/diy/ - Do It Yourself

you need gates in a real computer because a program consists of a series of steps and decisions, and gates can execute those. if you wanna use a conceptual paper computer, you dont need any gates, you need a finite state machine. FSMs are taught in early computer science classes, so you should find lots of documentation.

>>1115026
>a mechanical computer that can be able to run a word pad

You mean one of these?

>>1115052
A finite state machine would be built using gates among other things. FSM isn't some sort of piece of hardware; it's a logic concept you draw on paper then design hardware or software to implement it.

>>1115056
No. It has to have no paper.

>>1115052
I never said I didn't want gates. I just want logic gates that aren't conventionally used. Ones that are as simple as xor, not, and, or, ect. and not ones that are actually just substitutes for clusters of the basic gates.

>>1115076
Not possible.

>>1115079
Yes it is. In fact I already know my method I'll be using. Please troll somewhere else.

>>1115079
>>1115084
Let me rephrase this. I only gate about the logic gates and reasons behind them in other features of the computer. I already know about ALU and Memory. I need to know about clocks, program memory, decoders, ect. I know how to even convert data into my own custom computations for display purposes or other purposes, just not those things.

>>1115086
Different anon, I have no clue what you're trying to do. What exactly are your goals and what do you have done already?

>>1115078
>>1115026
When you say "a mechanical computer" do you actually mean a thing that uses gears and levers like Babbage's difference engine? That thing was a simple scientific calculator and took a decade to build. I think you might not be comprehending just how many gates you'd need to build a word processor.

>>1115092
This project is a very "cross project" kind of thing, but, essentially, I want to create a computer that is as simple to use as say, one of those old school computers that just had wordpad and a few other things on it, but, I also want it to have no need for electromagnetism. Instead, I want it to be powered by something more kinetic and microscopic. I don't care about learning the scaling as much as I just care about the actual program memory, clock, decoders, and other logic gate maps. That's all I need is the logic gate maps for specific task. The concept.

>>1115094
close, but, no. I already know how that works. It's not about how many logic gates you need because the pattern is just repeated. I think you don't know how specific components actually work of a computer and are just stating things without knowing anything.

>>1115086
>>1115097
You are not making an iota of sense.

>>1115097
>I think you don't know how specific components actually work of a computer and are just stating things without knowing anything.

Not that guy, but, even if you're right and he doesn't "know how specific components actually work of a computer", YOU don't know a damn thing about what you're trying to do. You freely admit that you don't know a damn thing about how a processor works, and, if I had to guess, are still high on the novelty of learning the absolute basics of digital logic and the fact that it isn't just applicable to silicon.

You are in no position to judge.

I'll be honest: I have no idea of what kind of total parts count you're looking at here. I DO know that it's going to be astronomical, especially if you try to adhere to any real-life standards, such that, theoretically, this thing could somehow be interfaced with a normal PC.

The most basic bitch, early CPUs still had thousands of individual components on their dies. Let's just pull the Z80 as an example. 8500 transistors. Even if you could somehow use half the number of mechanical components to create an equivalent logic gate, you're still looking at over 4000 components just in the CPU logic. You will then need to connect those logic gates to each other meaningfully, something that's taken for granted in a CPU and not included in any component count. Then the parts to fasten them together. Oh and the parts that fasten the gates to some kind of mechanical support. Then extra parts for odd linkages here and there. Then a frame to put this all in.

And then you have to do this ALL OVER AGAIN for the memory architecture, address bus, human interface, and whatever form of output or display it has. Then I hope you planned ahead and provided all the necessary bits and bobs to make all this shit move.

Unless you radically simplify what you define as "a word pad of sort", I will be astounded if you can get this thing working in less than 50,000 parts.

>>1115126

Just for reference, by a huge margin, the single most complex thing a consumer will ever own is a car. Some cursory Googling will turn up the fact that your average car has between 10,000 and 30,000 individual components.

And, by the way, modern computers don't run on "electromagnetic" effects. All of the business bits are reliant on electrical ones, with electromagnetism being reserved only for power-handling components.

>pls tell me how to be smart and design computer from nothing

This is literally what you're asking, in a nutshell. We cannot teach you this. No single person really could. And only with the most egregious naivety could one think otherwise.

>>1115126
like I said, you are just trolling. Also I'm just going to watch some lectures so I don't have to deal with autism.

>>1115129
Dude unless you want to spend thousands of hours assembling logic gates, it's impossible

>>1115026
>run a word pad of sort without being electromagnetic

>>1115076
Mechanical information boards use a set of painted cards and rolls them to the letter/character it needs. It uses electric motors but you can just replace it with a regular mechanism similar to a gear shifter system.

>>1115026
A processor is a very complex design, requiring many abstraction layers to fully grasp its inner workings. The reason that components in a layer are specified using subcomponents of the previous layer, and not using transistors or gates, is because the specific implementations of those devices is dependent on the actual hardware. For example, in CMOS logic, NAND and NOR gates are the most basic type of gate, and ALUs are built with this in mind. If, in your specific system, OR and NOT gates are the easiest to make, you'll probably try to build things around that.
With that said, I still don't know what are you trying to build. A binary mechanical processor, like a Z1? Or an analog computer? Maybe a denary computer, like the analytical engine? Are you designing the CPU from scratch, or using a known design, like MIPS? What type of logic are you using. latch-lever logic, rod logic, other I'm not aware of?

>>1115095
yeah, ok, good luck, not gonna happen.
Remember those old computers, that ran off paper cards, and used tubes? A whole building full of that shit was basically the power of modern scientific calculator.
If you want something, so complex as a notepad program you just gonna fail.
Maybe you just didn't vocalise what you really have in mind.
The most realistic thing you can do is various mechanical calculators, integrators and shit. But not a goddamn computer.

>>1115129
>I'm just going to watch some lectures so I don't have to deal with autism

Unfortunately, there's no known cure, so you're going to be stuck with it for the rest of your life.

And I take my original view back: You're not just ignorant, you're stupid. Worse, willfully so.

>>1115097
Are you the same guy who was trying to build Batman arm braces last week?

>>1115129
>like I said, you are just trolling. Also I'm just going to watch some lectures so I don't have to deal with autism.

sounds like you didn't understand a word he said

you're not going to do anything

>>1115188
I'm trying to copy a computer that already exist, but, without using electromagnetism if you will. This is much easier than actually designing your own parts. Don't get me wrong. I know about the ALU and how to build one manually and have already done so in a program called CEDAR. I want to make the same processes, but, in a make shift kind of way. Like water running down a mountain and gates which coordinate the stream, but, not literally. This is merely an example. It also much be automated with the ability for someone to put in user inputs.

>>1115744
>I'm trying to copy a computer that already exist
go for a Pentium-III

>>1115095
Look up the University of Illinois LC-3 (Little Computer 3). The hardware schematics are laid out plainly and the assembly language for it is intuitive. Building it mechanically would take years, but I think this is what you want.

>>1115866
Here's the hardware

You also mentioned that you wondered how the clock works. It deals with the expansion of a crystal due to temperature change when running electrical current across it. There are a few more ways to do this, but this is very compact, and has lead us to the computing capabilities we have today.

>>1115026
You're not asking the impossible, but if you want even a four function calculator, incredibly infeasible, and if you want a word processor, infeasible even for someone with Bill Gates's money and ten human lifetimes.

This is what you can hope to accomplish.

https://en.m.wikipedia.org/wiki/Digi-Comp_I

If you're clever, maybe even up to the level of some kennex computers you see on YouTube. Any time you bother to search.

>>1115129
and this is when i realized it was all bait.

>>1115127
There's really nothing else needed in this thread. Well put.

>>1115866
thank you. This is the kind of stuff I'm looking for.

>>1116284
Modular engineering.

>>1116284
>Minecraft Computers
Please... Clearly you are working in the caveman ages.

I knew a guy in HVAC trade school that built an air powered calculator using only air gates. He won the bet with the instructor but it would only do 2 digits and it took up an entire sheet of 4 x 8 plywood. Those parts were packed TIGHT too. You better have a LOT of room.

>>1115026
it will be huge and if you're just making a word processor you don't need to actually make a computer.

The first word processors weren't computers
https://en.wikipedia.org/wiki/IBM_MT/ST

The problem you are going to run into is the need for rewrittable memory, you need some way to store a lot of characters. At minimum I think you need to be able to store 5000 characters, at 8 bits to per character this corresponds to 40,000 bits. If you do this mechanically things tend to be a bit big.

Though supposedly, a swiss company did make a word processor that ran off of fluidic logic gates. It was huge, the size of a desk.

>>1117301
That should be no problem for OP since he's full of hot air.

>>1115026
OP, how are you dealing with the mechanical friction and how would it be powered?

For your initial question: Logic tables. You start of with a list of instructions you want to implement and write down, what should happen internally. After a while you end up with 'input-output' tables for each sub-component. These can be transformed to logical gates and if you want just down to NOR gates only.

>>1117507
kek

>>1115744

You haven't said anything meaningful. What EXACTLY do you mean by

>without using electromagnetism if you will

Here's your problem OP- you want to build a digital computer using non-electronic components. A digital computer is designed from the ground up around using transistors. If you use a digital computer design without using digital components then you're going through a hell of a lot of work for zero benefit.

Read up on analog computers or mechanical computers.

>>1115076
The most mechanical a computer has ever been was through the use of punch card storage. The first computers used tubes and were essentially a sharp calculator

>>1119374
The Z1 was a rod-logic (iirc) mechanical computer. It used perforated film as instruction tape, but it had more than that as mechanics. Its possible, just probably not really worth it unless OP sticks his head out of his ass and explains what does he want

>>1119496
The Z1 used a unique form of mechanical logic gates not rod logic. Rod logic has the distinction of allowing for reversible computation, the Z1 could not do this.

OP just wants to make a text editor though. Two hardest thing about that are the rewrittable memory and display. Doing the character decoding for the display compactly will be interesting

Here is a patent for what essentially amounts an air logic controlled text editor system that attaches to a type writer:
https://www.google.com/patents/US3466610

The rewrittable data storage here was a paper tape that could have indentations popping up or down

Bump

I know this is a troll thread but it's still kinda interesting.

http://www.halfbakedmaker.org/blog/58
Mechanical identity, NOT, AND, OR, NAND, NOR gates

http://www.halfbakedmaker.org/blog/76
Memory, registers, addressable memory

http://www.halfbakedmaker.org/blog/116
Full adder

http://www.zyvex.com/nanotech/mechano.html
More technical explanation, plus an idea of using "buckled plates" to make logic gates.

i l lain-chan

Has anyone here heard of micro-pneumatic computers?
http://www.instructables.com/id/Air-Powered-Fluidic-Ink-LEDs-and-Circuits/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2917228/#R15

>>1123266
yes, but I really don't like how you have to have vacuum and pressure lines. I also don't like the fact you can't have pressure activated normally closed valves.

>>1115026
>I want to become Babbage but I'm living in the wrong century

>>1115076
ghetto rig a typewriter to impression onto a Roman style wax tablet instead of inking paper.

>>1124775
Kek that would work

Do it OP

Bump

>>1115026
I'm not gonna help you because your pic implies you're going to use it to imprison Lain