Computer Finally Crashed

[Paul]

Solid state components don't generally "wear out", except that some diodes,
transistors, and IC's might eventually develop failures over an extended
time due to surface and other manufacturing defects. Electrolytic
capacitors, lamps, motors, switches, etc, are a different ballgame, however
(but they're not solid state). The lifetime problem of those components is
simply due to their mechanical or chemical nature. Resistors have no such
problem.

It depends on the operating conditions and the process used.
Electromigration is an issue. The clock speed of my P4 Northwood
gradually dropped during the time I was using it. And there were
a couple reports of a more serious clock rate decline on some
AMD processors. I don't know how significant electromigration
is, with respect to a target of a thousand years of operation.
One of the reasons processors have some "headroom" when they leave
the factory, is to cover parametric shift over lifetime. (One
AMD slide showed this headroom allocation as "500MHz".)

http://en.wikipedia.org/wiki/Electromigration

DRAM failures seem to be different. I've had DRAM fail while
it was stored without bias. It suggests chemical contamination.
DRAM up to and including FPM/EDO, seemed to be better than
the stuff we have today. At one time, I would have been
gung-ho for the "it'll run forever" camp, but the evidence
that is true is no longer there.

So realistically, if you asked how to make a computer last
a thousand years, I'd have to answer "self-repairing circuits".
And then leave the interpretation of that, to your science
fiction imaginations

Paul

 

[J. P. Gilliver (John)]

In message <[email protected]>,

That is true. In fact most solid state consonants actually fail
because of surges, ESD, heat etc,
That could happen the first day you own it.

Vacuum vowels, however, can last for decades if no-one drops them.

 

[Stef]

Stef wrote:

[big snip]
Never wear out? Great idea. So far, just a concept. The reality of it
has yet to be achieved. I doubt it ever will. Nothing lasts forever.

Last a 100 or a 1000 years? I wouldn't be surprised, too. I'd be very,
very, very skeptical, bordering on incredulous. ;-)

Stef

Solid state components don't generally "wear out", except that some diodes,
transistors, and IC's might eventually develop failures over an extended
time due to surface and other manufacturing defects. Electrolytic
capacitors, lamps, motors, switches, etc, are a different ballgame, however
(but they're not solid state). The lifetime problem of those components is
simply due to their mechanical or chemical nature. Resistors have no such
problem.

You're being too literal. Wear out as in stops working. Not wear out
like an automobile tire. How long have you (and BillW50) had this
obsessive compulsion for absolute correctness? ;-)

As far as resistors, I've seen them fail, too. Lots.

Nothing lasts forever.


Stef

 

[J. P. Gilliver (John)]

In message <[email protected]>,
(e-mail address removed)
writes:
[]
That is true. In fact most solid state consonants actually fail
because of surges, ESD, heat etc,
That could happen the first day you own it.

Vacuum vowels, however, can last for decades if no-one drops them.

I presume you mean Vacuum Valves. Valves is an early name for Triode
Vacuum Tubes.

Oh dear (-:! My "vowels" was supposed to be a jocular rejoinder to
gfretwell's "consonants", but nobody else spotted it!

I think the British still use the term "valves" for all vacuum tubes,
regardless of its type (e.g: triode, tetrode, pentode, or whatever). (not
just triodes)

Correct, we do. UK valve = US tube (or "toob").
Not electrons! That would just represent a charge. There is a _coating_
on some of the electrodes that can wear off in time.
Indeed; they weren't that reliable in portable use, especially
vehicular.

 

[Stef]

Stef wrote:

[big snip]
Never wear out? Great idea. So far, just a concept. The reality of it
has yet to be achieved. I doubt it ever will. Nothing lasts forever.

Last a 100 or a 1000 years? I wouldn't be surprised, too. I'd be very,
very, very skeptical, bordering on incredulous. ;-)

Stef

Solid state components don't generally "wear out", except that some
diodes,
transistors, and IC's might eventually develop failures over an extended
time due to surface and other manufacturing defects. Electrolytic
capacitors, lamps, motors, switches, etc, are a different ballgame,
however
(but they're not solid state). The lifetime problem of those components
is
simply due to their mechanical or chemical nature. Resistors have no
such
problem.

You're being too literal. Wear out as in stops working. Not wear out
like an automobile tire. How long have you (and BillW50) had this
obsessive compulsion for absolute correctness? ;-)

As far as resistors, I've seen them fail, too. Lots.

Nothing lasts forever.


Stef

Resistors fail IF subjected to adverse conditions (such as overvoltage and
running too much current through them, which exceeds their power rating).
Other than that, no, I don't recall any such incidents.

Admittedly, resistor failure is rare, but it happens, regardless of the
cause. Nothing last forever.

A penchant for correctness and detail probably comes from an engineering
background (alas, I was never an artist, but sometimes wish I were .

To a pessimist, the glass is half empty;
To an optimist, it's half full;
To a realist, it's half full AND half empty;
To an engineer, the glass is twice a big as it needs to be. ;-)

Stef

 

[Ken Blake, MVP]

I am old enough to remember TVs and Radios with tubes in them

Me too.

IBM also had tubes in the early computers.,

And so did all other computer manufacturers. IBM used three-digit
model numbers for computers with vacuum tubes, and four-digit model
numbers for those with transistors. For example, the 709, which came
out in 1957, had tubes, and the 7090, which came out in 1959, was the
transistorized "improved" 709.

The MBTF

You mean MTBF.

on some of those old computers was measured in hours, not
even days. Tubes go bad .,.. a lot.

Yes.


So much so, almost every drug store and hardware store had tube
testers and a fairly good stock of the common ones.

Yes, I remember it well.

 

[J. P. Gilliver (John)]

[Can you hint at your email in your posts? I know this thread has a life
of its own, but ... (-:]

David H. Lipman wrote:

[]

Not electrons! That would just represent a charge. There is a _coating_
on some of the electrodes that can wear off in time.

Mostly the cathode, I'd suggest.

Yes, I think so.

I have a question for you, John. We here in the US use a ---/\/\/\/---
symbol for a resistor, which I think is pretty apt, but you guys use a small
rectangular box for a resistor as I recall. Is there any reason you guys
didn't also adopt that neat ---/\/\/\/--- resistor symbol? Just curious,
that's all.

I don't think it's UK/US: it was in the early days of computer-aided
design, I think. We used to use the zigzag too. But early computer
drawing systems had difficulty with anything other than boxes, with the
result that resistors became boxes, those nice semicircular AND gates
became boxes, the invert bubbles on the output of NAND gates became
triangles, ... and, unfortunately, still are, on both sides of the
Atlantic (certainly, in TTL databooks for example, though they usually
do show the curvy-gate equivalent too, especially when showing the
equivalent internal diagrammes).

One for you: we do non-polarised capacitors as two parallel black
rectangles -||-, and polarised ones with one of them not filled in; you
do them as one curved and one flat plate -|(-, even for the
non-polarised ones. Any idea why?
--
J. P. Gilliver. UMRA: 1960/<1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

The main and the most glorious achievement of television is that it is killing
the art of conversation. If we think of the type of conversation television is
helping to kill, our gratitude must be undying. (George Mikes, "How to be
Inimitable" [1960].)