I was called out to a client's factory to repair a piece of machinery that had mysteriously stopped working. The client's standard maintenance procedures had failed to fix it. The machine welded two vinyl-coated fabric strips together using directed hot air followed by a pair of pinch rollers to force the melted vinyl coating together.

The system used an electric heater (made by Osram/Sylvania) to heat compressed shop air to about 450C. The hot air was then directed by a nozzle into the gap between the two strips of vinyl-coated fabric. The air heater wasn't working at all. When I got there, the machine operator had replaced the heater -- a common operation -- but with no results. The temperature was controlled by a standard PID controller and they had replaced that as well as the thermocouple monitoring safety relay. Still no results.

I checked the heater installation. It seemed OK. I checked the thermocouple circuit. It was an open circuit. Tracing the thermocouple leads from the controller to the heater yielded no results, but the process of elimination suggested the problem was in the thermocouple probe. It was either totally encased in stainless steel, or other refractory metal such as Inconel, or in the cable leading to the probe.

Well, that was easy enough to fix -- just replace the thermocouple probe. In this setup, the thermocouple probe, which is about 1/16 inch in diameter, is inserted into a hole in the side of the air heater barrel, just below the heating element. The tips of the thermocouples have been known to literally burn off in a hot-air environment.

When I pulled the thermocouple out, I ran my finger across the tip, expecting to find the usual rounded end of the welded probe tip. Instead, I found a sharp knife-edged tip. A close look made it plain that the tip of the thermocouple had been cut off with a pair of diagonal cutters. The machine operator said they often cut the thermocouples to length to more easily position them in the airflow. Apparently, in the past, they had managed to mechanically create a new thermocouple junction when the cutter crushed the stainless-steel sheath and the two thermocouple wires (Iron and Constantan; Type J) together. This time, the process failed.

A new thermocouple probe (accompanied by admonitions NOT to cut it to length again) solved the problem. In the past, I've made thermocouple junctions using an oxy-acetylene welding torch, and even soft solder (for low temperatures), but this was a new one.

This entry was submitted by Brooks Lyman and edited by Rob Spiegel.

Tell us your experience in solving a knotty engineering problem. Send examples to Rob Spiegel for Sherlock Ohms.