I dont understand that. I'm familiar with the bathtub curve etc but PCs
arent normally anywhere near the steep bit of the curve at that age. Re
pcchips, yes I got what someone else paid for, but its what we've got,
and are unlikely to be getting any more for a bit yet.
After a few years you have several cumulative effects-
- dust/dirt/residue/etc accumulation on contacts
- fans "might" be wearing out
- battery replacement. A battery is cheap, but the time to
order, stock, go around and open all systems changing the
battery then setting CMOS values, may not be so cheap.
Another cost weighed against value of system.
- another service interval is needed, the time (value of
it) offsets the value of the system which has been dropping
until the time is worth more than value of system
- hard drives aged, significant failure point
- motherboard, video, PSU, etc, electrolytic capacitors
wearing out. Cheap systems in the Pentium 1-2 era can be
expected to have a shorter life from this factor than
earlier that used a larger percentage of tantalum caps, and
higher voltage, lower current
- year after year of power surges... eventually take their
toll. Even a surge protector has to have "some" kind of
threshold, under which surges get through. Maybe it only
effects the power supply, but as with the time to do
maintenance, time to diagnose, time to buy and install, and
cost of a power supply are all weighed against the value of
the system.
- Expected lifespan or failure rate is merely an average.
Suppose the systems are "on average" expected to last
another 4 years but you have 25 of them. 4 might fail in
the next 2 years, or 15 might. What did the downtime cost?
Could be $0, or could easily be several hundred dollars per
event if not a LOT more.... Only you can know this, or have
enough info to attempt a calculation.
- M570 appears to have been released in late '97, even if
the system wasn't built or deployed at that point the board
may have been manufacturered and sitting, aging. Almost a
decade old, we actually don't have any data to compare and
determine that such systems WOULD last over 10 years. As
mentioned above, this was a newer era of switching onboard
power, higher currents and more extensive use of
electrolytic capacitors. These systems failing now are not
necessarily a symptom of a problem in lifespan that goes
against the "norm" but rather, are the beginnings of
evidence on what the expected lifespan of the technology
really is... if it were't that they were PCChips boards, and
possibly the systems are equipped with similarly low-end PSU
and fans?
it would not be a good option at all to pull these out now. Wrong time
entirely, hence trying to figure out how to diagnose/fix them. In past
it would have been no big, in future ditto, but not now
If they can't be pulled all at once, one at a time may have
to suffice. How do you suppose to "fix" them?
You had the previous dead system, yes? Did you trash all
the parts? Swapping known good parts for questionable ones
would isolate each variable. Perhaps you can pull each
board in turn and replace all the significant capacitors...
it is a non-trival amount of time (including downtime) and
cost as well... all to keep using systems of little to no
value that will need to be replaced anyway since having the
caps replaced doesn't eliminate any of the other issues.
I know theyre good, had the dvm on them and theyre right on the mark,
with no measurable ripple. Will try cmos defaults too.
You cannot alway conclude "good" by taking a voltage reading
with a DVM. It is a better way to indicate a problem rather
than declare there aren't any.
