On Mon, 12 Dec 2011 18:04:35 -0800, Jeff Liebermann wrote:

On Fri, 09 Dec 2011 08:31:16 -0600, Jim Yanik
wrote:

I think the solder crystallizes with age,then thermal cycling cracks it.

Nope. Alloys do not crystallize unless the component metals separate.
That's not going to happen in a fairly short period of time. What
might happen after perhaps 30-50 years is the solder form "grains"
thus separating the tin and lead. Add a little electrolytic action at
the grain boundaries, and we have a crack.

the automotive environment is very tough,hard on electronics.

Yep. The vehicle operating temperatures (-40C to +125C) are similar
to the mil spec electronics temp range (-55C to +125C).

lots of thermal cycling,shock and vibration.

Not thermal cycling. Copper and solder have almost the same coef of
thermal expansion:
solder = 13*10^-6 in/in/F
copper = 10*10^-6 in/in/F
The difference would need to be much more for the solder to
mechanically separate from the copper due to thermal expansion.

However, if the relay is mounted directly onto the PCB, with no
provisions for absorbing motion (shock or vibration), the combination
of PCB flex and mechanical stresses are going to break the joint. I've
seen PCB's where components will "retract" slightly when I hit the
connection with a soldering iron. That means the joint was
pre-tensioned or under stress from the original soldering and just
waiting to break. That's why I suggested a rubber washer or preform
under the big components.

Relay armature, diodes and resistors on this board are all clad Iron.

--
Boris