The Adventure of the Stressed-out Switches

A young engineer discovers that a materials problem requires more than finding the right solution — someone actually has to be willing to do something about it

By Peter Tabur, Contributing Writer

Quite a few years ago I had a consulting assignment to correct a problem with an automotive door lock switch. The auto maker had a failure rate specification of 1 per 100,000 and they were up in arms because this particular switch had a failure rate of 7 per 100,000. To make matters worse, it was an intermittent problem and I only had two faulty switches to study.

Upon investigation, I learned that this switch had been in production for about a year, and its construction was quite simple. There was a small round PCB with three large solder pads to which three wires were soldered and was held in a small plastic housing. The actuator consisted round disk, pressed into the housing, and heat staked to this was a beryllium copper set of bent sheet metal fingers. Rotating this moved the fingers to bridge a pair of solder pads making the contact.

I began by disassembling one of the switches and found no smoking gun. No fingers were broken or bent, and no evidence of excessive wear or corrosion. The switches had a small dab of lithium grease to minimize wear. This grease appeared unchanged and free of contaminants. A quick analysis revealed it was unchanged from new grease (perhaps it had coagulated so the copper fingers could no longer slice through it.) Before disassembly the switch never worked, but once I reassembled it, it worked 100% of the time! I was flummoxed.

I switched course and looked at the manufacturing line, looking for any changes in process, parts, materials, system adjustments, sourcing, etc. to no avail; all was the same…

Then something caught my eye. In a large bin were hundreds of actuators and I noticed that the BeCu fingers were bent at approximately 30°. The ones I had seen from the defective switch were bent approximately 15°. Armed with this insight, I took thirty actuators and measured the height of the contacts prior to assembly. These I had built into switches. I then carefully took them apart and measured the height again. They were all consistently lower!

Here was a clue. A fundamental of mechanical engineering is the stress-strain curve. In the elastic range, a part will deform in proportion to the force applied, and then spring back to its original position when the load is removed. If the yield stress is exceeded this will no longer be true. Simply put, the yield stress of the fingers had been exceeded in the process of assembling them. This suspicion was confirmed via finite element analysis.

I wrote up my results including a redesign of the contacts that still fit within the switch, applied the same contact pressure and would not exceed the yield stress of the material. The engineering manager was dismissive, saying, “This can’t be the root cause; the switch has always worked just fine.” I pointed out that the failure rate suggested that it wasn’t just fine. Also, the yield stress had been exceeded, not the ultimate tensile stress (the point where the part would break). The part was by definition in the plastic range of the stress-strain curve and one could make no predictions of the behavior. Many could behave well, while others not, it all depended on where the fingers finally came to rest.

He essentially told me, thank you very much, and that concluded my consultation. I have no idea if they implemented the design change. The reluctance, I suspect, was that the redesign would require a new, expensive stamping tool.

Contributing Writer Peter Tabur is an ME with 20+ years of experience, mostly in the high tech arena. He relishes creative design and problem solving. In addition to wood working and reading, he loves pestering the Design News Editorial Director with email.