My IC Memories - © Rob Wilson 2003
This series of ramblings is an account of my life in IC design. It's a living document, i.e. it might change from time to time.
I went through university (1969 to 1972) on a thick sandwich course - the idea being that you did a year with your sponsor before university, worked for them each summer vacation, and then did a year after university. My sponsor was the UK Ministry of Defence, which paid me a good salary - better than a student grant! But their training wasn't up to much, at least until my last year, which I spent at two research establishments. After this I went back to my "base" establishment and got involved with designing thick film circuits. This was a new in-house capability intended to show Ferranti, an early UK IC company, that we could compete with their custom analogue IC on a particular project. Thick film was really quite new then and seemed to hold a lot of promise; particularly useful was the ability to have resistors of any value and the ability to trim them, not just to a specific value but also to achieve some required function such as a filter cut-off frequency.
After a few years I'd had enough of working for the government and applied for anything going in the corner of England (rather arbitrarily) south and west of Oxford. After a gruelling interview at Normalair Garrett in Yeovil (then part of Westland helicopters), which did not result in a job offer, I had a more rewarding session at Pye TMC at Malmesbury. This company had been based in Dulwich, London, and had just moved out to the West Country.
The Pye group was quite large, being famed especially for professional wireless products, audio and TV, even a record label. TMC (for Telephone Manufacturing Company) seemed to be fairly autonomous though and was expanding fast. Its competitive advantage was the ability to design custom ICs - at that time very few businesses outside the semiconductor industry could do that.
I started work there in the autumn of 1977 as part of SBS (Small Business System) team, most of us newcomers, to develop a new office phone system or PABX. Unlike previous Post Office systems this was to have electronic terminals with programmable keys and "lamps" (LEDs). It was to be controlled by a central computer, which we were going to build from one of these new microprocessors (the Intel 8085 and Zilog Z80 were then the state of the art) - more on this later!
After we built a demonstration system, using standard parts and a Signetics 2650 micro, work started on the real thing. I was given the task of designing a chip which flashed the LEDs. Since a phone terminal could have 30 or more of these LEDs and the power consumed by the terminal had to be kept low, then the LEDs had to be stacked with as many as possible in series so that the current could be re-used. Now it so happened that the IC process we used then was PMOS (PMOS transistors were easier to build consistently than NMOS) and the supply voltage could be 17V. We would run these chips from +5 and -12V supplies, which at least allowed a fairly straightforward interface to TTL and other logic parts running from 5V. So we could have a stack of 6 LEDs. Each LED would have a PMOS switch across it which would shunt the current if the LED was to be off.
Designing the analogue stuff was all done by hand, with simulation being done as a final verification. At first we used an in-house simulator called Philpac (by now we had become part of the Philips empire); only later did we move to Spice. Simulations were run on a remote computer using a 75 baud teleprinter terminal! So you took care to get things right first time. Schematics were drawn on paper and netlists generated manually - a far cry from today's environment. Now the logic design was interesting, because we used a form of synchronous dynamic logic called four-phase, which did not need much black magic to get right.
Next: John Rhodes and four-phase
Useful links to interesting stuff: