The "Guru of Advanced Thinking About Advanced Radios" of Rockwell-Collins
gave an interesting talk .... One point he made was that (even) Collins
had to buy enough chips for their complete expected production run, and
eventual spare parts, right at the design-in stage. Otherwise either -
1) The part would be unavailable before the production run was completed, or
2) A part with the same number would be available, but it would be enough
different that it would either not work or would degrade performance.

How true, how true!

I worked at Fairchild Semiconductor's Research and Development Lab from
1967-71, where we designed and built SYMBOL, a computer with a radically-
different architecture (OS, compiler, editor, etc., were all in HARDware
[and I do NOT mean ROM'd firmware; just lots and lots of gates and flip-
flops]). When we began actual construction, we tried to stockpile --
from Fairchild Semi's production line -- enough of every IC type we
thought we'd need, but we ran short by about 100 flipflops, so we got
three more "tubes" from Production.

It is safe to say that those hundred FF's caused us MORE TROUBLE than
the other 4,000 FF's: after two years of production, those later FF's
had so much FASTER "setup" time that our design rules failed when we
put those FF's and earlier ones on the same printed circuit board!

--Myron.
--
Five boxes preserve our freedoms: soap, ballot, witness, jury, and cartridge
PhD EE (retired). "Barbershop" tenor. CDL(PTXS). W0PBV. (785) 539-4448
NRA Life Member and Certified Instructor (Home Firearm Safety, Rifle, Pistol)