Fw: Deck of IBM PLAYING CARDS GOES FOR $325

[William Donzelli]

> > I recall Neil Lincoln talking about the "box of Chiclets" approach to
> designing the Cyber 205--something like only 11 or 13 different IC types.
> FWIW, I don't know how many types the finished product had, but a small
> number of different types of packages was definitely in the air those
> days...

I seem to recall that the 205 was indeed made of a relative handful of
chip types, based off a standard simple gate array. I do not know who
fabbed the things.

Too bad no complete (or even remotely complete) 205s exist. Now that's a
knoife..

Anyway, 11 or 13 is way more than the (effectively) 3 types used in the
Cray-1. None of the ECL families were very big anyway, not the hundreds of
functions that TTL or CMOS did.

> On the opposite extreme, I can remember looking at the various databooks
> and seeing something like a SN74LS302985 "13-bit LFSR with parity and
> tristate complemented outputs with bottle opener" and thinking it would
> just be PERFECT for what I had in mind--if only there were a second source
> or three.  Then, as now, it wasn't uncommon to "pre announce" a chip that
> never made it into production.

Quite a lot of the really oddball parts that get into databooks are
probably customer specific - at least this is how the transistor industry
was. A customer would request a small tweak of a standard part - maybe
even so small that the fab process did not even change, but the tests did
- and BANG, a new number was registered. The tube guys played this game as
well.

William Donzelli
aw288 at osfn.org