i've been a dedicated computer nerd for about twelve years now. i started with a cast-off tandy "laptop" that ran BASIC and have been hacking ever since, through 386/486/pentium/etc.
now i find myself much more interested in computer systems that appear historically before my introduction to computers, and i'd really like to get to know these older computing methodologies more intimately. i'd like to try to build a computer from scratch - build my own processor, etc. - to gain a greater familiarity with the underlying technology.
so where do i start? poking around on the internet, all i can seem to find is vendors trying to sell "homebrew" computer parts which basically involves piecing together readymade components.
i know that the definition of "computer" covers a pretty wide continuum right now, but what i'm interested in building is just the basic machine: an electronic device that runs programs, whether it has a display, printer, or just an array of LEDs as its output.
maybe someone knows a book or something that covers this material.
I doubt you're talking something so basic as the once offered (perhaps still is) Radio Shack computer for kids. It had a bunch of LEDs that flashed with logic signals of sorts. Even further advanced but on same level as LEDs flashing or whatever - was the SWTP computer. SWTP - South West Technical Products. A box about the size of a bread box or so - with LEDs on the front (or an LCD readout - can't quite recall which) and a bunch of switches on the outside. In a subsequent model I believe they offered a keyboard or pad. I'm now 48 and this was all when I was 16 or so, so memory is faint there. That "would" take you back to the very basics. From there came the Tandy computer which I believe you've had already, as well as the Vic and Commodore models and the TI-99 and then Apples. Someone with a better memory than mine can elaborate. The Vic and TI-99 were pretty basic if memory serves me correct. Heathkit/Zenith also offered such a computer set up as I believe you're looking for. Maybe you can find one on E-Bay. National Radio Institute (NRI) did a computer something like this for their course in computers. I seen one recently - unbuilt with books on E-Bay. Not sure what it sold for. Lafayette Radio offered a similar version of the Radio Shack deal.
What an easy life you've had! In my day, I had to build it up solder joint by solder joint. 256 bytes of static RAM was all I could afford on the board. Metal bat handle switches developed calluses on your fingers. ASCII was entered with 7 switches and a push button -- those keyboards were for the silver-spoon, fancy-pants rich boys. ;) Used an AM radio nearby to help debug the programs by listening to the tones.
Hehe.
Well, you are in great luck, to be honest. There are good books today as well as cheap FPGA boards and free software you can use to lay out your own cpu. You almost won't even need a soldering iron.
Try "Bebop BYTES Back: An Unconventional Guide to Computers" from Doone Publications as a very good and very detailed introduction. For an introduction to VHDL and Verilog, try "HDL Chip Design" by Douglas Smith (who, last I heard, had moved from England to Alabama.) I'd recommend looking on the web for FPGA boards -- there is BurchEd and Xilinx and a variety of others making such boards. Some of them are in the $50 to $100 range and very, very good. Make sure you get software with them for writing VHDL or Verilog and where you can do floorplanning, later on (not now), and the like.
I'd recommend starting out with a simple board that includes some switches and some 7-segment LED displays. Try writing some VHDL code to perform various simple functions, like binary addition. That will get you started on some of the basic features of an ALU (and you won't need to understand sequential logic in VHDL, yet.) Work on busing data around, addressing and registers, etc. You can use the internal RAM in the FPGA for your program and data memory.
Just get doing it. It's not hard to do something good enough to get the basic ideas across and learn from. It is so much easier to do, today!
___ Jake- If you can stand starting at the microprocessor level, try getting your hands on "Build Your Own Z80 Computer" by Steve Ciarcia, BYTE Books, McGraw-Hill, circa 1981, ISBN 0-07-010962-1.
Sounds like it might be just what your 'e looking for. Cheers! Charlie
___ "Sic hoc legere scis nimium eruditionis habes." (If you can read this, you're overeducated.)
My first build-up was the Altair 8800 from a kit of parts. Years before the Z80 was available. I remember looking with envy at the
8085 when it finally came out, because it had been so simplified from the complex clocking monster that the 8080 was. And that was before the Z80. The Z80 was so simple to design a board for. Nice!
Well, I've some 65SC02's laying around. And I'm sure I could find an
8080A and a clocking chip for it, if I looked. ;)
The only problem I have with your answer to the OP is that the OP specifically asked this:
It was the 'build my own processor' part of the above phrase that really sounded to me like asking about how to design an ALU with latched input paths and output, access to registers for latching, a memory address latch, tri-state buffers on a common internal bus, etc. I could have been wrong, though.
But that is a great deal of fun to learn about and do -- especially in this day of cheap FPGA boards from a variety of sources in attractive variations on a theme. It's almost dead simple to get started now writing VHDL and getting a simple cpu up and going, entirely of your own making. You don't even have to worry much about routing and floor planning, as the tools will do a uniformly lousy job automatically but one that still gets something working for you. And with the huge resources available on these chips today, who cares if it all gets ruthlessly squandered by the planner while you are learning VHDL?
It will be interesting to see if the OP is really more about learning the internal basics of how a cpu works inside or more about learning the larger picture of a cpu-memory-i/o system. There are so many choices now that can teach at almost any level of interest -- from internal cpu design all the way up to pasting down one-chip-wonders that include all the I/O, code space, data space, and cpu processing power and leave very little left to learn about.
schreef in bericht news: snipped-for-privacy@f14g2000cwb.googlegroups.com...
Well, Jake,
A starting point is a matter of choice. Some early calculating machines used cogwheels and functioned pure mechanicaly. The first electronic computers used electron tubes by the dozens and required more power then your mains connection can provide. The first computer I worked on was build with discrete transistors, ferrite cores and lots of wire all packed in five 19" rack enclosures higher then a mans length. The first one I build for myself has a Z80 processor on 4MHz, 2k of RAM and 2k of EPROM. The latter contained a monitor program derived from the NASCOM. I build, also from scratch, a separate I/O card for it containing a UART that communicated with a dumb terminal. The next step was an I/O card that could write to - and read from cassette tape. Still works when I hook up a PC running a terminal emulator. At about the same time you could buy Apples or one of its clones. You could buy an empty board and fill it with components. Some time later you could buy empty PC- and peripheral boards to do the same. AFAIK the last computer building that required soldering. These days you can assemble your own machine even without a screwdriver. IMHO you can't go back but to the first microprocessors like the 8085, Z80, 6800, 6502 and some others I don't know well. Some stuff, like the Z80, is still available. Don't know about the others. Nevertheless, I don't think this is the way to go. The old times will not come back you know. I advise to look around in the world of microcontrollers. They have processor, RAM, ROM and I/O in one package but fiddling with the bits, assembler programming and even soldering are still required. There is a wide range of them from six pins SOT-23 to forty and more pins DIP all with eight bits processors. The latter at least as powerfull as the old Z80 and its contemporaries. If you want more there are much more powerfull sixteen bits micros available as well. You'll find more info then you ever can read on the web but you still can do plenty of things others have not done before.
Not "really" being wrapped up in computers back when the Z80, the Altair and so on came out, I had forgotten about the KIM and so on. My question is - trying to sort out my memory here - was the Altair 8800 sold be SWTP? OR a different monster altogether? I DO recall the Altair now that it has been mentioned - but may be getting it confused with South West Technical Products - products.
I've never looked closely at the book, but my impression was that it was to build a working system, circa whatever year it came out.
No matter how fancy a system was back then, it's going to be nothing compared to what you can find in the garbage today. Thus I think there's little reason to build a computer to that extent.
I think there's still plenty of reason to have something on the level of the KIM-1, ie a calculator style keyboard and readout, a good monitor that can single step etc, a cassette interface to save programs, and some sort of general purpose I/O so you can play with things. There were plenty of such single board computers back then, I use the KIM-1 as an example because I had one as my first computer.
But the single stepping meant you could run programs and see what happened at every step. You could even just put in a line of code, and check that out, really useful to getting a real feel for what the op-codes were supposed to do. Since there was so little in there, you didn't need to learn a whole lot of GUI stuff before you tried out your simple program to add some numbers. The monitor did have what you needed.
The general I/O meant that you could play with real things, like hook some LEDs onto it and learn how to control them. Or have inputs from something, to control the program.
These computers are of such a simple level that they are easy to build, and you will get ample use of it (as opposed to trying to build a full blown S-100 bus computer from 1976, which would have real limitations today if you wanted to run applications). In some ways, it's even easier, because whereas the KIM-1 needed 8 ICs to get 1K of memory, you can scrounge up a static RAM that fills the address space that will draw less current and require much less wiring. If you can live with hooking it up to a terminal (ie your home computer running a terminal emulator), then you can toss the readout and keyboard. Many of those old boards allowed for hooking up a terminal to an RS-232 port, though many of us didn't since the terminal cost more than the computer. Likewise, one could use the home computer to store the programs, which beats out a cassette interface in terms of speed and reliability.
ONe great project from the era was in Byte, though I can't remember the year, or even a general idea. 1978 and 1980 somehow come to mind. Someone wrote an article about bootstrapping an 8085. Jam a NOP onto the data bus, so the processor advances the address bus while doing nothing. By single stepping, this means you can load the RAM without any bootstrap ROM. The input mechanism was a piece of circuit board that he'd carved up with a hacksaw to make pads, and a "stylus" to connect a wire to the needed pad. Some LEDs on the data bus. Not much else. Build up the simple hardware, he just used wire to do it if I recall, and you can start playing right away. None of that fussing with a bootstrap ROM, but something you can infinitely play with it. Get some coding in, and then add that monitor ROM. Or dig up the listing from an old single board computer, and use that.
SOme years later, Byte put out a book about the 68000, and it had an article (which I think had never run in Byte because I never saw it elsewhere) using similar techniques to bootstrap the 68000.
Tha Altair came first, being on the cover of Popular Electronics for January 1975.
That's what set off the whole thing.
Of course, there was the Mark-8, an 8008-based computer that was on the cover of Radio Electronics for August 1974, but while there was interest in that, it somehow did not get the excitement and the attention that the Altair did.
Likewise, Scelbi had an 8008-based computer before the Altair, I can't remember exactly when, but at least for me, and likely others, it wasn't noticed till after the Altair hit. I know the Scelbi was advertised in the back of QST, but I only saw the ad when I looked for it after the Altair hit the market and the ad was mentioned.
The SWTP 6800 came a bit later, I can't remember exactly when, but of course it had the advantage that it was terminal based. No fiddling with switches and strings of LEDs on the front panel, and surely their absence made it simpler to build.
The Altair 8800 was 'advertised' on two issues of Popular Electronics circa the beginning of 1975. You can see it here:
formatting link
I bought mine in January 1975 and assembled it over the next month or so. It was sold my MITS, which was a company that used to make calculator kits, back in the days when calculators were expensive and selling kits actually made some money. They also had contracts with NASA, I think. But after the huge budget cuts of 1970, after we got to the moon, and the gasoline crisis, all hell broke loose and companies like MITS went into red ink and started looking for something else to do. Luckily, Intel had just had Intersil drop a contract (but largely paid for, already) for a cpu for a terminal, for which they were making the 8008, if I recall. Intel was rolling that into the 8080 to sell as a micro just about the time MITS was starving and looking for work. The Altair 8800 was the result.
But I'm open to corrections on the above. That's just what my poor memory reminds me of.
SWTP was, at least from my point of view, a later-on company. I hadn't heard of them until well after the Altair 8800 assembly I did.
They were a small company that started by selling some rocketry related electronic kits, and used "MITS" to suggest MIT/Massachusets Institute of Technology.
They went the usual route, selling kits and counting on articles in the hobby electronic magazines to run articles (which they wrote) as promotion.
When calculators were feasible, they started making them as kits. The November 1971 issue of Popular Electronics has their calculator, I think it was their first, on the cover. They were a new thing at the time, and even though expensive they were still a good price if one needed such a thing, compared to the cost of what came before.
But as things evolved, the big companies came into the market, just as with digital watches (I can't recall if MITS made those) where small companies were in early and then lost to big companies. MITS couldn't compete with TI and the other big companies in the calculator business after they came into the field, and then nearly went bankrupt as a result. Legend says the Altair was a last minute attempt to save the company.
It wasn't Intersil, it was Busicom that wanted a calculator IC. Intel convinced them a general purpose CPU would be the best solution, and the result was the 4004. The 8008 was developed for a terminal manufacturer, but not Intersil, who of course manufactured seminconductors, mostly analog. By the time they came out with the 8080, they weren't designing it for anyone.
And as I've pointed out, others had already come out with "home computers" before the Altair, and while we hobbyists likely hadn't hard of the 8080 before the Altair hit the cover of Popular Electronics, I think they already had a level of familiarity and usage in professional circles.
But one thing about the Altair is that it needed so much to get going. The front panel (which wasn't just the switches and the LEDs, but I gather rather extensive TTL), and the CPU board had no memory so you needed an extra board for that, and of course then you needed the motherboard so you could connect the boards.
That bootstrapping project leaves it relatively simple. Even something like the SWTP 6800 CPU board could be standalone I think, because it had memory however tiny and the monitor in ROM and a software UART.
One reason the Cosmac Elf was so popular, relative speaking, was that it didn't take many parts to wire up, and you could have it running as soon as it was wired. It was made for bootstrapping without ROM. The problem there is that 1802's were never that common back then, and scrounging one up will likely require much effort at this point.
SWTP was already in the kit business before kit computers hit the market. They sold kits of parts for a lot of Popular Electronics articles from the late '60s, on.
My memory is definitely fading! Thanks! I seem to recall reading some 'history' on MITS circa the late 1970's that included some comment along those lines. But I appreciate the correction and I'll try and kill that 'recollection' of mine!
Ah! Perhaps that was what my memory distorted over the years...
I remember seeing some tiny partial-column ads from them and thinking about buying one. But it was too much for me at the time. I just kept using a slide rule.
I've heard that, too.
No, I think you are thinking earlier than I was thinking. Isn't Busicom a Japanese company that started the whole ball rolling with Intel?
Anyway, I was thinking actually later than the late 1960's. And it was definitely something to replace a lot of discrete logic in the body of a glass terminal. Not a calculator.
Fully agreed, here.
Who was it, then? Do you know?
Yes, that's what I have heard as well. Something along the lines of 'productizing' what they had left off with on the terminal project.
Since I was just 19 at the time, I wasn't in those circles.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.