# Failure estimation

Discussion in 'Electronic Design' started by R.Lewis, Sep 30, 2003.

1. ### R.LewisGuest

How do I estimate the failure of a custom wound inductor to get a MTBF
figure.
(In fact a RM10 ferrite, Al 630, wound to give 50mH, varnish impreg coil
etc)

Thanks in anticipation.

2. ### Mark FergersonGuest

First, determine its mechanical and electromagnetic
properties (and their limits) as built but not installed.
Find out how it's supposed to behave when installed. Figure
out what it would take to change its properties out of
acceptable limits (as installed).

What circuit is it built into? How is it mounted, and on
what substrate? How much voltage will it see, how often, and
for how long? How much current will it carry, how often, and
for how long? Is there any airflow, and if so, how much, at
what temperature? Is it potted? Will it ever see water or
other contaminants (how much etc)? Is it subjected to
mechanical vibration or shocks (how much, etc)? Will it see
externally-generated magnetic fields (how strong, etc)? Will
it be near a voltage high enough to arc to it (how high,
etc)? Will it _cause_ any thermal, mechanical, or
electromagnetic interference with other components that will
feed back to it that could contribute to its failure?

(Did I forget anything?)

IOW, think about what in its environment and usage
conditions might cause it to fail, then compare against its
physical, mechanical, and electromagnetic properties _as
installed_.

If you don't know those properties, talk with the
designers/engineers/etc until you do. Do _not_ let them blow
you off.

While you're at it, figure out how many ways it can fail,
and what the circuit it's in will do as a result.

BTW, if you don't know how to do this, what idiot tasked
you to do it? ;>)

Guess. Build a few and test to destruction. Refine guess.
Repeat until you're confident enough to present the result
to the idiot, uh, your boss. If they don't like it, tell
them what to change to get it to last longer.

Get promoted. Buy me a beer. (worth a try)

Mark L. Fergerson