newbie building a scratch-built computer
ard at p850ug1.demon.co.uk
Tue Jul 31 15:55:44 CDT 2007
> Thanks for your advice, Tony. To address question #1, I'm honestly not too
> sure. I figure ha wih the early micros, there is documentation, where as
> something completely homebrew, there is little, if any. What I need to find
> is a good resource for electronic circuit design. I have a book called "How
The book I certainly recoemnd is 'The Art of Electronics' by Horrowitz
and Hill. It's expensive, sure. But I doubt you'll 'grow out of it'
quickly. Put it this way, I consider I have some knowledge of electronics
(and I trust eome others here agree with that), but I certainly don't
know everything in that book. I still refer to it quite often.
It starts from the 'what is a resistor' level, and goes on to
microprocesors' It's a very 'practical' book, very little maths.
There is also a 'lab manual' called 'The Student Manual for the Art of
Electroics'. Certainly the second edition  contains a design for
simple 68008-based microcmputer built on those solderless breadboard
things . There's no reason why you couldn't make it on stripboard or
by wire-wrap, or... and/or expand it a little. I forget what's in the
first edition of the book, maybe an 8085 or z80 baed design.
 I like the books so much that I bought both editions of both books...
 Personally I hate those things. Too many unreliable connections, too
much stray capacitance. There's so much you can't do with them, you might
as well solder the stuff up from the start!
> to Build Your Own Working Microcomputer" that I bought for under a dollar at
> a thrift store. I still need to finish it, but it doesn't seem to be much
> help in that respect.
> As for your second question, I would find it to be really fascinating to
> build a computer out of TTL, but again, I need some kind of resource to
> guide me through it. All my knowledge on electronic circuitry is very basic
Thinking about it, I would suggest making youre first homebrew somputer
using a normal microprocessor. You can get some experience of TTL
circuitry my making interfaces for that. When you've got said experience,
you will probalby be ready to have a go at ddesigning a processor.
If/when you get that far, you might want to read some of the technical
manauls and scheamtics for minicomputers, etc. Not to copy such a design
gate-for-gate, but to see how things were done and learn a few tricks. I
remember I learnt a lot from reading the CPU manual for the Philips P850
and then the one for the DEC PDP11/45 (Knowing what I do now, I'd not
recomend starting with either machine, but it's what I had...)
> right now. My father, an electrical engineer, tells be that he couldn't
> even do it after six years of college(although, he graduated 20 years ago),
> so I'm not very hopeful that I could.
All I can asy is that I don't have any qulaifications in computing,
electrical enginering or electronic engineering, and I've designed a
simple processor. It's _possible_
> Third, commercial software would be nice, but I don't need it. I would very
> much like to learn machine code. I figure that when I start college
> interviews in a year going toward something in the technology field, I'd
> like to have as much pre-knowledge of computer science as I can. I figure
I'm not sure how much good it will do you. Certainly the courses over
here (UK) seem to have little, if any, hardware or machine code
programming in them. But perhpas I've just been unlucky...
> that there's no better way than to build and program a computer. It will
> take me a long time, but it would be worth it.
TO be honsest, I doubt you'll get it all done in a year. This doesn't
mean uoi shouldn't start, of course, and to be honest, I would consider
what you learn from this to be a lot more valuable than the end product,
and of course you'll be learning all the time so you will have something
to talk to the interviewer about :-) I know that if _I_ was interviewing
somebody I would be impressed if they were in the process of
designing/building a computer., sicne I would realise all the practical
stuff they must have picked up on the way.
> Finally, I would most definitely use chip sockets. I don't want to risk heat
> damage to components if I can just solder in sockets and install the actual
> ICs later. It may not be as good for reliability, but it is better for ease
> of design and construction
I've never overheated a chip by soldering it...
But the real advantage of using sockets is that you can, say, pull a chip
and force the pins to high and low states. You can pull your processor ,
ROM and RAM chips and check for shorts on the data bus with an ohmmeter.
That sort of thing.
One other point. Use decoupling capacitors. At least 1 0.1uF between
power and ground close to each IC. That may be overkill, but I can assure
you the cost of some disk ceramic capacitors is a lot less than your
time in finding a nasty glitch. I speak from bitter experience here.
More information about the cctalk