Back then, 100k ECL cost roughtly ten times as much as TTL and Gigabit's GaAs cost ten times as much as the 100k ECL. More important, there were three suppliers of 100k ECL - Philips, National and Motorola, none of whom were going to go out of business, while there was just one Gigabit. I really would have much preferred to go with National's 300k ECL - the masks were made with a Cambridge Instrument EBMF 10.5 electron beam microfabricator, whose beam-steering electronics I'd personally up-graded to meet the demsnds imposed by the new, fast electron beam resist that National were using.
I was digitising the trigger time delay position vis avis the 800MHz clock by digitising a period between 1.25 to 2.5 nsec to eight-bit accuracy and feeding the digtised result into the digital time delay generator, and using the low order bits of the output to generate a delay with the same resolution. the whole process - from input to output - took about 40nsec, which rather restircted the A/D converters and the D/A converters that I could use.
At the time 100k ECL would have supported a 200MHz clock and 10psec resoluton in 5nsec is 9-bit accuracy, which wasn't on offer at the time. The guy running the project was primarily interested in being able to sell the machines after they had been built, and the 10psec specification would have made that part of his life easier, though in fact it slowed down the development enough that he was never in a position to actually sell a machine.
I've still got my databook, but no samples. They were a bit too expensive to snaffle for souveniers.
I've used Motorola's ECLinPs parts, back when they were good for
500MHz synchronous counters. Much easier to use than GaAs, and from a much more reliable source.The electron beam tester would have sold for about half a million dollars a unit, and we spent something like five million dollars on the development. You can't design a lot of that kind of product every year, and you can't turn them around fast, particularly if you push the envelope as far as we - foolishly - did.
-- Bill Sloman, Nijmegen