In order to have a design change approved by the customer's design engineers, we would have to prove the cause of failure, and show that the proposed change would fix it. For this, we had to bring in the big guns. Fortunately, MIT was right around the corner. We found a professor in the Department of Materials Science and Engineering who specialized in semiconductor materials.
He was able, in short order, to set up test equipment to measure the complex impedance of the thermistor. Comparing good ones to failed ones, he concluded that there were zones within the bead where the platinum wires and the ceramic pellet had physically separated, reducing the electrical contact and increasing the resistance. Now it all made sense.
In our plant, sensors were always stored in ovens set to the operating temperature so they could be tested more quickly and easily. When the sensors were stored in the customer's warehouse, they got cold and the mismatched tempcos caused stresses to arise in the assembly. The epoxy eventually stress relieved (“cold flowed”) and equilibrated at this temperature, since neither the metal nor ceramic/glass was stressed to anywhere near failure.
When a spare sensor was installed in the system, it was powered up and it reached operating temperature in a few minutes. This rapid temperature rise caused the epoxy (which had a larger tempco than the metal body or the thermistor) to expand from its equilibrium state and exert compressive stress on the glass bead, ceramic pellet, and platinum wires.
Glasses and ceramics are very strong in compression, but platinum is ductile, and the leads were, for lack of a better word, squished. After some hours at the new temperature and under stress, the epoxy would again stress relieve, releasing the compressive stress on the thermistor. This allowed the ceramic to expand away from the now-smaller platinum wires and to lose electrical contact in some areas.
We submitted a design change to replace the rigid epoxy with a much softer one, which would exert very little force on the thermistor due to temperature changes. This solved the problem, and, at least while I was with that company, there were no thermistor failures in sensors with the new epoxy.
This entry was submitted by A. David Boccuti, P.E. and edited by Rob Spiegel.
Dave Boccuti has been a mechanical design engineer for more than 30 years and has worked in a number of different industries. He is currently developing disposable diagnostic medical devices at Seventh Sense Biosystems in Cambridge, Mass.
Tell us your experience in solving a knotty engineering problem. Send to Rob Spiegel for Sherlock Ohms.