Localized Thermal Testing - John Dunn, Consultant, Ambertec, P.E., P.C.

Imagine that you have this really complex circuit board which is going awry in some way when it gets hot and you can't seem to figure out which component is the culprit.

You might be able to isolate the problem to some one particular part if you could just heat up that one part by itself without also heating up the nearby surrounding parts.

You could try making a small tool and then do this:    

 

If applying this very localized heat to a suspect part brings about the problem, you've found your culprit.

Note that although you  could use the soldering iron itself to heat a suspected part, the soldering iron might dribble some molten solder, an event that would likely be bad news. The hot nail won't do that!

 

Comments

[Anthony Stewart EE'75]

I recall 35 yrs ago simply using a thermocouple to calibrate myself to 2 methods I used. In that era, Epoxy and Ceramic DIPs were most common so the package was fairly big. and might have a 10~30 second time constant for a small area heat source faster with a large area source like a hair dryer even faster but harder to control with a heat gun. After I discover a thermal issue quickly, I could zero in on the culprit with a soldering iron tip.

Method
First hold for a second then release and tap it quickly to raise the chip temp slowly in 10 seconds or so until it was hot enough. If you want to simulate thermal shock, that may trigger different failure modes unlike slow thermal margin failures. THe finger test would teach you which parts were hot from self heating, which is also a good way to find problem devices. A clean tip with a moist sponge is good and if the epoxy is melting, you are holding it too long... so tap tap tap.

Many soldering irons are grounded but safe irons are floating with a Mohm resistor for static leakage.

I used this method to debug an HP rack mount gold-plated circuit board that failed in the rack but never failed on the card extender. So I said to myself... aha self heating problem. Got out the hair dryer then soldering iron and found the defective chip. Sometimes chips and LEDs fail due to the 10 micron gold bond wire being broken and separating with thermal switch characteristic. In this case, the card extender provided sufficient convection cooling to prevent debugging so it never failed until heated.

Anthony Stewart EE'75