Page 2 of 3
Posted: Sat Aug 05, 2006 10:22 pm
yes, I recall and it has been written that the 4004 was the first widely used "LSI" chip microprocessor. Fellow named Faggin did the original work on it while at Fairchild, but Intel later hired him.
The company for which I worked in the 70's built process control computers that were all TTL chips, usually wire-wrapped boards (automatic machines did the wire-wrapping).
One of these process control computers ran the on-air switching for a major TV network for many years.
later, we "graduated" to microcoded CPUs, these being, essentially, wide ROMs, like 60 bits, where some bits caused the data paths to be steered and functions like ADD done, and other bits were the next-ROM-address (a state machine).
Because as a young man I spent so many hours debugging computer designs - like errors in the multiply and divide logic, it was many years before I, writing programs, would attribute a subtle bug to me rather than the computer hardware!
Posted: Sun Aug 06, 2006 6:08 am
OK - this is my version - I have no way of knowing whether it is less or more accurate than the others - but it is more complete.
About 1960/70 three PhDs' Gordon, Moore & Noyce left Fairchild or to go work for Brattain ? They couldn't stand that so they set up their own Co. (Intel) specifically and only to make solid state memory - hitherto Ferrite Core - then solid state and the worlds first solid state memory (The 1101 - 1,000 BITS) - before they could start making money from SS memory they took on sub-contract work - One of the jobs they took on was to build a chip-set for a hand-held calculator. I forget the name of the guy in charge of the project/job Let's call him B4004. - but I don't think it was Faggart/Faggin ? - After a short while considering doing the design in the then conventional way- all solid state logic to cover every possible keyboard operation, B4004 (B4) expressed the opinion that this was not the way to do this i.e. with expanded chip designs iy was possible to build the logic and then write a program (micro-program in the first instance) with the instructions. Busicom at that time had a (Japanese)_ representative in the Intel establishment who (Later joined Intel) argued against this because (a) it was new and (b) If it did'nt work his job would be on the line. B4 won out hence the 4004. Then people at Intel began to think of the infinite possibilities of the 4004 with re-writeable, customised programs. The problem was - they did'nt own the design. Busicom did ! This required an approach to Busicom and some careful negotiation - like ' Hi you guys - do you mind if we use this design for something else ?? It might seem hard to tell who won because rumour had it that Intel had to pay more to get it back than they got for making it ? Who won out in the end ? Busicom went out of business and Intel became the biggest/greatest semiconductor manufacturer in the world. They quickly realised that since, among other reasons, the electro-mechanical ? ASR 33 Teletype, which was at the time about the only conveniently available I/O device, was 8 bits (7 bits plus parity) the 4004 quickly became the 8008 - that didn't last long. probably because it was +/- 12 volts (and 5 volts /) hence the 8080 which was (I think) 5 volts - the rest is history - including the fact that the worlds' first micro-processor was at one time owned by the Japanese.
That's my version guys E & O E
Posted: Sun Aug 06, 2006 6:25 am
Ok, i'll try to get a basic schematic posted on the web.
I was also going to suggest that you simulate it first so you
can get an idea what you are up against
There is some logic simulation software somewhere but darn
if i didnt loose the address...perhaps someone else knows here?
I'll also ask around.
Posted: Sun Aug 06, 2006 7:20 am
Yes, the 8008 might have been built on TTL logic but it was a single "integrated" circuit and what I think you may be thinking about is how to build one from discrete logic, i.e. a number of simple chips with basic gates, clocks, latches and ROM.
Actually, the 8008 CPU in the DATAPOINT 2200 CTC wasn't a single integrated circuit. It actually was built out of 100's of TTL chips
. Back before the DATAPOINT 2200 CTC was released, Intel and TI were contracted to build the CPU into a integrated circuit. TI failed and wasn't able to create one and Intel didn't finish in time for Datapoint's deadline, so Datapoint (not manufacturing integrated circuits) instead used 100s of TTL chips instead of the microprocessor.
For my project, I'll just start out with a simple 4 bit computer implementing the 74LS181 as the ALU.
Posted: Sun Aug 06, 2006 8:07 am
Great post, Will, on the story behind the 4004/8008. As I write this I am holding in my hand a weathered and worn MCS-8 User's Manual published in 1973 (2nd edition) for Intel's 8008 CPU and chip set including the Intellec 8, development system. It was a wonderful guide, especially the section on modifying an ASR-33 Teletype, but I opted for the Motorola 6800 eval board, blew out the first 6800 and replaced it for a mere $90.00 (just the chip). I'll never forget the thrill when I hit the reset button and the refurbed TT clanked out the monitor prompt.
Now I've gotta go rummage around and see if I can find that ancient Motorola kit user's guide. Hmmm, maybe it was filched by a "friend" in our early computer club who also never returned my issue number one of "Byte".
Posted: Sun Aug 06, 2006 2:55 pm
I still have a SWTPCo 6800 "PC" with the "SS-50 bus". Later upgraded CPU to 6809. Same era as the MITS.
Posted: Sun Aug 06, 2006 6:44 pm
Here's a schematic of your basic 8 bit computer. Most of what
is shown is the CPU.
Posted: Sun Aug 06, 2006 10:49 pm
we can assume that the goal architecture is not Harvard (separate address spaces for data and program)
Posted: Mon Aug 07, 2006 1:23 am
Power-On-Reset (POR) can simply clear the counter...why not?
Also, with only one instruction you can program to do anything, guess which one.
COPY address1 TO address2
No need to waste a byte on the instruction code itself because it would always be the same.
You just need a few "magic" addresses, at least the counter, a meek ALU,
and your "terminal". (switches and lights, or keyboard and display?)
A TTL 74154 would manage up to 16 magic addresses.
And really clever programming skills too.
Anyone out there have Altair Basic?
Posted: Mon Aug 07, 2006 5:25 am
Back in the 1970s/early 80s just after the 8080 had been developed, a friend was working at Texas Instruments designing the guidance system for Harpoon missles. He so much wanted to use the ÂµP, but it wasn't MILSPECed, so he had to stick with TTL. He designed the entire guidance computer around the 74181 ALU.
Posted: Mon Aug 07, 2006 7:40 pm
Thanks MrAl for the schematic. Now I have to look for that digital logic simulation software...
Apollo Guidance Computer
Posted: Mon Sep 04, 2006 12:51 am
Here's a link that may be of interest to a lot here. It's a project in which someone built a complete working Apollo Guidance Computer (block I) entirely using TTL chips. This is the computer used on the moon rockets. (the original computer was made up of about 4000 NOR gates)
The site includes pictures, schematics, firmware listings, etc. Very interesting read!
Posted: Mon Sep 04, 2006 10:26 am
That Apollo Guidance computer looks cool. Took the guy however four years to build. I just finished building, I guess it's a prototype , my 4 bit computer. It uses 74181 for ALU, 74367 for I/O, 74175 for registers and is clocked using a 555 timer.
Posted: Mon Oct 02, 2006 9:21 pm
Sorry to bring back the subject, but stumbled across this:
and remembered this thread.
A Aussie Computer from the 70's EDUC-8
Posted: Thu Oct 05, 2006 7:50 am
I remember building this many years ago!