At 15:07 3/12/01, Frieder Faig wrote:
Have you ever wondered of MTBF-values ? Very comon as value for trstiness
of computer disk drives. 500 000h live time for a hard drive?? Well it is
not the time a drive will last between two repairs. It is a rahter
theoretical value. Interesting is the reciprocal value, which is the
likeliness your equipment will fail. Too sad, Ive no MTBF values for OM-2S
circuits. But this MTBF may give an idea of reliability of (good)
electronics.*)
Frieder Faig
Failures are randomly distributed around a mean life expectancy. This
means my OM-2S may have a longer or shorter life span than someone elses .
. . even with the same usage rate and care. (Excluded are failure inducing
events such as dropping it.) If you were to study enough of them, you
could build a probability distribution.
Other than mechanical wear which tends to "tighten" the probability
distribution around a fixed life, the MTBF of most devices is modeled as an
exponential distribution. It represents the lifetime of a device with a
constant probability of failure and models most electronics fairly
well. (Reliability studies often use a Weibull.) The attractiveness of
the exponential distribution as a model is its inverse, a Poisson
distribution. The Poisson can model the probability of how many failures
will be encountered per unit of time (e.g., failures per year). None of
these looks anything like a "normal" distribution (bell curve).
What does this mean (pun completely intended)?
Users would be interested in the exponential model of the MTBF as it would
tell us what to expect for a lifetime, and how much the lifetime for one
may vary for another.
OTOH, John H. and Clint, as repairers, would be interested in the Poisson
model which, given an estimated population size, would tell them what to
expect in failures per year (i.e. how much business they can anticipate as
some percentage of all failures) and how much that could vary from year to
year.
-- John
< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >
|