/diy/ - Do It Yourself

>>2792916
>I want to build my own DIY computer from scratch, but I don't really know where to start.
somebody post the bingo card

https://eater.net/6502

>>2792916
> from scratch
> pre-made CPU with 10s of thousands of transistors
Ngmi

>>2792941
and whatever OP does, he needs to absolutely plug his thousands of transistors into solderless breadboards like his pic.

>>2792916
Just how far down are you going. Every gate is gonna be 3 or 4 transistors, plus associated parts:)

Make an analogue computer instead, it'll be more fun.

>>>/g/

>>2792941
>>2792959
>>2792963
Not really like that when I mean "from scratch", I don't want to use discreet transistors, or I'll never do it. I want to use ICs for this.

>>2792967
I never thought of this, I'm mostly familliar with digital ones so that's why I want to build one like this. Any suggestions about analog one?

>>2792916
>The main purpose of it is to learn more about computers and how it all works down there.
>Is it better to pick up something like 65c02 and build around it or made everything by myself?
If you want something you can call your own, develop your own instruction set instead of trying to emulate an existing processor.
>Not really like that when I mean "from scratch", I don't want to use discreet transistors, or I'll never do it. I want to use ICs for this.
Just so you know what you're in for, It can take several hundred discrete logic ICs to make a primitive processor capable of running a calculator. This is how computers were built in the 1960s.
One of the first, if not the first, IC processors was the 4004 and it had a few thousand transistors.

>>2793110
> use ICs
The maximum “pre-made” complexity that I’d accept as “from scratch” is using something like a 74181.

> 4004
The 4004 is a pre-made CPU. Might as well just buy an old C64 and play with that.

“From scratch” means that it doesn’t have a pre-existing instruction set designed by someone else.

— t. /mcg/

>>2792916
>want to build an MSX2 clone
>find all the needed parts beside the VDP for around $50
great !
>find the VDP chip alone cost more than $50

>no garentee any of the douzens of chips work
>if a single ship is crap the whole system won't work and i'd be wasting over 100 bux on nothing

i really want to do it, but im too short on cash to gamble on ebay selles

>>2792916
Have you ever played this?
It made me fall in love with computing architecture, even when sometimes it is ambiguous in explaining stuff.

Simulators are probably the most cost effecting thing you can do.

>>2792916
i made ben's 6502 project a couple years ago, and i also made a basic microcontroller with a 32-bit ST chip. the 6502 was fun and easy, especially with the ben video, but the ST project is definitely not for the faint of heart. no single-step debugging, theres like 14 steps you have to do before you get even one hint of feedback whether or not youre on the right track, and the manual is like 1600 pages long. very proud of myself for getting it working tho.
the 6502 manual is like, what, 50 pages? super easy.

>>2793370
heres a pic of the ST. i used a header only printf implementation and wrote the put() function to output on the chip's UART. the pi then picked it up and printed it on my desktop. quite proud of that one.

>>2793089
Well the thing is they're in a way simpler but also more complex depending what you want.
https://www.youtube.com/watch?v=gSX-fV7yLTI

>>2793389
The simplest analog computer example I can think of would be the slide rule.
The point is that an analog computer is just a device that varies continuously unlike digital logic.

>>2793391
>The simplest analog computer example I can think of would be the slide rule.
ah yes, the literal boomer being literally stupid. I bet you got used to eyes glazing over when you pontificate about 50 years ago.

>>2792916

>>2792916
Check out gigatron.io which is also a kit.

>>2793396
Refute me faggot, don't just call people names.
A slide rule is not a simple device but it is a simple example of one.

>>2793006
/g/ is way too stupid for this stuff. I used to browse there a lot but it's all threads about consumer electronics and beginner programming these days, and troll/jokes threads of course. whenever a thread about actually making something with electronics pops up they point the OP to /diy/

>>2793089
I've built analog breadboard computers. Here's one generating a circle using the pythagorean theorem.

An analog "computer" isn't a computer in the modern sense. It's a circuit that solves a single equation using a sequence of things like voltage adders, voltage dividers, logarithmic amplifiers, etc. Once you understand how to do simple things like multiplying and subtracting voltages with op-amps you can build a circuit that solves most equations.

If you start getting into it you might stumble on analog multiplier chips, then realize they cost hundreds of dollars. You don't need these. You can do multiplaction with log and antolog opamp circuits.

An analog computer is really not a beginner project. Start with something much simpler like a headphone amplifier.

>>2792916
https://hothardware.com/news/uk-man-builds-massive-computer-out-of-discrete-transistors-to-play-tetris-and-other-classic-games

>>2794068
That's insane. Reminds me of Scanimate. I've pondered building something like that with Eurorack VCO modules which have a hard sync input, so you can generate ramps. I hate breadboards.

>>2792916
I recommend the Rosco M68K kit. 68K is the absolute best architecture if by chance you want to do assembler. Also, when you get bored with that, you can run EmuTOS on it and play with Atari ST stuff.
Otherwise maybe get an MSX2 kit? Again, when you get bored with it, it'll at least play games.
There's CP/M Z80 stuff. I dislike it. No one has a good 80-column solution for it, and the crux of CP/M was 80-column software.

>>2793366
I'll give it a try, seems very interesting


>>2794068
What amount of effort and time you put into this? Looks scary and interesting af

>>2794207
I'll think about both options, thanks for information, but what is a good MSX kit out there? Quick search results in GR8Bit and Omega MSX, but maybe it's better to save money for Rosco. As for assembly I'll be learning it at uni, motorolla one, so it'll go perfectly if I get my hands on it.

>>2793121
Yeah, logic ICs look like my way to go, they're fun and understandable, almost like lego but with electricity.

>>2794207
>Rosco M68K
>Regular price: £149.99
>MC68010 16/32-bit CPU
>1MB RAM
>512KB Flash ROM (Up to 1MB supported)
>On-board UART (115200 BPS as standard)
>SD Card support (requires 5V Arduino-compatible SD adapter, not supplied)
Not OP, but even though the idea of assembling a computer from scratch sounds interesting, I don't think I'd be willing to drop 150 pounds for something that's got inferior specs to a Raspberry Pi Pico in all but storage capacity.

>>2794590
At least you’ll know how it works and that the chinese aren’t listing to your conversations.
You can’t know that they aren’t on a raspberry pi.

>>2794892
>You can’t know that they aren’t on a raspberry pi.

If you don't know how to monitor a network why do you have an opinion?
If security really matters, don't use a computer to communicate since the five eyes pwn the backbones.

Why would anyone use the internet for anything compromising in the first place? The best security is to resist temptation and not follow the herd.

https://nandgame.com/ is a pretty good resource to learn from, It guides you through all the steps of building a computer starting from relays

Personally, I've been working on building one in minecraft. All the ALU is missing is a circuits for bitshifting and jumps and it should be ready for programming

>>2792916
i'd love to build an MSX replica but there are too many fake chips out there
an MSX requires many chips to work if any of em turn out to be fake/defective you can kiss your project goodbye
allot of time and money wasted

Hi OP. I was once played with redstone, I watched all of ben eaters videos, tried ordering a bunch of discreet parts to plug into a breadboard, etc. I ended up going to school for this

I would not try to learn going the ben eater route with discreet parts or ICs. You will swamp yourself with a couple hundreds of dollars in parts, reading each datasheet, and it wont turn on when you run into parasitic capacitance on the breadboard or some shit (idk I'm not a EE).

What you need to do is get an FPGA and toolchain and learn how how to implement things in verilog. If you're not familiar verilog basically lets you define registers and wires on an FPGA with a C-like syntax. You can first simulate(!) verilog code on your computer and then synthesize(!) it onto the FPGA to get any hardware you want in real life. Hopefully you can see how buying 1 $50 fpga is better than drowning in 100s of different parts from weird vendors for $$$. Learn how verilog works with a

For the actual processor design a starting point I recommend is with choosing an ISA. this is like x86 or ARM, I recommend riscV or MIPs (what we used in school) because they are much simpler. What you can look up for inspo is you want to design a Single Cycle MIPs processor. Then you can work your way into pipelining, caching, multicore, etc. MANY college courses follow this route and you can maybe follow one if it's open source

Here are some resources you can look up:
These lectures look great: https://courses.cs.washington.edu/courses/cse378/10sp/lectures/
Here's a book I used in college, you can get it on libgen: Computer Organization and Design MIPS Edition: The Hardware/Software Interface

I haven't explored the game turing complete. nand game is a nice game, but not something to build a whole project on. Minecraft works and is fun but you're limited by your redstone and creative building tools knowledge as well as compute speed.

>>2797795
Yeah I'm a final year Computer Engineering student and FPGAs are definitely the best way to go if what you want is to make your own processor, they aren't the best if what you want is something that you can show off and looks cool but it's what is used in the real world when you need to prototype some chip, if you have the money the basys 3 development board is pretty comfy

>>2797795
I completed turing complete up until assembly challenges, which is optional.
The game leaves you reinventing everything, it's challenging and sometimes ambiguous when explaining stuff (at least for someone like me who never touched logic circuitry before) but still satisfying, you build an 8-bit computer starting from scratch, you also build another one that is more sophisticated than the previous.

The problem is it won't give you a good overview on how irl components work, so you end up building memories out of a "delay" component, and counters out of adders and a registers. Some components like the "program" one is given without understanding how it works, and irl problems like how to deal with flipflops saving stuff are not present at all.

I say OP is far off better getting a simulator like logisim and a good information source, then moving to an fpga and languages like you said.

Still, turing complete is a fun puzzling toy.

>>2792916
>I want to build my own DIY computer from scratch, but I don't really know where to start.
We have a FAQ:
https://wiki.installgentoo.com/wiki//aig/_Alternative_ISA_General

>>2793391
No thats no their definition.

An analogue computer is a computer that builds a physical analogue to whatever discrete problem you want to compute. Be it pulleys, complex water systems or analog electrical circuits. The theoretiocally infinite variability of working with non discrete states as such is just a side effect.

In the slide rule example the graticules have the operations encoded in their varying distances. It would still be an analogue computer even if there were a finite number of ratcheting graticules on it.

tl,dr: An analog computer translates an abstract problem into a physical one. piggybacking of laws and relationships to perform computation.

>>2798888

neat, also checked

>>2792916
This is a nice book about this topic. It has a chapter on building a digital computer with a schematic (using 7400 series logic chips) and an in-depth explanation of how it works.

>>2792931

>>2792933
Was about to suggest that, that dude is doing god's job.