste energy as heat like steaming, over-pressurized water towers. And erasing bits releases heat instead of dropping the 1 onto the negative or ground "return path"

I agree that Spintronics is a great game for mechanical circuits, which are in certain sufficient ways like electronic circuits, which can't persist qubits for any reasonable unit of time.

Terrible analogies are superior to great ones.

I can shout great analogies at my team all day long and they will just roll their eyes

But the moment I toss one out there that's terrible, they give me a dozen better ones back, and I win because I got them to swallow thing I was sugarcoating with analogy

So you are saying it is as if cat herding potentially exists to create a binary appreciation fallacy for set-theoretical classification of analogies? Mind blown.

Reply on news.ycombinator.com