been following his videos and built a similar version with upgrades; 256 bytes memory, 8bit PC and MAR, more instruction cycles, Input output (I can load in a program from an external ROM attached onto the IO) and a few conditional instructions. Built on solder boards instead of breadboards (more reliable and permanent). Had my father help build a wooden case to hold my logic cards (5 in total, with a 50pin IDC ribbon up the back as the backplane) and made a front panel, check my channel for a few vids on it
256 bytes isn't enough for me. I'm going to put 64 KiB (65536 bytes) with a 16-bit address! A bit more difficult, but a lot more RAM. And add some dedicated video logic to it! BTW, I'd like to see your implementation of the program loading from ROM. I've got some ideas on how to do that, but it'd be interesting to see various possibilities.
I'm sure we'll find a way down the road to do a direct keyboard hookup that fits within the minimalist spirit of the rest of the system. The challenge is to do it without adding dedicated chips.
The system could make every key one bit in its memory. Would cost about 13 Bytes. Decoding the keyboard into that state is a bit more difficult. PS/2 usually works with interrupts, but this system does not have any...
As of late September 2018 (?), there is a PS/2 keyboard interface for the Gigatron. It uses a tiny microcontroller chip to do the work, and the chip also includes Tiny BASIC in its ROM that can be loaded using the Gigatron's Loader utility. BASIC programs you write can be saved to the controller chip's non-volatile RAM. There is also a new EPROM for the Gigatron that has the Tiny BASIC, as well as the WOZ monitor and some other software built in. The adapter is very inexpensive, about 10 Euros, in the form of a kit that has only a handful of components. It just plugs into the rear of the Gigatron using the game controller connector. It replaces the normal game controller, with keys mapped to the same functions. With this simple accessory, the Gigatron becomes a useable computer instead of just a fun demonstrator toy.
Wow.....this one ticks all the boxes. Stunning work. Comprehensive was my first thought but that doesn't even cut it. Well done and thanks Dave for the review!
Not missed in the sense that we didn't think about it. We've just not figured out yet how to do direct keyboard hookup in the same minimalistic spirit of the rest of the I/O. Even PS/2 is pretty though at this level, and adding a few chips for just that feels wrong. We did think about the cartridge interface. The ROM is easily replaced however.
Very Nostalgic. That's like the original 1960's era minicomputers I had to learn on,. All "flat-pack" or DIP TTL logic gates. We used to troubleshoot down to the chip and gate level to effect repairs. I remember having to manually going through instruction decoding. and the like.
I can't speak for the implementers, but I think Dave's assessment of this is a little off point; and perhaps too harsh. From my perspective, this is an excellent demonstration of early computing. I mean *really* early computing. This has a gate count on par with some advanced computers still in the tube era. Having a RISC processor and interpreter puts us into the Apollo AGC timeframe, which helped land humans on the moon. (Having an interpreter was a way balance speed and storage for complexity of calculation, hence the Mandelbrot demo.) This is a computer (definitely not personal in any way) that reminds us of the history of computing. Today, we rely on computers that have roughly a hundred million times more gates than this kit does. It's easy to forget just how much functionality can be implemented with a few thousand gates. Video, audio, and other peripherals from the modern era (e.g. the 60's) just don't fit with this system.
Totally agree, I just got a 4008 full adder talking to a 4051 BCD input to 7 segment display output and it felt like I had invented the wheel, it can only add up to 15 digit numbers, but that's not the point.
I would argue his criticism is on point here: really the only thing he really ragged on was the manual, which I was also disappointed in. Sure, it looks great, but is there any value added by the manual? Why print another guide to soldering, which is probably better covered elsewhere, vs. a detailed explanation of how their RISC works? I've met plenty of programmers who would be very interested in something like this kit, but there's nothing in that manual they would be interested in. Anyone who can follow just the schematics doesn't need a kit like this to understand basic system architecture. Although I thought Dave was off-base in criticizing the limited instruction set, that could actually make this a better learning experience. It's just a shame they obviously spent so much time on the manual, and ended up producing an entirely useless document. Who, even kids, is going to type in source and compile it vs. download? You absolutely *need* the internet to even be aware of this product. Oh, and you are way off base with your time scales: think late 70's (e.g. altair 8800). They are calling it a microcomputer, after all.
Exactly. It has a lot of potential. Personally still don't understand that price. Although it looks cool, I also don't understand the dozen of different prototype iterations. I mean this is something you can typically simulate, calculate or even prototype with a ucontroller first. The rest is just schematics and PCB layout. I personally don't see were the 200 bucks are coming from. It's not the parts or very special board layout. So yeah, once again cool project with potential if it was just half the price. Even for schools 200 bucks is a bit much to ask for.
I really wanted one of these, until I saw Ben eaters 8 bit breadboard TTL computer. It’s not cheaper by any means, but it teaches a lot more than this. It really teaches the absolute basics of a CPU, and with that you can expand upon the idea. People have made some really cool modifications to the design he uses.
I believe the Gigatron Control Language is to abstract you away from the timings needed to successfully output a VGA signal. The 'Operating System' handles that for you.
Just found this command in github repo, was it mentioned in manual and Dave missed it? Sadly since its Harvard you have to burn your own eprom before being able to execute anything low level :( so its useless out of the box. Alternative is copying Xerox Alto, dividing code space into rom and ram for microcode patches.
This is a treat to watch for the generation like me who learned microprocessor and cpu architecture from Intel SDK85 kit. I built one like this on a universal PCB and enjoyed creating sound synthesizer. Brought back my memories and time i had lots of fun time.
The part with the watch and the monkey was incredibly exciting! Also, the Gigatron build was pretty entertaining too! ;P The 4 hour stream was actually cool! I'm rarely up late enough to catch streams from your side of the world.
150 Euros is NOT EXPENSIVE! Good lord we are Engineers! This is a bare bones Harvard architecture TTL based microcomputer! The time they put into this is WAY worth a measly 150 Euros! I am already working on a basic compiler (not interpreter but come on we all have rom flashers). This is VERY well done. Oh, and I am thinking of putting together a pic based solution to translate a standard usb keyboard in through the controller port.
Kevin Lyon "We" are not all engineers. You might be, but some people are students or hobbyists etc and frankly, for a kit like this, I'd expect the later to be the target market. So I think Dave's criticism has merit. It would be a great investment however, if they took onboard his suggestions about improving the educational aspect of the product. It seems they may only need to adapt the presentation material slightly in order to make this a reality. In my opinion this review should be viewed as invaluable constructive criticism.
They have to end the documentation somewhere. josscheng - why don't you write up the documentation to improve the educational aspect of this product. Just do it if it so easy!
Sure I'll do that jim, if they want to pay me. I don't know why they would want to pay someone else, since, as I have said they would probably just need to slightly adapt the presentation that is on their own website. On their website they say "Learn what happens inside a CPU" this review has pointed out that they don't really provide you with this information and has suggested they may amend that. I don't know what your problem is. Here is a comment from one of the designers: "As one of the designers, it is only good that Dave points out exactly what it is and also what it is not. The board is its own strange thing with some quirky self-imposed constraints. That makes it not really cheap and it can't do all of the things you might expect, or do them in the same way. It is also a hobby project for us, and it is only great that there is an audience that can appreciate it. But best point out the pros and cons rather than have disappointments down the road. I'm really flattered by Dave taking so much time for doing this episode and doing the research."
Absolutely excellent. Think you're being a tiny bit harsh with the review too Dave. It's obviously designed to be a learning experience and fits somewhere between Ben Eater's series which for many is a steep learning curve and a cheap gaming console where nobody learns anything. Having the games to play at the end already in the EPROM is clever and will attract young kids, as for the Mandlebrot generator, it's astonishing even at that speed. Bit banging the lo-res VGA output and audio is very cool. I'd like to build one and see if I can write a compiler for it. I've built something very like this in the past, but didn't spend countless months writing all that software. Well done guys - a truly brilliant, educational product.
If anything, the Gigatron is missing two things: a bus interface, and maybe an experimental section of the PCB for additional circuitry. This computer is purely for software geeks, and encourages absolutely ZERO hardware expansion via TTL. As a kit, what the Gigatron should have included is a logic probe. They're cheap enough to be thrown in. If the student has problems with the build, the logic probe would be really handy for troubleshooting. Maybe dedicate a chapter in the manual for troubleshooting - maybe add in some test points on the PCB, and document those test points in the documentation.
Awwww! I was hoping to see real time video of the actual construction, not just a time lapse, even if it wasn't the full 3 hours or so from the live show.
Now we need the similar kind of DIY kit for Mechanical Relay Computer. TTL is fun, but switching relay music is 100x more fun! Gigatron, let's go down more into the roots.
Classic pics from early computers too... I remember looking at some of those on my old PC/XT. Also, the "can't do anything else with it" point Dave mentions is actually quite familiar to anyone who's used the old, old micros. So many of them were limited by not having a tape drive or floppy (hard drives just didn't exist at reasonable prices at the time) or by needing an "extra" hardware card for graphics.
A few extra features would push this from a good learning tool to a great one as you touched on. A push button to clock an instruction cycle through each instruction with LEDs indicating the movement of the opcode+data from the EEProms to the IR and then into the various registers would be a great way to learn. These days a CPU is a closed circuit and can only be seen by a student in a simulator or by implementing one in an FPGA. It's nice to be able to access the internals and scope it out and step through it manually to fully understand the workings of a CPU.
It is. But one could certainly have fun with it after it. Dave was just too freaking lazy to hook up less than 10 jumper wires to one of his 1000 Arduinos and download own programs to it.
This is a good review, but please note they are have updated the kit with a new v3 ROM and an adapter that allows for PS/2 keyboard input. The new rom has a Basic interrupter, Tetris and Break out games to boot.
Dave, you mentioned in your live stream about over clocking. I think that since the microprocessor is generating all of the VGA signals overclocking may result in video problems outside of a very narrow bandwidth. I agree that the designers missed out on a great teaching opportunity. This would be an ideal type of platform to teach how microcode works. Instead of "Cylon" style blinking LED's it would be more useful to have an accumulator and program counter display with stepping and load switches (like on the old IBM 1130 computer). They could have taken a lesson from the old RCA "Cosmac" 1802 based computers. Those spawned an entire subculture of hardware and software hackers. OK, I'm showing my age now. Cheers, Ian.
EEVblog Not much. Most "later" VGA monitors are "multi sync" (tm) or similar, autodetecting and adapting to the sync frequencies within a broad margin.
One guy on Twitter overclocked it to 8MHz and it appeared to work with video. He replaced some IC's with faster ones and pushed it to 10MHz, but at that point the monitor didn't accept the signal. Would need a different ROM there. twitter.com/Monsonite/status/989913209418248194
It would be nice to have a kit like that that could run software that was written for historical 8-bit computers, kind of a new build-it-yourself 8-bit computer kit. Allow for 6502 and Z80 CPUs and the ability to make them behave like an Apple, a Commodore or whatever else.
It is amazing what you can do with some TTL logic. I've been watching someone build a CPU with logic gates, and it does seem like an interesting kind of project to think about. Personally it makes me want to try and design an old school graphics chip. The kind of thing a 8 or 16 bit system might've been using. Not quite as featureless as VGA, but hopefully also not so complex that it requires thousands of logic chips to implement. XD I did wonder about timing though; video output is rather timing critical, so you can't for instance get away with designing a graphics chip that runs at lower frequencies. You can do that with model processors, but not with graphics output. I considered a way around this might be to cheat. Create something that produces a VGA-like signal from a framebuffer at high speed, but then couple a much slower system to it that updates at it's own pace. The high speed section ensures the signal output timing is correct, while the low speed section does all the interesting stuff... Still, I guess the starting point would be creating a valid video signal first. XD
This is a lovely small CPU implemented with an impressively small number of gates. One thing that I'd point out though - there's not a single TTL part on this board. These are all 74HCT parts, which are CMOS that use TTL signal levels. Switching the whole thing to 74XX (or 74S or 74LS or 74H) parts would make it true TTL, and would up the power requirement quite significantly. Still a neat little computer - I wish they'd included an expansion port (bus) and a serial port - skip the modern video and keyboard interfaces.
A nice kit, I watched Dave assemble this the other night and it worked first time. It is what it says on the box and I doubt it could be expanded. Though a Basic rom would have been nice and a keyboard port. But that would have required some Serial interface to be added and then it has become something else.
yeah i get that...pritty strange still.. ive been live streaming alot lately and they all show up soon as the stream is finished..usually show the "processing" tag for a while..even tho its still entirely viewable before the processing is complete
Interesting! However, if I'm making an 8-bit computer, I want to customize it. So I'm following Ben Eater's 8-bit computer series and the improvements made by James Bates. And I'm going to improve it further. And there's really no way of programming it in assembly without reprogramming the EPROM or something?
One name ben eater check him out! I really enjoyed building his 8bit micro and the best thing about it is you can add to and modify the design highly recommend it
For the same price, you can get the IBM 5150 clone kit from MTM. On one hand, once you finish assembling it, you get a functional IBM-compatible PC. On the other hand, I don't think it comes with such a lovely and complete documentation, so it might not be as suitable for beginners.
What do you think of it as an educational tool? I think I did better when I started with some breadboards and a handful of logic gates, latches, and timers!
Made a design for a Brainf*ck CPU once, using only TTLs. It would take quite a few chips, though, mainly due to only 4 bits per counter IC and I wanted a 16-bit address space.
Kudos to Gigatron. Your reviews are brutally honest but they are your honest opinions. While I disagree that it is not worth the $200, I can agree that Gigatron missed the boat a bit (and revenue potential) by not making it more expandable and with more interfaces. And with more LEDs to show what is going on.
Seems like it wouldn't be *too* hard to build a keyboard interface for it... but you'd need the firmware to make it programmable. Also hacking in some storage might not be as easy
"These are purposely shipped with 'positive' misalignment to facilitate insertion" Fairchild TTL data book (dead tree edition) Are there machines that grip these chips in such a way as to make this true ?
DIP IC insertion tool: even Radio Shack sold them. Maybe turn on safe searching before googling. Of course, you were putting these in sockets because you didn't want to damage them and it made repair easier. That was the idea back then anyway.
Jeez Dave, those guys did a fantastic job on the manual and kit. They put way more effort into it than most legit companies do. Why do you feel the need to spend 1/4 of the time of the video pointing out all the trivial negativism's? Cut them a break.
I said it's a nice kit and manual and praised it as such many times. But as a *computer*, especially one at this price, it's not very useful, so they are hardly trivial. If you can't differentiate comments in regards to the kit itself, to those as a computer then there is nothing I can do for you.
As one of the designers, it is only good that Dave points out exactly what it is and also what it is not. The board is its own strange thing with some quirky self-imposed constraints. That makes it not really cheap and it can't do all of the things you might expect, or do them in the same way. It is also a hobby project for us, and it is only great that there is an audience that can appreciate it. But best point out the pros and cons rather than have disappointments down the road. I'm really flattered by Dave taking so much time for doing this episode and doing the research.
Well let me add my congratulations, it's a thing of beauty. Most people who aren't very experienced in software or hardware design will probably give you a 'meh' comment, but those of us who are will be praising your effort and skill. What you've done is remarkable, in fact better than any early 1980s machine based on a Z80 or 6502. Been reading through the code on your Github repo and it's awesome to see people already building a little ecosystem around it.
Gigatron runs at 6.25Mhz, that is not slow for 8bit retro computer, and if I'm not mistaking it fetch data of next instruction when still working on instruction, like the 6502 does I think and not like the z80 where date always get fetch after the instruction is done.
@@EEVblog Yes - I presented a Gigatron to Chuck Moore last year - on the occasion of his 81st birthday - just in case he wants to do a Forth implentation on another cpu ;)
That seems very familiar. I had designed computer from logic in software. I wanted to build it, but I was put off by the need to route roughly 1000 quad gate chips. At least I know why there are millions of transistors in modern CPU.
Less soldering than an Altair with 32k memory and a front panel! A lot less. Believe me, I know. I used to build them for people and debug those built by others.
I love the Philippine mahogany/lauan case. All the looks and feels of real mahogany, none of the pesky whinging from conservationists. It's one of my favorite tonewoods for guitars.
Something you might be interested in is the Colour Maximite which is a microcomputer with a BASIC interpreter using a PIC32. You can by pre-assembled kits, solder yourself kits, or just buy the parts yourself and build one. Here's the page with info on the project: geoffg.net/maximite.html Here's a video The 8-Bit Guy did on it: th-cam.com/video/XQA8lowEKOo/w-d-xo.html
Been thinking of getting one. Build a keyboard interface and add BASIC to the ROM. Although, would it be cheating if the keyboard had a processor in it?
Depends on the keyboard type. Matrix keyboard might be easy. Repurpose 3 LEDs to drive the matrix rows (use one 3-to-8 decoder), and put the output lines on the bus with a bus buffer chip. Even PS/2 will not be easy to do with just simple chips. The problem is the timing and the two-directional data line. Adafruit has a small PS2-to-TTL connector selling for $17.95. There is a small controller in it. It just outputs ASCII codes in TTL levels I think. The Arduino route will be quickest, but clearly cheating. First something that works. Than something that is minimalistic.
It's more 'Don't send me what I don't need or have space for'. David doesn't have a warehouse to store everything after all. He uses the donations to further his channel related projects.
Brek Martin his list of stuff that he wants is so he can make the videos he wants to for the channel. It's better than just getting people to spend tons of money sending him things he doesn't have any use for and will just sit and collect dust.
Makes you wonder why they don't just interlace, then you get the faster speed, and no apparent lines? Would be nice to mod this and add a video output chip, don't know what kind of external expansion this has... worst case, you could put it on the CPU bus with enough VRAM to hold a frame buffer, a few microcode/ROM changes so graphical instructions are handled differently. Certainly do-able, and begs the question why they didn't already?...
Ewen Wordsworth at Microbee technology has announced the release of a limited run of the new Microbee kit computers coming out in a couple of days Dave, just in case you are interested......
I must admit with the DIP chips and their splayed legs, I wonder why Manufactures still do that. Like every DIP socket, DIP TH PCB, even BreadBoards need them squeezed in.
Hi Dave, could you try overclocking it? They say on Gigatron FAQ that by installing faster RAM and changing clock you should be able to get over 10MHz (6.25MHz stock)
Oh, probably. I don't have the knowledge to foresee such complications, I've just read their FAQ. Overclocking was the first thing that came to my mind when I saw how slow that image was loading
Cool, but way too expensive. It should include a video chip, keyboard input and some sort of non volatile memory for the price. It would probably be pretty quick if it didn't have to bit bang the vga output.
EEVblog Or maybe implement the classic trick from the ZX81: Provide an even faster mode with the video output off, so a program can turn off the video, run it's calculations at the full 6Mips, then display the result. Also assembly language support would have been more fun than the interpreter.
It would have been very nice had they decided to include either a card edge connector or a port connector that brought out all of the data, address, clock, and interrupt lines (an old Centronics printer port would likely work nicely for such). Had they done this then you could expand upon the Gigatron through expansion projects that could have added video generation, I/O ports (incl. keyboard), external storage, etc. and build it to be whatever you wanted it to become, including the opportunity to design custom projects. Even better if they buffered that expansion port to protect the Gigatron itself so that design mistakes would at most cost the owner a new buffer chip to repair any serious mistakes. I also agree that it would have been very nice to have LEDs on the various I/O lines along with a variable clock so that you could slow the clock to actually see data moving around on the various bus lines and register states...though a variable clock would break VGA video generation as implemented unless an external video generator expansion with its own dedicated clock were available and used instead.
Steve Pocock it was the ZX80 that had FAST and SLOW modes, it was one of the bigger difference between the 80 and the 81 (especially in terms of running 80 basic on the 81) Though of course you could turn your 80 in an 81 for the most part by dropping an 81 ROM into it.
Oh yeah I'm well familiar with that from experience. I know TTL is a term that's basically synonymous with any 74/4000 series logic, even if they are CMOS chips. It's just a bit of a confusing term because personally when I got started in electronics I assumed all TTL chips used BJTs. It blew my mind when I found out the HC/HCT series among others were CMOS versions. But that's why it's so important to read the datasheet, I suppose.
Instruction set reminds me a bit of BrainF-ck, except it’s a bit more verbose than BF. Only hardcore engineers use BF! Google it and then tear your eyes out learning what exactly constitutes a Turing-complete ISA...
Mitch Coté Yes, there were 8 opcodes and each opcode was a single character, making reading BF “programs” (I prefer the term abominations, ha) absolutely miserable. Yet people still went on to write various things in it, the site esoteric.sange.fi/brainfuck has a ton of compilers, examples, and history on it. I don’t think too many people actually use BF, more of an “Imza e7331 hax0r” (And oh how spellcheck fought me, trying to write that expression out) type of thing, though it was obviously popular just because it was so obscure and difficult.
![[LIVE] : ONE ลุมพินี 80 | คู่เอก "รักษ์ vs ยอดนำชัย"](http://i.ytimg.com/vi/ckLMRorsMeY/mqdefault.jpg)
If you want details on 8 bit Computer look at Ben Eater "Building an 8-bit breadboard computer" play list.
been following his videos and built a similar version with upgrades; 256 bytes memory, 8bit PC and MAR, more instruction cycles, Input output (I can load in a program from an external ROM attached onto the IO) and a few conditional instructions. Built on solder boards instead of breadboards (more reliable and permanent). Had my father help build a wooden case to hold my logic cards (5 in total, with a 50pin IDC ribbon up the back as the backplane) and made a front panel, check my channel for a few vids on it
256 bytes isn't enough for me. I'm going to put 64 KiB (65536 bytes) with a 16-bit address! A bit more difficult, but a lot more RAM. And add some dedicated video logic to it! BTW, I'd like to see your implementation of the program loading from ROM. I've got some ideas on how to do that, but it'd be interesting to see various possibilities.
Here's hoping for a second generation with a native programming environment!
It would be nice, but in its current stage, a PS/2 keyboard would be pretty difficult to decode.
I'm sure we'll find a way down the road to do a direct keyboard hookup that fits within the minimalist spirit of the rest of the system. The challenge is to do it without adding dedicated chips.
The system could make every key one bit in its memory. Would cost about 13 Bytes. Decoding the keyboard into that state is a bit more difficult.
PS/2 usually works with interrupts, but this system does not have any...
As of late September 2018 (?), there is a PS/2 keyboard interface for the Gigatron. It uses a tiny microcontroller chip to do the work, and the chip also includes Tiny BASIC in its ROM that can be loaded using the Gigatron's Loader utility. BASIC programs you write can be saved to the controller chip's non-volatile RAM.
There is also a new EPROM for the Gigatron that has the Tiny BASIC, as well as the WOZ monitor and some other software built in.
The adapter is very inexpensive, about 10 Euros, in the form of a kit that has only a handful of components. It just plugs into the rear of the Gigatron using the game controller connector. It replaces the normal game controller, with keys mapped to the same functions.
With this simple accessory, the Gigatron becomes a useable computer instead of just a fun demonstrator toy.
I'm working on building a RISC computer myself. I hope to have it finished within a couple of months. I'll definitely be sending you the kit!
Wow.....this one ticks all the boxes. Stunning work. Comprehensive was my first thought but that doesn't even cut it. Well done and thanks Dave for the review!
Not missed in the sense that we didn't think about it. We've just not figured out yet how to do direct keyboard hookup in the same minimalistic spirit of the rest of the I/O. Even PS/2 is pretty though at this level, and adding a few chips for just that feels wrong. We did think about the cartridge interface. The ROM is easily replaced however.
Very Nostalgic. That's like the original 1960's era minicomputers I had to learn on,. All "flat-pack" or DIP TTL logic gates. We used to troubleshoot down to the chip and gate level to effect repairs. I remember having to manually going through instruction decoding. and the like.
I can't speak for the implementers, but I think Dave's assessment of this is a little off point; and perhaps too harsh. From my perspective, this is an excellent demonstration of early computing. I mean *really* early computing. This has a gate count on par with some advanced computers still in the tube era. Having a RISC processor and interpreter puts us into the Apollo AGC timeframe, which helped land humans on the moon. (Having an interpreter was a way balance speed and storage for complexity of calculation, hence the Mandelbrot demo.)
This is a computer (definitely not personal in any way) that reminds us of the history of computing. Today, we rely on computers that have roughly a hundred million times more gates than this kit does. It's easy to forget just how much functionality can be implemented with a few thousand gates. Video, audio, and other peripherals from the modern era (e.g. the 60's) just don't fit with this system.
I agree. As someone who spent a lot of time with TTL back in the day, I thought this was amazing.
Greg Krekelberg . Agree with you, although 200 bucks is just an awful lot.
Totally agree, I just got a 4008 full adder talking to a 4051 BCD input to 7 segment display output and it felt like I had invented the wheel, it can only add up to 15 digit numbers, but that's not the point.
I would argue his criticism is on point here: really the only thing he really ragged on was the manual, which I was also disappointed in. Sure, it looks great, but is there any value added by the manual? Why print another guide to soldering, which is probably better covered elsewhere, vs. a detailed explanation of how their RISC works? I've met plenty of programmers who would be very interested in something like this kit, but there's nothing in that manual they would be interested in. Anyone who can follow just the schematics doesn't need a kit like this to understand basic system architecture.
Although I thought Dave was off-base in criticizing the limited instruction set, that could actually make this a better learning experience. It's just a shame they obviously spent so much time on the manual, and ended up producing an entirely useless document. Who, even kids, is going to type in source and compile it vs. download? You absolutely *need* the internet to even be aware of this product.
Oh, and you are way off base with your time scales: think late 70's (e.g. altair 8800). They are calling it a microcomputer, after all.
Exactly. It has a lot of potential. Personally still don't understand that price.
Although it looks cool, I also don't understand the dozen of different prototype iterations.
I mean this is something you can typically simulate, calculate or even prototype with a ucontroller first.
The rest is just schematics and PCB layout.
I personally don't see were the 200 bucks are coming from.
It's not the parts or very special board layout.
So yeah, once again cool project with potential if it was just half the price.
Even for schools 200 bucks is a bit much to ask for.
The latest update of the rom, support keyboard and basic
Absolutely fantastic. Good stuff Dave.
I really wanted one of these, until I saw Ben eaters 8 bit breadboard TTL computer.
It’s not cheaper by any means, but it teaches a lot more than this.
It really teaches the absolute basics of a CPU, and with that you can expand upon the idea.
People have made some really cool modifications to the design he uses.
I believe the Gigatron Control Language is to abstract you away from the timings needed to successfully output a VGA signal. The 'Operating System' handles that for you.
this is true, but they could let you run small pure microcode snippets, limited in size to Vblank/raster
The SYS instruction does exactly that
Just found this command in github repo, was it mentioned in manual and Dave missed it?
Sadly since its Harvard you have to burn your own eprom before being able to execute anything low level :( so its useless out of the box. Alternative is copying Xerox Alto, dividing code space into rom and ram for microcode patches.
I solemnly swear not to underestimate the power of TTL ever again.
From ESD protection point of view, soldering capacitors first actually makes sense.
As limited as it is, This is the kind of stuff that inspired me to get started in hobby electronics, give it to a kid and move on to bigger projects!
I'm glad that Dave dud this short review video as I can't see myself sitting through the full 4 hour video.
This is a treat to watch for the generation like me who learned microprocessor and cpu architecture from Intel SDK85 kit. I built one like this on a universal PCB and enjoyed creating sound synthesizer. Brought back my memories and time i had lots of fun time.
The part with the watch and the monkey was incredibly exciting! Also, the Gigatron build was pretty entertaining too! ;P
The 4 hour stream was actually cool! I'm rarely up late enough to catch streams from your side of the world.
That manual is a thing of beauty. As is the finished thing itself. Pity it's so limited.
150 Euros is NOT EXPENSIVE! Good lord we are Engineers! This is a bare bones Harvard architecture TTL based microcomputer! The time they put into this is WAY worth a measly 150 Euros! I am already working on a basic compiler (not interpreter but come on we all have rom flashers). This is VERY well done. Oh, and I am thinking of putting together a pic based solution to translate a standard usb keyboard in through the controller port.
Kevin Lyon "We" are not all engineers. You might be, but some people are students or hobbyists etc and frankly, for a kit like this, I'd expect the later to be the target market. So I think Dave's criticism has merit.
It would be a great investment however, if they took onboard his suggestions about improving the educational aspect of the product. It seems they may only need to adapt the presentation material slightly in order to make this a reality.
In my opinion this review should be viewed as invaluable constructive criticism.
They have to end the documentation somewhere. josscheng - why don't you write up the documentation to improve the educational aspect of this product. Just do it if it so easy!
Sure I'll do that jim, if they want to pay me. I don't know why they would want to pay someone else, since, as I have said they would probably just need to slightly adapt the presentation that is on their own website.
On their website they say "Learn what happens inside a CPU" this review has pointed out that they don't really provide you with this information and has suggested they may amend that. I don't know what your problem is.
Here is a comment from one of the designers:
"As one of the designers, it is only good that Dave points out exactly what it is and also what it is not. The board is its own strange thing with some quirky self-imposed constraints. That makes it not really cheap and it can't do all of the things you might expect, or do them in the same way. It is also a hobby project for us, and it is only great that there is an audience that can appreciate it. But best point out the pros and cons rather than have disappointments down the road. I'm really flattered by Dave taking so much time for doing this episode and doing the research."
When you can get a full computer for 5 USD, 150 EUR IS expensive.
CrimSun - Then don't buy it. No one cares if you don't.
I love the time lapse 👍
Wow, that's retro - takes me back to the early 70's.
Absolutely excellent. Think you're being a tiny bit harsh with the review too Dave. It's obviously designed to be a learning experience and fits somewhere between Ben Eater's series which for many is a steep learning curve and a cheap gaming console where nobody learns anything. Having the games to play at the end already in the EPROM is clever and will attract young kids, as for the Mandlebrot generator, it's astonishing even at that speed. Bit banging the lo-res VGA output and audio is very cool. I'd like to build one and see if I can write a compiler for it. I've built something very like this in the past, but didn't spend countless months writing all that software. Well done guys - a truly brilliant, educational product.
If anything, the Gigatron is missing two things: a bus interface, and maybe an experimental section of the PCB for additional circuitry. This computer is purely for software geeks, and encourages absolutely ZERO hardware expansion via TTL.
As a kit, what the Gigatron should have included is a logic probe. They're cheap enough to be thrown in. If the student has problems with the build, the logic probe would be really handy for troubleshooting. Maybe dedicate a chapter in the manual for troubleshooting - maybe add in some test points on the PCB, and document those test points in the documentation.
Awwww! I was hoping to see real time video of the actual construction, not just a time lapse, even if it wasn't the full 3 hours or so from the live show.
I have an hour and 15 minute video on TH-cam showing my own Gigatron build, shows a lot more than Dave's one here.
Have you seen Ben Eater's breadboard computer and all the other variations he has inspired?
Now we need the similar kind of DIY kit for Mechanical Relay Computer. TTL is fun, but switching relay music is 100x more fun! Gigatron, let's go down more into the roots.
Just finished mine last night!
Classic pics from early computers too... I remember looking at some of those on my old PC/XT.
Also, the "can't do anything else with it" point Dave mentions is actually quite familiar to anyone who's used the old, old micros. So many of them were limited by not having a tape drive or floppy (hard drives just didn't exist at reasonable prices at the time) or by needing an "extra" hardware card for graphics.
I like old-school, but not THAT old-school!
A few extra features would push this from a good learning tool to a great one as you touched on. A push button to clock an instruction cycle through each instruction with LEDs indicating the movement of the opcode+data from the EEProms to the IR and then into the various registers would be a great way to learn. These days a CPU is a closed circuit and can only be seen by a student in a simulator or by implementing one in an FPGA. It's nice to be able to access the internals and scope it out and step through it manually to fully understand the workings of a CPU.
I recognize the photos from a classic screensaver I had in the mid-1980's.
Seems like a cool display piece. Would be fun to build. I can't see it being much fun after that.
It is. But one could certainly have fun with it after it. Dave was just too freaking lazy to hook up less than 10 jumper wires to one of his 1000 Arduinos and download own programs to it.
I think it would be neat to try to use the faster chips from potatosemi.
Oh, fantastic! This would help me get to a next level.
I propose you would do a 2nd revision: now this puppy can do a lot more things, including emulation of 6502 and even Apple-1 !
This is a good review, but please note they are have updated the kit with a new v3 ROM and an adapter that allows for PS/2 keyboard input. The new rom has a Basic interrupter, Tetris and Break out games to boot.
Why not have a third video to explain what each section of chips does?
Maybe this will help? th-cam.com/video/QUfdASs82Lw/w-d-xo.html
Yes! That is excellent. I learned NOTHING from Dave's video - not even what "TTL" was. Lots of giggling and laughing, but no useful info.
A cheap kit computer with a basic interpreter would be really cool...
(Like a spectrum or something)
Dave, you mentioned in your live stream about over clocking. I think that since the microprocessor is generating all of the VGA signals overclocking may result in video problems outside of a very narrow bandwidth.
I agree that the designers missed out on a great teaching opportunity. This would be an ideal type of platform to teach how microcode works. Instead of "Cylon" style blinking LED's it would be more useful to have an accumulator and program counter display with stepping and load switches (like on the old IBM 1130 computer). They could have taken a lesson from the old RCA "Cosmac" 1802 based computers. Those spawned an entire subculture of hardware and software hackers.
OK, I'm showing my age now.
Cheers, Ian.
Yes, video timing would be a problem
EEVblog Not much. Most "later" VGA monitors are "multi sync" (tm) or similar, autodetecting and adapting to the sync frequencies within a broad margin.
video timing would be a problem when single stepping = no accumulator and program counter LEDs
video is generated by the "cpu" itself
One guy on Twitter overclocked it to 8MHz and it appeared to work with video. He replaced some IC's with faster ones and pushed it to 10MHz, but at that point the monitor didn't accept the signal. Would need a different ROM there. twitter.com/Monsonite/status/989913209418248194
96Hz Vsync should work fine on good CRT :)
It would be nice to have a kit like that that could run software that was written for historical 8-bit computers, kind of a new build-it-yourself 8-bit computer kit. Allow for 6502 and Z80 CPUs and the ability to make them behave like an Apple, a Commodore or whatever else.
It is amazing what you can do with some TTL logic.
I've been watching someone build a CPU with logic gates, and it does seem like an interesting kind of project to think about.
Personally it makes me want to try and design an old school graphics chip. The kind of thing a 8 or 16 bit system might've been using.
Not quite as featureless as VGA, but hopefully also not so complex that it requires thousands of logic chips to implement. XD
I did wonder about timing though; video output is rather timing critical, so you can't for instance get away with designing a graphics chip that runs at lower frequencies.
You can do that with model processors, but not with graphics output.
I considered a way around this might be to cheat. Create something that produces a VGA-like signal from a framebuffer at high speed, but then couple a much slower system to it that updates at it's own pace.
The high speed section ensures the signal output timing is correct, while the low speed section does all the interesting stuff...
Still, I guess the starting point would be creating a valid video signal first. XD
This is a lovely small CPU implemented with an impressively small number of gates. One thing that I'd point out though - there's not a single TTL part on this board. These are all 74HCT parts, which are CMOS that use TTL signal levels. Switching the whole thing to 74XX (or 74S or 74LS or 74H) parts would make it true TTL, and would up the power requirement quite significantly. Still a neat little computer - I wish they'd included an expansion port (bus) and a serial port - skip the modern video and keyboard interfaces.
A nice kit, I watched Dave assemble this the other night and it worked first time. It is what it says on the box and I doubt it could be expanded. Though a Basic rom would have been nice and a keyboard port. But that would have required some Serial interface to be added and then it has become something else.
Perhaps the most secure computer in the world!
cool!..was just watching your 4hr build video on this!
The video still does not show up on my creator studio list of videos. Yay TH-cam...
strange...its showing for me on ya channel page...all 4hrs there too...
My video creator page is different to my channel page, only I can see it, and it's were all my videos are listed, stats, edit etc.
yeah i get that...pritty strange still.. ive been live streaming alot lately and they all show up soon as the stream is finished..usually show the "processing" tag for a while..even tho its still entirely viewable before the processing is complete
Interesting! However, if I'm making an 8-bit computer, I want to customize it. So I'm following Ben Eater's 8-bit computer series and the improvements made by James Bates. And I'm going to improve it further. And there's really no way of programming it in assembly without reprogramming the EPROM or something?
One name ben eater check him out! I really enjoyed building his 8bit micro and the best thing about it is you can add to and modify the design highly recommend it
That's the edit I wanted, now I'm happy. Thank you Dave'o. lol
For the same price, you can get the IBM 5150 clone kit from MTM. On one hand, once you finish assembling it, you get a functional IBM-compatible PC. On the other hand, I don't think it comes with such a lovely and complete documentation, so it might not be as suitable for beginners.
Can't wait to order this. Pay day needs to hurry up.
Get the gigatron rom version 3! It has basic and some more games!!!
What do you think of it as an educational tool? I think I did better when I started with some breadboards and a handful of logic gates, latches, and timers!
Well there isn't really education material to go along with it, unless you count kit assembly education as a thing.
Assembly is essential, but it's not EE.
Made a design for a Brainf*ck CPU once, using only TTLs. It would take quite a few chips, though, mainly due to only 4 bits per counter IC and I wanted a 16-bit address space.
Kudos to Gigatron. Your reviews are brutally honest but they are your honest opinions. While I disagree that it is not worth the $200, I can agree that Gigatron missed the boat a bit (and revenue potential) by not making it more expandable and with more interfaces. And with more LEDs to show what is going on.
Not only do we have DaveCad, but now... witness the DavePickNPlace running at full speed!
Seems like it wouldn't be *too* hard to build a keyboard interface for it... but you'd need the firmware to make it programmable. Also hacking in some storage might not be as easy
If you fancy a similar level soldering kit, try one of my Minstrel ZX80 clones. Once built, you can run all the ZX80 and some ZX81 software.
"These are purposely shipped with 'positive' misalignment to facilitate insertion"
Fairchild TTL data book (dead tree edition)
Are there machines that grip these chips in such a way as to make this true ?
DIP IC insertion tool: even Radio Shack sold them. Maybe turn on safe searching before googling. Of course, you were putting these in sockets because you didn't want to damage them and it made repair easier. That was the idea back then anyway.
Geez Louise! How about a 1980 Holden automobile in kit form? Rasperry PI should come with a manual as good as this kit.
Jeez Dave, those guys did a fantastic job on the manual and kit. They put way more effort into it than most legit companies do. Why do you feel the need to spend 1/4 of the time of the video pointing out all the trivial negativism's? Cut them a break.
I said it's a nice kit and manual and praised it as such many times. But as a *computer*, especially one at this price, it's not very useful, so they are hardly trivial.
If you can't differentiate comments in regards to the kit itself, to those as a computer then there is nothing I can do for you.
As one of the designers, it is only good that Dave points out exactly what it is and also what it is not. The board is its own strange thing with some quirky self-imposed constraints. That makes it not really cheap and it can't do all of the things you might expect, or do them in the same way. It is also a hobby project for us, and it is only great that there is an audience that can appreciate it. But best point out the pros and cons rather than have disappointments down the road. I'm really flattered by Dave taking so much time for doing this episode and doing the research.
Well let me add my congratulations, it's a thing of beauty. Most people who aren't very experienced in software or hardware design will probably give you a 'meh' comment, but those of us who are will be praising your effort and skill. What you've done is remarkable, in fact better than any early 1980s machine based on a Z80 or 6502. Been reading through the code on your Github repo and it's awesome to see people already building a little ecosystem around it.
Damn... You build that way faster than the David "The 8-Bit Guy" :P Took only 30 seconds!
The routing on that board is beautiful. I make a mess just routing one TQFP32 chip.
Gigatron runs at 6.25Mhz, that is not slow for 8bit retro computer, and if I'm not mistaking it fetch data of next instruction when still working on instruction, like the 6502 does I think and not like the z80 where date always get fetch after the instruction is done.
I think they could have implemented a forth interpreter rather than spend time on gcl. I'm biased I guess :-) Pretty cool, though.
Wow, there are still Forth fanboys...
We are legion.
Also, graying.
Yes we are.
@@EEVblog Yes - I presented a Gigatron to Chuck Moore last year - on the occasion of his 81st birthday - just in case he wants to do a Forth implentation on another cpu ;)
That seems very familiar. I had designed computer from logic in software. I wanted to build it, but I was put off by the need to route roughly 1000 quad gate chips. At least I know why there are millions of transistors in modern CPU.
I think they said do the caps first because you can't really destroy them with heat and such
maybe they should write LLVM component to output binary code, instead of having an interpreter.
It looks like Schematics are made with Kicad
Less soldering than an Altair with 32k memory and a front panel! A lot less. Believe me, I know. I used to build them for people and debug those built by others.
I love the Philippine mahogany/lauan case. All the looks and feels of real mahogany, none of the pesky whinging from conservationists. It's one of my favorite tonewoods for guitars.
Similar idea to the ZX80.. while that used a Z80 CPU, everything was done by the CPU including the display.
Something you might be interested in is the Colour Maximite which is a microcomputer with a BASIC interpreter using a PIC32. You can by pre-assembled kits, solder yourself kits, or just buy the parts yourself and build one.
Here's the page with info on the project: geoffg.net/maximite.html
Here's a video The 8-Bit Guy did on it: th-cam.com/video/XQA8lowEKOo/w-d-xo.html
Been thinking of getting one. Build a keyboard interface and add BASIC to the ROM. Although, would it be cheating if the keyboard had a processor in it?
Depends on the keyboard type. Matrix keyboard might be easy. Repurpose 3 LEDs to drive the matrix rows (use one 3-to-8 decoder), and put the output lines on the bus with a bus buffer chip. Even PS/2 will not be easy to do with just simple chips. The problem is the timing and the two-directional data line. Adafruit has a small PS2-to-TTL connector selling for $17.95. There is a small controller in it. It just outputs ASCII codes in TTL levels I think. The Arduino route will be quickest, but clearly cheating. First something that works. Than something that is minimalistic.
As others are saying, check Ben Eater's breadboard computer out! It's amazing!
Defo a fun project, and probs fun to code for it! :D
Maybe mistaken, but IIRC 8 Bit Guy’s was sent fully assembled. Funny that!
Brek Martin I seem to recall him having trouble identifying resistors due to color blindness. Or maybe that was for another through hole kit.
John Francis Doe I’m not a subscriber to know. I don’t like his “Send me what I want” style. Comes over as self, rather than viewer motivated to me.
It's more 'Don't send me what I don't need or have space for'. David doesn't have a warehouse to store everything after all. He uses the donations to further his channel related projects.
I'm 99% certain either he assembled it himself or had a friend do it. There was a close-up of someone's hands doing the work.
Brek Martin his list of stuff that he wants is so he can make the videos he wants to for the channel. It's better than just getting people to spend tons of money sending him things he doesn't have any use for and will just sit and collect dust.
I could live without the lack of software, but man, why did they not add more blinken lights! at least some 7-segment displays smh
i fell asleep durring the live stream xD
Makes you wonder why they don't just interlace, then you get the faster speed, and no apparent lines?
Would be nice to mod this and add a video output chip, don't know what kind of external expansion this has... worst case, you could put it on the CPU bus with enough VRAM to hold a frame buffer, a few microcode/ROM changes so graphical instructions are handled differently. Certainly do-able, and begs the question why they didn't already?...
Hey Dave The 8 Bit Guy beat you to this!
8:10 - An OVAL planet!? Set the monitor for 4:3!
8 bit guy just did this to.
Too bad there's no real-time solder whistling ;>)
thank you 👍👍👍
Dave, have you checked out the Microbee Premium Plus kit? (It is still available!)
I can't see it for sale.
EEVblog Fair enough, maybe Ewen is getting the next kit run ready.
Ewen Wordsworth at Microbee technology has announced the release of a limited run of the new Microbee kit computers coming out in a couple of days Dave, just in case you are interested......
I must admit with the DIP chips and their splayed legs, I wonder why Manufactures still do that. Like every DIP socket, DIP TH PCB, even BreadBoards need them squeezed in.
Their GCL looks like a form of Forth.
Hi Dave, could you try overclocking it? They say on Gigatron FAQ that by installing faster RAM and changing clock you should be able to get over 10MHz (6.25MHz stock)
Oh, probably. I don't have the knowledge to foresee such complications, I've just read their FAQ. Overclocking was the first thing that came to my mind when I saw how slow that image was loading
I got mine up to 12.5MHz using 74F (Fast) logic, 35nS ROM and a 10nS RAM. Marcel got up to 15 or 16MHz
I'm interested in this sort of thing but finding a good video chip for one is kinda hard.
Cool, but way too expensive. It should include a video chip, keyboard input and some sort of non volatile memory for the price.
It would probably be pretty quick if it didn't have to bit bang the vga output.
Making dedicated video hardware out of TTL chips would have been cool.
EEVblog Or maybe implement the classic trick from the ZX81: Provide an even faster mode with the video output off, so a program can turn off the video, run it's calculations at the full 6Mips, then display the result. Also assembly language support would have been more fun than the interpreter.
I thought that was the ZX80, or am I confused due to old age?
It would have been very nice had they decided to include either a card edge connector or a port connector that brought out all of the data, address, clock, and interrupt lines (an old Centronics printer port would likely work nicely for such). Had they done this then you could expand upon the Gigatron through expansion projects that could have added video generation, I/O ports (incl. keyboard), external storage, etc. and build it to be whatever you wanted it to become, including the opportunity to design custom projects. Even better if they buffered that expansion port to protect the Gigatron itself so that design mistakes would at most cost the owner a new buffer chip to repair any serious mistakes.
I also agree that it would have been very nice to have LEDs on the various I/O lines along with a variable clock so that you could slow the clock to actually see data moving around on the various bus lines and register states...though a variable clock would break VGA video generation as implemented unless an external video generator expansion with its own dedicated clock were available and used instead.
Steve Pocock it was the ZX80 that had FAST and SLOW modes, it was one of the bigger difference between the 80 and the 81 (especially in terms of running 80 basic on the 81)
Though of course you could turn your 80 in an 81 for the most part by dropping an 81 ROM into it.
Not that expensive for what it is, but I wish there where instructions to do custom ROM (maybe there are, I didn't actually check)
Done in Kicad . . .
It says TTL, yet all of the chips are HC, which are CMOS.
74HCT logic actually, compatible with 74LS logic, and using TTL logic levels.
Oh yeah I'm well familiar with that from experience. I know TTL is a term that's basically synonymous with any 74/4000 series logic, even if they are CMOS chips. It's just a bit of a confusing term because personally when I got started in electronics I assumed all TTL chips used BJTs. It blew my mind when I found out the HC/HCT series among others were CMOS versions. But that's why it's so important to read the datasheet, I suppose.
@1:45 Why are DIP IC pins not bent at right angle to begin with?
Xy Zi For automatic insertion. So the wedge themselves in the machine head and don't fall out during the transfer from shipping tube to PCB.
Instruction set reminds me a bit of BrainF-ck, except it’s a bit more verbose than BF. Only hardcore engineers use BF! Google it and then tear your eyes out learning what exactly constitutes a Turing-complete ISA...
flomojo2u I remember hear about this years ago. Isn't something like only 6 characters you can use?
Mitch Coté Yes, there were 8 opcodes and each opcode was a single character, making reading BF “programs” (I prefer the term abominations, ha) absolutely miserable. Yet people still went on to write various things in it, the site esoteric.sange.fi/brainfuck has a ton of compilers, examples, and history on it. I don’t think too many people actually use BF, more of an “Imza e7331 hax0r” (And oh how spellcheck fought me, trying to write that expression out) type of thing, though it was obviously popular just because it was so obscure and difficult.
the instruction-set looks more like a microcode-set than an opcode-set.
Somebody needs to port the hardware for this over to a cheap FPGA and stuff it into a graphing calculator :P
WOULD MAKE FOR A NICE FPGA STARTER AT LIGHT SPEED!!!
Is it Open Hardware?
Just for trivia: Walter is on of the fastest Dutch lock picker and lever lock picker.
I expected it to be like 300€ or something, it's actually only 150€. Thought case + parts are mutch more expensive.
Ya missed one Dave!!