Non Distructive testing (letting the smoke out)

[Roger Heflin]

On Mon, 11 Jan 1999, Bruce Plecan wrote:

> Nother thing, what would be a "safe test" for how long overloading
> a driver would take.  If I wanted to see if something was going to
> work on injector drivers, how long in running them would it take to
> see if they fail?.
>   I have two items to test, running 16 injectors off of a 8 cyl TPI ecm
>   Running 4 P+H injectors where 2 are normally fittted.
> I would figure if the ecm gets only warm to the touch that would be one
> clue.
> Cheers
> Bruce
> 
> 
Probably by default it will get warm under normal conditions.  You
would probably need to find out what temp would damage the driver
transisters.   If you know the max temp, you would then need to
measure the temp on the driver.  I guess if you where worried you
could put a sensor on the drivers and have it warn you if you
where getting close.  Also if you cool the drivers (fans
generally) that will most likely increase the lifetime of the
drivers.  Also to properly test, you would have to do the test
under at least two extreams (hot 108 maybe, and cold) either extream
could make the drivers burn out depending on how the other components
increase or decrease the current.    

On overloading there are several different ways, one way is to
overload and smoke almost immediately, but there are alot of other
situations where the lifetime of the driver is significantly reduced,
and this one is generally what will get you.    You could reduce the
lifetime from years, to months and just have it burn on you while
driving.

Semiconductors failure mode from being old is having the parts that
make it a transistor migrate to other parts where they are not
supposed to be.   Higher temps increase the migration rate, so running
it over temp may not fail it immediately, but can result in it failing
much sooner.

I have seen automatic (motion) light sensors that had 2x the power ran
through it.   The electriction replaced it twice and each one burned
out, before he figured out that it was massively overloaded, but took
days to actually burn out.   The how close things get to the max temp
will probably be the best way to tell.   If possible adding a bigger
heat sink will also improve chances.

			Roger