On Tue, 16 Jan 2007 12:12:21 +1100, "Rod Speed"
wrote:

lsmartino wrote
Rod Speed wrote

Please explain how the manufacturer of a light bulb, fluorescent
lamp or CFL can provide an estimate of the lifetime of the lamp.

That is trivial to do by running an adequate sized batch of
incandescent light bulbs under appropriate test conditions etc.

Don´t say "They can´t because it´s impossible".
Explain exactly why it´s impossible.

Its up to those who claim that its possible to design a device
to die one year after the warrant expires how that can be done.

Well, it can be estimated how long will a power semiconductor
run if you leave it without a proper heatsink.

Not in a domestic environment it aint, because
the ambient temperature varys so much.

I got a bit of a start relatively recently when someone was having
overheating problems with their PC to discover that they were one
of the few in this area who were silly enough to have no form of
cooling whatever, not even a swamp cooler, in an area which can
see 10 days over 40C some summers. We had one just last week
and it got to 44C, and it was like walking into a furnace walking
outside my airconditioned house.

Open any Samsung TV, for instance, to see for yourself
how important transistors are left bare, dissipatting heat
to the air. I don´t see why it should be difficult for the
manufacturer to know that these particular transistors
left overheating will fail within a finite number of hours.

There is no finite number of hours, because the
ambient temp varys so much in domestic situations.

Those who dont have any cooling at all in an area which
can see a week over 40C wont survive the warranty
period and those who have decent air conditioning will
find that the TV lasts long past the warranty period.

Also, you will notice that the same circuit will have electrolytic
capacitors near heat sources, when it´s a well known fact that
heat shortens dramatically the life of electrolytics caps.

In practice that isnt a significant problem with domestic appliances.
Essentially because you dont see many electros in that situation with them.

The manufacturer know how to properly design an electronic
circuit in order to provide a long life, but it also it knows how
to design it to fail within a short term under certain conditions,

No they dont on that silly claim about surviving the
warranty fine, but failing immediately after that expires.

and accordingly they estimate a warranty just long
enough to cover the product for a safe term, a
safe term for the manufacturer, not the user.

Have fun explaining how come not a single electronic device
I have ever owned has died just after the warranty has run out.

And that includes my latest gigantic widescreen TV too.

Of course it´s impossible to predict exactly how many years
the TV will last, but the manufacturer count with statistical
data which says, for instance, that a TV set is turned on
10 hours per day for instance, and taking that into account,
and estimating how long the weakest part of the TV will last
under these conditions, they can determine the warranty lapse.

Pity about the TVs that get left on all the time.

The claim is completely fanciful and those making that sort of
claim have obviously never actually designed a damned thing.

And only the stupidest manufacturer would deliberatly design
their product to die as soon as the warranty has expired
anyway, because the bulk of those who had bought such
a dud wouldnt be buying another from that manufacturer.

They also do product reliability testing to see
how long on average it is before a product fails.

No they dont with domestic appliances.

They dont even do that with mass market hard drives anymore.

Yes they do.

No they dont.

They quote the useful lifetime of a hardrive in MTBF hours.

That is calculated, not measured. Convert that MTBF
to years and you will discover why they cant possibility
have tested them to get those numbers.

Don´t answer "it´s impossible" if you are not prepared to give
a real explanation. Samsung, Seagate, WD... any decent hard
drive manufactures gives an estimate lifetime of their products.
These estimates are provided in the datasheet of each harddrive.

And they are ESTIMATES, not measured results.

Exactly, these are estimates,

Pity the claim was about TESTING, which doesnt happen, like I said.

and most of the time very accurate,

Like hell it is. Have you actually tried converting
the MTBF of a current hard drive to years ?

The average quality EIDE drive has a published MTBF of 400,000 hours.
That wouild be 45 years on my computer. I've had LOTS that never made
3 years. If they test 1000 drives for 400 hours and get one failure,
they have their MTBF of 400,000 hours - 1 failure in 400,000 hours of
running. They will actually do a larger test sample over a larger time
span Likely 2500 for 500 hours. That gives them 125,000,000 running
hours and if they have 3.125 failures they have a 400,000 hour MTBF.-
but that's how the numbers are arrived at if they are not just using
statistical analysis methods.(predictive failure). Today's hard drives
with S.M.A.R.T. technology can predict their failure date quite
accurately. (using third party software).I just pulled 2 drives from
service because they predicted their own death in less than 60 days.
One was made on the 123rd day of 2003 (seagate), the other the last
day of January 2004 (wd).
Being a WD Caviar retail drive it has a 1 year warranty. If it was a
"distribution" drive, it would have a 3 year warranty. Might have
lasted 2 years - but I don't take a chance on my data.
The Seagate has a 1 year warranty, and was in a computer that only
runs a few hours a week - and lasted less than 3 years.
I used to work for the (then) largest hard drive distributor in
Canada.

specially those concerning the maximum number of startups/stops
the drive can tolerate before the heads get completely worn.

Wrong again. Its such a round number it cant have been produced
by TESTING, and the number of starts and stops dont produce
any wear of the heads with modern hard drives anyway.

I don´t want to imply that all manufacturers are dishonest per se, but
I can easily see how a given manufacturer can produce different items,
with differents level of quality of design and manufacture. And these
differences *will* impact the useful lifetime of the final product.

Separate matter entirely to the claim that they do reliability TESTING with domestic appliances.

They dont, and dont with mass market hard drives either to
produce the MTBF or the number of start stop cycles either.



--
Posted via a free Usenet account from http://www.teranews.com