18 Jan
2004
18 Jan
'04
3:55 p.m.
:-)). In fact, I read somewhere that the PROMs can
suffer
from bit-rot
where the fuses actually grow back. And of course the
transistors, etc,
in the address decoder and output buffers can fail.
A few years ago I had a customer issue with a DSW42
synch card in a MicroVAX 3100-85; just one particular
customer and one particular machine but with a group
of different DSW42s. After the usual fiddling around
I eventually reproduced the problem and sent a box
across the Atlantic to the h/w team who were doing
support for this. The answer came back that the PLD
(IIRC) was bad and they'd had a preliminary report
form the manufacturer indicating that for that
particular batch of the suspect part "crystal regrowth"
(or some similar phrase) was a known failure mode.
I think if you wait long enough all ICs will fail,
through solid-state diffusion if nothing else. At
least with the off-the-shelf parts it will be possible
(in principle) to remanufacture a facsimile. With
modern ASICs (and probably FPGAs) this will be pretty
hard to achieve.
Antonio
--
---------------
Antonio Carlini arcarlini(a)iee.org