A question for the future

(C Kingsbury) wrote in
om:

Andrew Sarangan wrote in message
.158...
"Neil Gould" wrote in
news
Recently, Andrew Sarangan posted:

The same thing was said about glass cockpits. Now even the ubiquitous
172 comes with a glass cockpit. It won't be long before the backup
vacuum driven gyros are removed from their panel.

And with good reason: just as transistors proved more reliable than
vacuum tubes, solid-state AHRS look to be far more reliable than their
replacement. There's a huge benefit to safety, utility, and ultimately
cost by making the move.

However, I would not want to give up the non-powered wet compass,
altimeter, and airspeed indicator completely. Not until we prove that
mice can't chew through wires, etc. What's the benefit of tossing
these completely?

Likewise, I suspect navigating by reference to magnetic north will be
one of those charming anachronisms that our descendants 2000 years ago
will talk about in the same way that people today talk about how
railroad gauges were based on roman roads. (I know it's true only in a
loose sense, no need to rehash that here)

You can't count the number of components in a circuit and assign
failure modes to each one of them. If that were the case, your
computer will not be able to run for even a minute. There are
millions of transistors inside your computer, with million different
failure modes. The traditional method of counting failure modes of
mechanical parts do not apply to highly integrated electronic
products. Yes, there are a few failure modes, but not as large as you
make it out to be.

Um, I'm calling bull**** on this assertion. Isn't the real key here
that the odds of a particular part (say a transistor gate inside a
CPU) failing are simply infinitesimally small? Because if one
capacitor on your PC motherboard smokes out, it's quite likely that
the whole shebang will in fact not work properly.

Best,
-cwk.



The real key is that all the transistors on a chip are made on the same
substrate. They share all the same characteristics. The failure of one
transistor is not an uncorrelated event from another transistor failing.
They all fail more or less at once, or they all continue to function.
They behave as single component. Just because there are millions of
transistor on a chip does not mean that there are millions of different
failure modes. There are only a handful of failure modes regardless of
the number of transistors on the chip. This is what is fundamentally
different about integrated circuits compared to discrete circuits.