Quirky chips

1/30/2012 1:41 PM EST

Interesting article.

Back in days gone by it was a bit more common to see people design their own CPUs. One place I worked for in the 80s built a vector display processor themselves using an EPROM and a PAL or three.

As generic micros have become more capable and FPGAs have become available it is easier to just use generic components.

1/30/2012 2:08 PM EST

How about Intel's iAPX432, the Ada machine?

P.S. the 1802 was my first micro; the instuction set was so simple, you could 'assemble' machine opcodes in your head... :-)

1/30/2012 3:00 PM EST

If I recall correctly the other main difference between the 2900 and 3000 was the 3000 could only implement a 2 address machine but the 2900 could do 3 addresses in one microinstruction. Doesn't seem too important now but remember we were all CISC-oriented back in those days. I also remember from TI the TMS1000 8-bitter which I recall doing some development on (later rechristened TMS370) and the graphics series TMS380, there was a frightful amount of hair-pulling in some circles when the TMS380C20 went EOL just after some vendors had finished polishing their carefully-optimized assembly code implementing the X11 "window" server for Unix (especially for early military graphics apps)...! Also I recall manufacturers writing code to turn the 8X300 into a tape or floppy controller, and several of these WERE successful products, if you can imagine what kind of world it was before very much became available in the way of useful high-level peripheral support chips (late 70's-early 80's).
Things were primitive BUT at least there were lots of job opportunities unlike now!

1/30/2012 3:08 PM EST

Oops my bad - the graphics series was TMS340 not 380, guess I meant to say TMS34020 went EOL, sorry!

1/31/2012 5:14 PM EST

Its impressive to read the history of microcontrollers. I’ve recently been thinking that math, physics and science in general is part of our human culture. Usually we think that a man who has culture or knows culture doesn’t necessarily knows STEM, whoever, I do think that STEM is also a form of culture. Now be reading this article, I get the idea that history of technology can be also considered part of the culture of man-kind. Thanks for sharing this interesting brief of the development of electronics in the US.

2/2/2012 1:15 PM EST

Interesting to read of a processor with no subroutine call instruction. Today we regard this as an almost indispensable part of a usable machine.

I remember being lectured at university by Prof David Wheeler, who worked on the first subroutine library for EDSAC in Cambridge in the 1950s. That had no subroutine instruction either, so he came up with a very clever method by which each subroutine modified itself on every call to return to the right place.

I am told that some places still call subroutine calls "Wheeler Jumps" to this day...

2/2/2012 1:32 PM EST

As a lover of the 1802 I find these comments slightly unfair. It was the best and fastest CMOS processor at the time - all others needed a lot more current - 1802 was the fastest on a limited power budget. We supported one of the first intelligent telephones in Europe - alternative was a TMS1000. For details see the Design Ideas Book BMP802 http://homepage.mac.com/ruske/cosmacelf/Design%20Ideas%20Book%20for%20the%20CDP1802.pdf
I wonder if there are VHDL implementation that fit it all into a small FPGA.

2/2/2012 2:13 PM EST

Nice to see Prof williams and some of his kit.
He tutored me as a graduate student at Manchester University in the early 50's.
What few computer engineers would be happy with today is only computing with your babies overnight.
The Williams CRT memories were directly in line with the new microwave comms link, setup by the GPO to the north of the computing labs and pointing south!
The signals randomly corrupted the CRT memories.
So Freddie and his crew had to work at night when the GPO engineers were sleeping.
The door in the photogr

2/2/2012 4:03 PM EST

Intel introduced the 8048 in 1976 while the PIC1650 was introduced in 1977. Why would anybody consider the PIC1650 be the first microcontroller?

2/2/2012 7:00 PM EST

The Williams tube should be called the Williams-Kilburn tube since both contributed equally in its development and received equal royalties. In reality, it's just a standard cathode ray tube with a sense plate in contact with front face. W & K used VCR97s mainly (military designation of Mullard ECR60). These were of variable quality, so they had GEC make a version in clean conditions (so the screen was pure) which was also 2 inches longer for better focus.
In the Manchester Baby replica built recently in the UK, VCR97's are used.

2/2/2012 7:12 PM EST

I've had the "privilege" of working with many of the quirky devices you mention (plus some others as well).
One of my favourites was the RCA Cosmac. I worked with an RCA eval kit using the original silicon-on-sapphire version of the 1802 processor (beautiful white ceramic devices on a royal blue PCB) which had the uncanny ability to retain its state overnight without applied power. The 1802 had a single output pin (Q) controlled by set and reset instructions and there were rumours that this was intended to fire the squib in intelligent armaments.
The Signetics 8x300 was great fun too and you could warm your hands on it during long winter debug sessions in a cold draughty lab.

2/7/2012 1:48 PM EST

Quirky and the least favourite processor that I ever used was the Scenix SX18AC. Scenix became Ubicom who then licensed the chip to Parallax, if I remember.

The device was pin and code compatible with the PIC 16C5XX series. The only peripheral it had was an 8 bit timer clocked at the same speed as the instructions executed which made some sense for interrupts (of which there was one). The RISC processor ran at 50MHz (and later 100MHz if memory serves) and the idea was that you could make any peripheral ("virtual peripheral" was the terminology) in software because the execution speed was fast enough.

I made a remove voltage monitor with a built-in 300 baud FSK modem (the analog parts had to be added on). I can't believe that I bought into the philosophy for a while.