Electrical Component Thermal Fatigue?

[Foreman]

I'm a structures guy, but I've been asked to look at thermal fatigue for a surface mounted transistor. To give you some background, it is company policy not to use surface mounted components on electronics boards. But the electrical guys want to use a specific surface mounted transistor, but cannot perform a thermal cycle test on any board with the transistor until they have some analysis showing the transistor good.

I'm pretty ignorant to electrical components, so any information would be greatly appreciated. I have a book by Steinberg, called Preventing Thermal Cycling and Vibration Failures in Electronic Equipement, but I've heard his method is flawed and is not neccessarily a conservative analysis.

How does one go about doing one of these analyses? Does anybody have fatigue properties for Solder? Does anybody have good advice for me?

Thanks in Advance

 

[saxon]

Foreman, Sounds like you've got a mind set problem. They want empirical grade data, without running the empirical analysis. Good luck! And the Electrical guys could never figure out why I always wanted Instrumentation and Control boards "burned in" for thirty days prior to acceptance. Their answer was always, " But our calculations show ...". When burned, there never was a short term board failure and system collapse.

Sorry I couldn't be of more help.
saxon

 

[Foreman]

Saxon,

I agree, it's a Catch 22. We can't test the boards until we have analysis showing that it will pass the thermal cycling qualification. And we can't do an analysis without good empirical test data.

I do have some empirical data for thermal fatigue of solder. The Engelmaier curves from Bell Labs, relate fatigue cycles to strain. And, I can calculate the thermal strain using the Steinberg method I mentioned in my first post.

The problem that I have, is that I have heard the Steinberg method is flawed and may not give acceptable results. I am searching for a "better" method, to calculate thermal fatigue.

 

[IRstuff]

how many cycles are we talking here?

TTFN

 

[Foreman]

12 Thermal Cycles for acceptance testing from -45 deg C to 125 deg C.

 

[IRstuff]

Anyway, it sounds like your company is paranoid about surface mount for no good reason.

Issues with surface mount solder and fatique were solved nearly twenty years ago.

TTFN

 

[hacksaw]

while there has been a remarkable amount of progress in component assembly, i don't think that solder and fatigue problems have been solved, better understood might be a better description

sounds like the poster has two issues at hand: the fatigue limitatations of the transistor (the manufacturer will have to help you there) and the fatigue limitations of the surface mount/board/and fastener technology.

Designing to 12 thermal cycles falls in the category of low cycle fatigue, where failure is defined by the peak thermal stresses (i.e. peak operating temperatures). sounds like you have a fairly simple design issue to sort out.