At 04:25 PM 7/8/2012, neil n wrote:
>Cool. The explanation and insights help.
>
>Ok. So when one pops open their ECU, they aren't just looking for
>failing solder joints due to time, use and heat (normal wear and
>tear).

No, that's what you're looking for.  Unfortunately it takes a strong
microscope (35X Stereozoom is what I use) to see small failures, at
least with my eyes.  My point about the Bosch (and to lesser extent
T/A) solder is that it's so ugly looking that it makes it harder to
see the fatigue problems.

>I had another look at your Picasaweb album here:
>
>https://picasaweb.google.com/117189706757545167023/SolderFatigue
>
>in particular this image:
>
>https://lh3.googleusercontent.com/-YdpQ04ceKbs/TWxSLphU3nI/AAAAAAAAH4s/Ieu85LxefaY/s720/IMG_0556.JPG
>
>That's a lot of crystalized (or crystalizing) joints. I had assumed
>these were due to time/use/heat.

Those are straight fatigue failures from vibration.  The solder is
tearing roughly along the horizontal axis of the photo.  That's one
of the ribbon interconnects to the other board (I would be inclined
to resolder those on general principles.  Also any heavy components,
any that are raised off the board, any that get hot.  Trouble is that
leaves a huge number of joints that are all suspect).  Probably all
those joints in the photo are still making contact, but not absolute
certain; and there will certainly come a time when contact is
intermittent or fails entirely.  In the ignition drive circuit the
voltage and energy are high enough to cause arcs that will burn away
the contact once it becomes intermittent, but for the rest they sit
there looking innocent and not working.  On a severe case with a
freestanding component you can move the component and see the lead
move on the other side of the board.  By that time you can likely see
the failure with naked eye.

A further difficulty is that once the bond fails, the lead may
oxidize and be unwilling to be wetted with new solder, even with normal flux.

>  If the saw process was used on
>(some?) Bosch ECU's to cut leads,

It wasn't.  If you ever see a sawed board you'll know it right
away.  All the leads are bent over like trees in a storm, and have
flat tops with nasty sharp points and a bur.  Hateful things to work on.

>  maybe this helped spur on poor joint
>quality? (different solder introduced to manufacture process,
>manufacture process not changed) Anyhow.....

I don't think the joints were actually bad from the factory, just
bad-looking.  I could be wrong about that, and no place I've been
around would be satisfied to see something like that coming off the
line; but my big complaint is it makes it much harder to see the
fatigue problems.

I think hard-mounting the ECUs to the chassis was a mistake, and a
rubber mount would have preserved them.

>A comment I recall reading in the archives. If one was certain, or
>nearly certain that they were dealing with a poor or failing ECU
>(solder joints), while symptom present or crops up, give the ECU case
>a rap.

Why wait to be certain?  Easy to do.  No change, no meaning; but if
behavior changes then you've nailed the ECU or connector dead to the
floor.  Try it when the beast is running right...

Yours,
David