Build an 8-bit computer from scratch (2016) | Hacker News Reader

Build an 8-bit computer from scratch (2016)

eater.net

398 pointsjcartw62 comments

I always upvote Ben Eater's project(s).

I got his 6502 kit and worked through his YouTube videos to get it up and running with LCD display and all. I want to actually frame it and get it running, hanging on a wall.

If anyone is new to Ben, find a comfortable chair or couch, make some tea, take a breath, and begin your descent into his long-form breadboard series of nerd-splaining how a CPU works, beginning with basic logic gates.

His follow-on series is a little higher level. He's still at it with the breadboards but he allows that this time the CPU (a 6502) is already built for us. He then walks through how the address bus works, how to attach RAM, etc.

It makes me wish I could be transported back to the 70's when things could still be understood.

His series on building a video adaptor is a particular favourite of mine.

+1 for the 6502 kit. It’s the perfect project. I think the full 8-bit computer is kind of daunting. The 6502 is just big enough to be a fun medium size project. Plus there is a perfect extension project with the video card if you want to add that on.

I also did the 6502 and added a Pi Pico for VGA output and USB input.

I'm still working on ways to add more to the design -- it's been a fun hobby. I actually enjoy cutting the wires and physically putting it together.

If anyone wants to avoid adding a full scale computer just to get TV out I recommend the late Don Lancaster's The Cheap Video Cookbook and Son of Cheap Video.

See https://archive.org/details/donlancaster_socv1/page/n3/mode/..., https://annas-archive.org/isbndb/9780672217234

I did the whole thing, was able to get the Fibonacci numbers appear and learned a lot during the process. However, I would not recommend other people to try this today, because --

* It is not necessarily the best way to spend your time and money. You'll be looking at tens of hours on building it plus over $300 in parts, for a very slow and basic computer that you probably will forget and throw away at some point. Cutting and laying out wires are some of the most tedious and frustrating process. There are other cheaper, more efficient ways to learn how a computer works. It may or may not be worth the effort to you.

* The tutorial is really old, with some very questionable design choices and no updates from the author. Some of the designs are just plain wrong -- e.g. floating inputs or missing resistors. It is very unlikely you can reproduce it by strictly following the tutorial. You'll need to spend time debugging those issues and go to reddit to see other people's experience with this, potentially seeking help as well. No doubt debugging is an important part of designing and understanding the circuit. But only if you have the foundational knowledge and patience, of which I happened to have, but I can imagine that someone who does not understand digital electronics well enough can easily get lost and feel defeated.

* To make the previous point worse, some of the parts are hard to source, depending on where you are. (You'll likely fry or physically break a few components when building it.) Many of the parts are not very commonly used these days, and not exactly easy to acquire, if you only need 1 or 2 of them. I had to go to eBay to get some of them, which took about a week, during which I had to stop and wait for parts to arrive.

With all that said, this may still be the best resource out there that lets you build all these on a breadboard, as I am not aware of an alternative that addresses all the issues above.

I’ll play devil’s advocate.

Following Ben’s videos and building the 8-bit CPU is the first thing that ever helped me “get” electronics and circuitry.

Honestly, way more than half of the fun has been in finding the design flaws and designing fixes for them as well as improving and expanding the design itself.

I have learned so much by going down rabbit holes. The original clock design assumes a make-before-break switch but the one shipped with the kit is the opposite, and this can put one of the flip flops into an invalid state. I switched to CMOS chips and learned why you shouldn’t drive loads like LEDs directly from digital outputs, and dove into the easiest way to power 8 LEDs off 8 IC outs without a mess of transistors. I followed another person’s expansion to a full 8-bit instruction register and a more featureful ALU (with my own changes of course). I worked to minimize power consumption. I’ve learned how to use an oscilloscope. I’ve improved upon nearly every module, and built and rebuilt them all a handful of times.

None of this has any real benefit for the CPU. It will never run anything more than toy programs. But that wasn’t ever the point.

All the flaws in the original design have been perfect as launching points for learning more and more.

As a side note, reading "But How Do It Know" for a total beginner like me can help to scratch the itch of understanding the basics. I really loved the book and will buy my 3rd copy (lost one and gave away another).

Another popular book "An introduction to microcomputers" by Osborne, Adam, 1980 also has great detail on this topic..... Its on the internet archive https://archive.org/details/introductiontomi00adam/mode/2up

A game such as Turing Complete will give the same sort of understanding of the fundamentals of computing, although not a physical learning experience.

Although I haven't followed his project at all, one fully free alternative which gives similar insight (how logic gates build up to a full computer) is the nand2tetris[1] course. It starts with just a nand gate and then creates all necessary components to make a simple programmable computer, all simulated virtually so you don't need any physical components.

1. https://www.nand2tetris.org/

An alternative might be to write or find a gate level simulator, then write a library for each IC like hex inverter, quad nand, etc. Then you could pick up where Ben starts at the pin level, without messing with wires and breadboards, and get much of the same understanding.

Yeah, I started on it and felt the same so I built

https://git.sr.ht/~vhodges/cputhing

There's

  * https://nandgame.com/
  * https://github.com/logisim-evolution/logisim-evolution

too

I was going to ask, is there anything like it where you can build a physical computer out of discrete parts? but you answered that in the last remark.

I would love a physical / homebuilt computer of sorts though. Ideally with clicking switches or lights and the like.

https://m.youtube.com/channel/UCeRXQ_B5WZD3yjPly45myvg

This might be close, but I don't believe he offers a kit.

I would be shocked if anyone is really looking for anything practical out of this, its more of a lab exercise that lets you learn by actually doing things. I learn best this way. It's one thing to draw a circuit, or even put it in a simulator, but the act of hooking up the chips, and watching the blinking lights really sticks with me, and far more deeply ingrains the concepts in my head. And while the 8-bit kit is expensive, and time consuming, I found it worthwhile. There is probably a lot more value to be had in the 6502 series though, if for no other reason that the 6502 is iconic, and it was something of a thrill to me to really understand at a basic level the chip that powered the NES I spent so many hours with.

and you might want more than 16 instructions in your programs!

Speaking as a down-in-the-dirt circuit-slinger who works alongside a lot of ivory-tower engineers who couldn't diagnose a loose wire if their life depended on it, I think that's the whole point.

* Tens of hours and under $1000 is an extremely cheap route to this level of understanding if compared against university courses. I'd give my left nut to have my coworkers go through this process.

* Cutting and laying out wires is time-consuming and error-prone. Yup, hardware is like that. Sweeping out the dojo is menial but important. Developing an intuition around when a weird behavior might actually be a hardware problem, is priceless and absolutely essential for any embedded engineer. And you will never, ever get that from a simulator. (As much as I love the ideas behind nand2tetris, it's entirely done in simulation and that misses half the point, IMHO.)

* Sad to say, the outdated components thing is only getting worse as more basic stuff goes out of production in favor of more highly integrated components. (Even the epic VULCAN-74 caved and used modern RAM.) It might be possible to rework this project to use newer parts but keep the educational value. I suspect there's enough people who've pursued this project and done bits and pieces of that, someone just needs to unify it.

> just plain wrong -- e.g. floating inputs or missing resistors

TTL logic gates have no problem with floating inputs. Those are interpreted as logic high. [1]

This is fairly self-evident from looking at a typical TTL gate schematics. [2]

Some "TTL-compatible" CMOS parts (including 65C02) have internal pull-ups on inputs, to match this quirk. [3]

Granted, it is a bad habit to develop, given that nearly all "modern" logic (i.e. of past 30+ years) is CMOS with no "TTL compatibility", and floating inputs are a big problem there.

[1] https://electronics.stackexchange.com/questions/683051/outpu...

[2] https://upload.wikimedia.org/wikipedia/commons/thumb/d/d4/TT...

[3] https://wilsonminesco.com/NMOS-CMOSdif/

> tens of hours on building it plus over $300 in parts

In other words, quite inexpensive.

To you, sure. To others, maybe not. Whether something is expensive (in money or time) is relative.

Relative to other hands on, high tech educational experiences would be a smart way to think about it, if you insist on adopting the sucks in a long hit of marijuana smoke “everything is just relative, mannnnn” attitude.

Pretty strong disagree with your take (though seems like you end with "this is a bad path with no viable alternatives"?). Laying out the breadboards and wiring this has frustrations but that's part of the experience. Same with the design choices; one of the most valuable things my most-digital of minds learned was that you can never escape the analog world completely. I don't think you should be looking for cheap and efficient in this realm, but what I did to source everything was make a big ali express order about 3 months before I planned to tackle this (over a Christmas/winter break), sometimes ordering the same part from multiple suppliers, and then forgot about it, until the packages started to trickle in. You probably already have a lot of the basics, so you can do this for < $200 pretty easily. Meanwhile I completed the NAND to Tetris project, which was very complimentary. I didn't find any parts particularly hard to source and have more basic gates than all ever use.

You are free to disagree. The project took me way longer than I expected, and while it (building the whole thing) was fun, the experience barely helped deepen my knowledge of electronics or computers -- the videos themselves were already very educational. I wouldn't have done it if I had a better idea of what I was getting into, but anyone can make a different choice if they value their time and money differently.

$300 is somewhat inflated estimate - for example, the breadboard's price is estimated as $4-$9 but I usually buy breadboards at $1.5 each (well, there is a warning that they might be worse quality but I am not used to expensive breadboards so don't notice any issues and I don't care if they are slightly different shade of white).

The Ben Eater computer is an interesting starting point. With a few modifications, such as an extended program counter / address bus, a combinatorial ALU programmed in flash, and a few more registers, it can be upgraded to a more powerful, if slow, computer.

Further, these circuits can be directly simulated in an FPGA if you want to teach yourself an HDL like VHDL or Verilog. You really want to learn better and faster ways to organize a CPU, but it's a fun way for a beginner to bridge into this kind of work.

At the local hackerspace, we actually taught a class that assembled a hybrid 8-bit / 16-bit variation of this computer with 1MB of RAM, an 8-bit ALU, a 16-bit data bus / registers, and a 20-bit address bus. With this additional capability, we could program it over serial with a very basic bootloader, and it was powerful enough to compute pi to 1000 digits. Granted, this took about 18 hours, but it could have run faster with a printed circuit board and a crystal oscillator. We just never really got to that point.

As for how powerful such a computer with discrete logic can be, consider that the KB-11A CPU for the PDP-11/45 was built using 74-series logic ICs. It wasn't until after this system that DEC moved on to VSLI. So, this is a way to play with technology found in computers in 1972 / 1973.

As a stepping stone to digital logic and FPGA work, Ben Eater's tutorial isn't bad. Just get yourself a copy of Don Lancaster's TTL Cookbook or CMOS Cookbook to bring his tutorial up to best practices.

Note that even the "Update" video is nine years old.

Also note that yes it's a computer, but it has 16 bytes of memory to hold both the program and the program's variables.

BYTES. Not KB, not MB, not GB. Bytes.

The demo Fibonacci program is pretty much the most complex thing you can do with it.

But it is in fact a genuine von Neumann computer. And understandable.

It feels almost more useful than BASIC on the Atari 2600, at least.

The MOS 6532 RAM-I/O-Timer chip used by the 2600 (and KIM-1 and others) has 128 bytes of RAM. Game cartridges could add 256 bytes more, along with the program ROM.

For an experience today closer to these machines, you can get a $0.10 CH32V003 microcontroller with 2k RAM and 16k flash.

Here's a 1 Euro retro-computer kit using one that has included in that price support for PS/2 keyboard and VGA video output, all included in the 1 Euro -- even the connectors! All from one 8 pin microcontroller chip:

https://www.olimex.com/Products/Retro-Computers/RVPC/open-so...

https://www.youtube.com/watch?v=dfXWs4CJuY0

That's astonishing. They must make a loss on every one they sell. I bet they don't sell many though, they want 21 EUR to send it to Norway!

Oh! I expected it to be cheaper shipping within Europe. I live on completely the other side of the world (antipodes of Spain or Casablanca to be precise)and it was a similar price. I took 18, which doubled the overall cost compared to getting just 1 (shipping was 30.92 EUR).

Ben Eater's breadboard CPU is a good way to learn the basics, but since the focus is pedagogical, it's very underpowered for the quantity of chips it uses (and you better enjoy forming wires). For a much more powerful computer built around the same number of TTL chips, see the Gigatron.

And no microprocessor!

But as I understand it the kits are no longer available :-(

Sadly, Marcel van Kervinck (one of the two creators of the Gigatron) passed away in 2020. You can still buy PCBs on Tindie, and the parts are easy to procure, it's all standard 74HCT chips. You can find a detailed bill of materials on gigatron.io, and the PCB files are there as well.

The one unavailable part is the fancy wooden box.

RE .... See the Gigatron. Google link https://www.google.com/search?q=Gigatron+CPU&oq=Gigatron+CPU gives lots of good explanatory links . (I had never heard of it , it looks quite interesting .... )

Building up a full CPU from logic was one of my favorite things in university! The class was digital logic in the electrical engineering department and was a lot of fun. We even got to write machine code programs for it.

Wasn't there someone who did a physical discrete logic implementation of the NAND to Tetris design?

I feel like that could be a good choice for someone doing this today because there's now a kind of a large ecosystem of resources for that design (not that it's inherently better somehow).

yep https://youtu.be/L-azf9ecvfo

Always love seeing Ben Eater’s “build [X] from scratch” His YouTube is one of the best channels to understand how stuff works from fundamental building blocks.

I can trace back the moment I saw YouTube as a tool for learning when I first saw your 8-bit computer series. I was honestly amazed with the quality of the videos and how clear you can explain things. Top content. I'm a subscriber and a fan to this day. Keep up the good work.

I kinda judge YouTube channel by the inverse of the speed I need to scale down/up.

Ben Eater is always among the top 10.

Ben's videos are fantastic, learned a lot. For HW beginners like myself I suggest to skip the computer but just build the clock - very easy and great testing tool!

Here's my question: Obviously there's tremendous interest in kits like this. Why doesn't Ben or someone else create small company to:

  * Modernize the components 
  * Update the videos
  * Add additional modules to make the computer a bit more powerful

People would pay money for this. I would.

Isn't this essentially what Sparkfun and Adafruit do? There's very little margin in the components & kits, and the market is very tiny. Also if you're looking to make this computer "more powerful" you should look at something else. The specs for this are a toy computer, barely enough to demo the basic functions.

It is a rite of passage, as electronics wizards say. I remember at least another good example shared in the wild: https://github.com/vascofazza/8bit-cpu

Check this yt channel out: https://youtu.be/ENlcDdyUHC8?si=cq7zob46DX-oRRaj

RE "....Cutting and laying out wires are some of the most tedious and frustrating process....." I would be investigating to use wire wrap. I built some circuits using wire wrap, back in late 70s ,early 80s. I designed and built a circuit on a wire wrap prototype board to put a 68008 DIL chip on the Apple II bus. I'm not sure of the cost of wire wrap parts now - may be quite expensive , the special wire wrap sockets etc .....

Just wanted to add that there is "a homebrew computer ring" project [1] and you will be surprised how many ideas for DIY computers you can find there.

(I also made a two-breadboard CPU with ~5 instructions myself but had yet no time to finish the software)

[1] https://www.homebrewcpuring.org/

James Sharman has been getting quite advanced with his homebrew cpu. He's got audio and VGA working too and was demonstrating a spinning 3D cube recently. https://www.youtube.com/@weirdboyjim

I tried something similar when in college. Built a 8bit GPU on a breadboard, i don't remember all the details now, but it was such a fun project.[1]

[1] https://nabraj.com/blog/8-bit-gpu/

If figuring out how computers work starting all the way from logic gates interests you, but you don’t want to start by ordering parts and soldering, check out the ”game” Turing Complete.

Of course it’s not the same thing as actually building the electronics! But it can be a nice introduction.

I'm a big fan of all the builds and videos Ben Eater puts together, I've learned more watching his demos than a couple of years doing hardware design in college.

I wonder how would it compare to an old 8 bit CPU like 6502 or Z80.

It is vastly underpowered and has a huge bill of materials compared to those chips. It only supports 16 bytes of memory, has an impoverished instruction set, is slow as molasses, and tends to be very flaky.

Even as a pure-ttl building exercise, it trades off efficiency and usefulness in favor of understandability and fancy blinking-lights so you can see what is happening.

It is useful only for toy programs at the most.

But as a learning exercise it is absolutely first rate. 10/10 would recommend.

Love me some "Ben Eater does this and crushes it"

I built this kit. It was a lot of fun but very tedious. I also have a computer engineering degree so I didn't really learn much. It was fun to show to people though.

If you don't understand what a register is for example though, OH MAN. You'll love this!

Another "interesting" thing to do with this project is build it in a "Digital Logic Simulator"

You need to enable JavaScript to view this site.

Nope. Why does this page need to be an interactive webapp? Ideally it'd be viewable on the 8-bit computer you built from scratch too.

Yo. It's 2025. We don't need to rage at JS, a little sprinkling is OK.

Also, can one do a TCP stack in 8 bits?

a little sprinkling is OK.

This isn't progressive enhancement, the page is completley unreadable without running JS. It rather irks me that someone who clearly knows about the basics of computing would do such a thing.

Also, can one do a TCP stack in 8 bits?

Yes. Lots of low-end IoT devices have an 8-bit MCU. Also, this:

https://www.c64-wiki.com/wiki/Contiki

The site was assembled using webpack and at least one element has a `data-react-helmet` attribute, so my guess is it's a React or React derivative such as Next.js. You're right -- it's not progressive enhancement and it's not necessary to present a simple tutorial.

Maybe there are other reasons, though? Maybe Ben wanted to learn the framework? Maybe Ben wanted to integrate a storefront and a comment section -- which he has done via Shopify and Disqus -- and his choice of framework made that easy due to the existence of official React libraries from both vendors? Maybe Ben's using cloud-native serverless deployment tools alongside his React application because some of the derivative frameworks like Next.js do really well at minifying, compressing, caching, and serving only the content that's needed?

I can understand the purist argument -- I really can, especially from a security standpoint given what we've learned in the past few years about cache-timing side-channel attacks -- but is it possible that Ben is simply a fan of learning and he's as excited about building a web application in React as he is about building an 8-bit computer from scratch?

It would be more efficient to implement this with an FPGA. An FPGA will scale with more complex designs.

The finbonacci on this computer should be faster than rust.

Reply on news.ycombinator.com