Ask Hackaday: Who Likes Retrocomputing?

Last week we posted a link to Project Kiwi, a homebrew Motorola 68008-based microcomputer built by [Simon] that includes Ethernet, a very good display adapter, an interface for IDE hard disks, two Commodore SID chips (for stereo chiptunes), a floppy disk controller, and an already existent software library that will make it very easy to develop your own software for this wonderful computer.

After thinking about [Simon]’s Project Kiwi for a while, I’ve been thinking there really hasn’t been a homebrew computer made that is so perfect for a proper Open Hardware release. There are more than enough peripherals in the computer to make development very fun. I’ve suggested doing a group buy to get Kiwi PCBs out into the wild and into the hands of other retrocomputer fanatics, but [Simon] would like a little more feedback.

Of course, this means turning to you, the wonderful Hackaday reader. Would any of you be interested in your own Kiwi microcomputer?

[Simon] tells me there are a lot of problems for turning the Kiwi microcomputer into a Open Hardware project. His prototype PCB cost €300, greatly reducing the number of people who would be interested in making their own Kiwi. Also, there are a few problems on the current PCB design (easily fixed for the next revision), and [Simon] would like to add a few features like DMA and a proper framebuffer.

Despite all those problems, I can’t see a better way to learn about computer architecture the hard way (i.e. 80’s microcomputers as opposed to futzing around with a Raspberry Pi). You’ll also get a really wonderful computer system that will show the power of 80s-era electronics, with the very hopeful goal of spreading the gospel of retrocomputing with the venerable Saint MC68000.

If you’d like to add your two cents – if having an Open Hardware 80s microcomputer is a good idea, or some technical requests such as adding a proper 68000 CPU to future designs, leave a note in the comments or on the forum [Simon] set up on his Kiwi page.

I think it’s a cool idea, but then again I’m probably blinded by how cool an 80s computer of this caliber is. The fate of this project is now in your hands.

54 thoughts on “Ask Hackaday: Who Likes Retrocomputing?

  1. I think it would be awesome to have my own kiwi computer to fiddle around with. If it was somehow possible to source all the “obscure” parts he managed to get(the Yamaha V9990 comes to mind)… I think this would be the hardest part in making these in mass. A kit you solder together yourself would be awesome, a hark back to when “real” computer people built their own computers!

    1. I would love to see this but I agree. You could replace the SID chips with a propeller emulating the SID. Heck you could probably get a SID per COG out of it.
      You could probably replace all of it with an FPGA and while some people say where is the fun with that I really like the idea of someone making a really Open computer. Even some of the Intel graphics chips have limited function with open drivers. Maybe even play with different ISAs. I know that there are free MIPS cores but how about a SBC that uses an IBM 360 ISA? Or even a Cray ISA?

  2. N8VEM is probably the most established homebrew computer project. The base board uses a Z80 and is CP/M compatible.

    I built Ultim809, but the materials cost is still quite high, and I never got around to writing a decent OS and filesystem support.

    The biggest issue with homebrew retro computers is the scarcity of peripheral chips. The “purest” ones uses vintage sound and video chips from the ’80s, but they can only be found on eBay, and are not cheap. The V9990 used in the Kiwi costs $48 by itself!

    Of course, you can always pair a vintage CPU with an FPGA or microcontroller, but it ruins a bit of the retro magic…

  3. This would certainly be awesome! But I have serious doubt about availability of parts. Is the 68 maybe made in some form still? The only place I can really get them from is ancient HP Printers, which I am fortunate enough to have a few of plus one 68HC000. But I don’t think others are this fortunate. So what other chip could one base such a project on in a modern world? A few Microcontrollers come to mind, but they’re all too Integrated again.

    1. Getting a 68000 is no problem at all. There’s a bunch for sale in the Jameco catalog in front of me. You could always tear apart a Sega Genesis, if it comes down to that.

      Getting the Yamaha display adapter might be a problem, but we’re only talking about sourcing a hundred or so. Not tens of thousands.

      1. Jameco has them? :D I didn’t know that…
        So it’s only the Yamaha chip thats the problem, and slightly the pricing of it all. But then again this is for Diehard Computing Nerds only.
        Now to what extent could the Yamanha be supplanted by something else, there has to be some other chipo/alternative wiring?
        Also I’m still wondering, with what modern sort of hardware one might best be able to recreate the experience these old systems give?

    2. Someone here was telling me that they think some of 68k series are still in production because they aren’t marked as being out of production on freescales site.. I think that all of them are out of production now and there is a stockpile of parts for long term support and they are expensive.

      Having said that: You can get NOS 68k chips pretty easily. I see trays of 68k series chips going on various auction sites most weeks and I’ve ordered 68SEC000s without problems a few times. Getting enough chips to complete a small run of boards wouldn’t be hard.

      1. @adsf

        You’ve said that before.. but I really doubt they’re shifting the numbers of parts to actually keep them in production.. and all the date codes I’ve seen would suggest that none have been fabbed for sometime…

      2. Dug around some more, and the official life cycle status is “NOT RECOMMENDED(DECLINING) – Capacity beginning to decline. – Orders and shipments permitted. Product not recommended for new designs.”

        There is still one stage before production is stopped, and for a product as important as the 68000 Freescale would give notice well in advance. Obviously production volumes are low at this stage, and you may have to be a special customer for them to crank a few wafers through production, but they’re still making them.

      3. @asdf

        I guess now is the time to stock up on any m68ks you want then. The 68SEC000 is a really nice chip IMHO. Because it’s 3.3v you can strap it to a cheap FPGA dev kit and have a customisable m68k machine in no time at all.. You can put TG68 on an FPGA but it’s a bit buggy and means you need a massive FPGA to get anything done… They aren’t too expensive either if you haggle for them.

      4. At my current job they are still producing replacement boards with 68302 chips on them. So I’m sure you can still buy that chip. (but that might change in a year or two, as our demand is quickly reducing)

      5. @daid303

        Could you tell us some date codes? There is a “Product Change Notification” about the 68302 moving fab to Malaysia in 2003 so they were still cranking them out in the 2000’s.. Some other 68K MCU style products are also listed on the PCN.

        For the 68HC000 the last was 1995 by the look of it, for the 68EC000 and 68SEC000 the last was 1999 about a mask change and moving to the Tohoku fab in Sendai which seems was damaged beyond repair in last years earthquake/tsunami.. so I reckon classic 68000s are out of production unless they moved fabs without issuing any notifications.

  4. His accomplishment is quite remarkable, I want to be clear about that before what is coming…

    If I wanted to develop a computer from scratch, then buying a premade board is hardly a the way to go. If I want to work with basic old tech, then there are tons of options that cost less all together (including an LCD) than the bare board for a Kiwi.

    Again, his build is very nice…but to me it just seems like something he did to develop his skills and show off a little; not to sell…not when I can get an ARM development board that blows it out of the water for $5.

    1. I hate to say it, as I love the 68k from my days with my Atari ST, bu you’re spot on.
      I would much rather work on modern cheap arguably more capable h/w. There’s no shame working with ARM devices, esp. for me as its a British invention! ;-) licensed around the world and in millions (billions?) of devices.

  5. A homebrew computer would be great, but I don’t like using harvested parts. Not only is availability uncertain (an old machine popular in some part of the world may be completely unavailable elsewhere), but I also think old hardware should be preserved. It may not be “authentic” retro, but I’d much prefer using an FPGA for video/peripherals.

  6. To me, the charm of 80’s computers was all about programming close to the hardware, rather than on top of some complicated, bloated multi-gigabyte OS. And there’s a certain joy derived from writing quality code, which the inherent limitations of the machines often required you to do; rather than just banging out anything that worked, or calling someone else’s library.

    I still get the same thrill as I did from my first computer (an Atari 800) every time I work with an MCU. But I do not live in the past. I have no desire to work with a real or emulated Atari 800.

    If I want to work with a computer, with a screen, keyboard and so on, I work with a modern one. If I want to work low-level, I work with an MCU. The middle ground of a retrocomputer has no emotional or practical appeal for me.

    I’d consider designing my own retrocomputer, as Simon did, a fun challenge. But I’d soon lose interest once it’s done (or nearly so). And not much fun in buying or building someone else’s design.

  7. You build one of these computers not because of cost, compatibility, or usability, but because you want to learn something.

    If you want something functional, you build it based on a microcontroller or small computer – something that has some compatibility with the world so you don’t have to develop *fracking*everything* from scratch.

    If you’re going to build something, be sure that it can also be useful.

    – – –

    I cut my teeth on an S-100 computer based on an 8085 with 32k static RAM,, then later a Z80 with 64k static RAM. It ran CPM at 2MHz, drove two 8″ floppies, and a monochrome 40 col x 16 line video display.

    It was fun, and interesting, but it was never really useful. After I added a serial terminal board to the keyboard and monitor, I used the resulting 80 col x 24 line monochrome terminal with a 110 baud modem to get online with Compuserve and Bulletin Board Systems (BBSs were much more satisfying). The original computer hasn’t been used since….

    The 110 baud modem was eventually replaced by 300 and 1200 baud modems over time. Somewhere in there, I got a somewhat compatible PC clone, then a turbo PC (4.77/8.00MHz) and a 20MB hard drive.

    All before (about) 1990.

    I gotta get another hobby….

    1. Nothing stops you from going to ebay and getting a C64 or Amiga or some Atari. They are easy to come by, they have state of the art floppy emulators developed for them and their video output is still compatible with TV sets, TV tuners and some of the more advanced monitors with composite/S-Video inputs.

  8. I think retrocomputing needs parts that will be around for a while but I think that the hardware also needs to be sped up but if you can’t get on youtube, less people will be interested.

    Users want computers that are fun and easy to program and sometimes chips with 2500 pages written for only engineers don’t appeal to everybody that doesn’t have an engineering degree. Users want to program and not have to wrestle with large manuals because they would rather spend their time programming.

  9. As much as I like the m68k.. and I like it a lot, it’s nice having a chip that is both simple enough to not have to continually refer back to the reference manuals and has GCC/gdb/newlib support.. If you want to learn how computers really work any simple micro with an external SRAM style bus will be as good as the m68k.

    I personally like this board:

    The H8/3069 has all the stuff you want.. classic CISC architecture, DMA, UARTS, DRAM controller and there’s a really nice rom monitor that allows you do debugging on the chip itself, assemble into memory and it has a gdb stub for remote debugging. The board above is relatively cheap and you could interface an old skool display processor in about five minutes.

    1. Renesas are very bad with providing info about debugging interfaces and such, is that also the case with the H8/300? Renesas also still have a few romless low-end SuperH CPUs that could be contenders.

      1. For newer parts like the RX etc.. if you don’t want to use Renesas’ debugging stuff (expensive as hell) yeah avoid them.. but with the H8 (not the H8S) etc there isn’t really much internal debugging hardware. Renesas(I guess hitachi because of the age of these things) supplied the source for ROM monitors for a lot of the H8 and early SuperH parts. For the H8/3096F this monitor is great .. the manual is all in Japanese but the interface itself is English and very easy to get on with.

  10. I grew up in the days of 8 and 16 bit machines. We built systems with wire-wrap and proto boards. Everything from Apple II+ clones to Steve Ciarcia\’s designs, from Z80 to NS32532. It was a awesome time, but today there is way more opportunity, hardware, software and more fun than just retro.

    Microcontollers can be simple or as complex as you want them. You can build and code a machine in assembly or you can load Linux and Python. FPGAs, DSPs and LCDs are affordable, SDR is doable with consumer hardware. You can emulate, virtualize or syth hardware. You can design and order quality PCBs with free software and little cost.

    Now is the Golden Era, pick a project or niche and hack away!

    1. “SDR is doable…” Someone is pushing my buttons.
      Implementing SDR to many people equates to having an audio ADC input, and driving a kit like the SoftRock. And relying on a PC for the heavy duty math.

      You want to make a nice kit? Try making something nobody else makes (nobody cheap, anyway). Find yourself an ADC or two, 14-bit or more, sample rate of 80Msps or higher. Then grab an FPGA or whatever to process input, LCD to monitor it, and some more IO to dump it out.

      Final touch: It need to be self-contained ie. the compile / debug does NOT require a PC, just a keyboard.

      1. SDR is doable, depending on what type of bandwidth you want and how much you want to spend. Take a look at QSR1 as a reference, it uses the computer as a controller but really doesn’t need it to do much of the work in some use cases. There’s a few other projects floating around the net, but no low cost killer project has grabbed this space… yet.

  11. I much prefer using an actual 80’s computer, though I think the 68K project is really awesome.

    Being a Propeller fan, I often use the Prop in the way I would old computers. It’s a little RAM constrained. The next generation of it won’t be for retrocomputing purposes though.

    One thing that would really get me thinking about the 68K computer would be whether or not it could potentially be used to develop for existing micros. The RasPI can do this, and it’s on my short list.

    Otherwise, I think a self-hosting existing micro-controller type computer would be “retro” in feel and applicability. Very intriguing idea all around. Reading with interest to see where it goes.

    1. >It’s a little RAM constrained.

      There are m68k machines out there with hundreds of megs of ram.. I had one with 128MB for quite some time. The beauty of 68000 is that they made it look like a 32bit machine from the start (The 68012 can address 2GB of memory apparently, never seen one in real life though). If you’re going to compare the m68k to anything it would be other 32bit machines like ARM.

      >develop for existing micros.

      Machines with an MMU can run Linux and GCC.. but you would be insane to use such a machine to do any amount of development.

      If you wanted to do m68k development on the raspberry pi .. I have a simulator that runs on the beaglebone, so it would probably run on the pi too :)

  12. The 68k isn’t retro enough. You have to explain why you’d build one of these instead of putting external ram and peripherals on a flexis coldfire chip… “Everything is in a DIP” isn’t that good an excuse. (and… I’m so done with floppies…)

  13. If the design files still exist for the Commodore SID, that’s a component that might draw enough interest on its own to get produced again.

    Instead of creaky old original chips for video, how about a modernized version?

    Check out the F18A, a drop in replacement for the TMS9918A (NTSC composite out) and TMS9929A (PAL component out) Video Display Processors.

    It’s a TMS99xxA chip in an FPGA with added features such as 80 column mode that’s compatible with the Yamaha 999x VDP series, which were based on the TMS99xxA series. The F18A also includes a true bitmap mode, eliminates the issue with more than 4 sprites on a line and enables all the sprites to be displayed at once, plus more useful sprite features the original lacked.

    I don’t know if the FPGA on the board has room to upgrade it to full Yamaha 999x capability. The 9995 used in the Myrarc Geneve 9640 supported almost all the functions of the TMS9918. There was one video mode only a couple of programs used, which wasn’t in the Yamaha chip.

    The F18A is “plug n play”. Remove the original TI chip from any computer or game console equipped with one, install a socket if needed, plug the F18A in and install a 15 pin VGA plug somewhere.

    The original analog video out will no longer work as the F18A does not have that capability, those pins don’t connect to anything. VGA is VGA anywhere so there’s no need for NTSC or PAL output.

  14. Indeed I did this project to learn about hardware design. I know about FPGA based projects but writing VHDL code is different. For those who just want to program a small, modern platform there are FPGA or microcontroller projects available. Ready to program for. For those who just want to use a ready computer or game console, there are plenty consumer products available.
    Actually Kiwi is different. As it has some hardware issues I would have to address before I would consider the PCB ready for production. Thus every part of the design can be altered. This redesign could be a community project at a lower level.
    For the parts we probably talk about dozens of each part, not thousands. I would assume it is possible to get even obsolete parts in this quantity. And prices will probably drop with a higher quantity of bought chips.
    Everyone interested is invited to discuss all this in the Kiwi forum. Joining forces could be fun and end up with a better design and more software I could do allone.

  15. I like retro computers. But I would be motivated to build a retro like computer myself only with modern hardware. Retro systems built with retro hardware doesn’t make sense because the parts are difficult to get. Recently I bought parts to build my own KIM-1 retro computer, and I had serious problems to get all parts together.

    My modern retro computer would not only be a study object to learn programming on the hardware level on the new mcu but it would be a useful tool for daily work as well. The system would support addons (fpga etc.) to extend it to a sophisticated oscilloscope, logic analyzer, and so on. Such a concept would motivate many people to contribute.

  16. It could be a very very big problem to get enough SID-Chips for all….

    I think this is a etical question, because for one of this hardware, two c64 will die…maybee you should develop your own soundchip or use a fpga with one of the available verilog/vhdl sid replicas…

    1. I think SID has been successfully emulated on the Parallax Propeller, using a single core. If this is true, you can:

      1) Save two C64’s and solve the availability problem.
      2) Keep the sound generators in socketable DIP package.
      3) Emulate both SIDs (or more) with one Propeller.
      4) Probably absorb the function of some other support chips too.
      5) If someone wanted to change the way the sound chip works, and knew neither Propeller or FPGA programming up front, learning to program a Propeller would be the less intimidating option.

      Although I acknowledge having emulated components doesn’t appear to be in the spirit of this project. And if you make an exception for the SIDs, #4 is a slippery slope.

  17. Maybe the answer is to replace the SID chips with something easier to obtain and more widely used.

    Not sure about availability but the Yamaha YM family of sound chips are very well known and widely used (many arcade boards and consoles used the YM sound chips). Combine a YM sound chip with a raw sound chip of some sort to handle sound effects and other things that the YM cant really do and you would have a nice sound system that doesn’t require the sacrifice of perfectly good C64s to do it.

  18. i’d say takings parts from a C64 is ok as long as it’s spare parts form an unworking C64. C64’s are may be anything but rare, but defacto was the best of the 8-bit era. let the still living venerable C64 alone.

  19. The Kiwi project is an excellent homebrew project, but as pointed out it is far to expensive and somewhat over complex for the average DIY homebrew type person.

    Ideally a design would use about 10 ICs.. cheap legacy stuff. I’ve sourced ICs from China and found that I can buy a complete kit of ICs for about $15 or so..

    CPU Z84C0020 20MHz , 128K SRAM, 64K EEPROM, CPLDs, V9958+VRAM, MAX232, SN76489 or SAA1099 or AY-3-891x sound chip .. all for $15 other CPUs are W65C02Px , HD63C09, MC68008FN12.

    It should in theory, once a PCB has been designed to make a complete kit for $50 (hopefully).. The hard part is designing a simple yet powerful system. Probably with a connector to add a Daughter board for extra stuff !

    1. Lez, would you please stop using my project for advertising your commercial (!) project. I asked you when you spammed on youtube. Everywhere I read about your computer you pretend you have “found” a kit.
      I have had to help your customers because you don’t offer a full design. So be fair and use your time to help your customers instead of spamming other non-profit projects. You make money from it…


Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.