By Gus S. Calabrese
While working on an installation at Florida Nuclear Power Plant, I ran into a problem with a constantly failing intelligent multiplexer (IMUX) that my company had installed. There were a number of these IMUXes being installed. My company was upgrading the security system at the power plant. The multiplexers handled a variety of tasks. They supported a ring network which kept working given a single point failure. They reported their health. They connected with secure door controllers, surveillance camera controllers, perimeter intrusion detection systems and more.
None of the other IMUXes were failing. This particular IMUX failed every few days. Usually the failed part was one of the modems for the network. The network used a pair of shielded wires to create a two-way RS-422 point to point network. This was in the 1980s, a time before there were all the network choices designers now have.
There was a lot of lightning at the site. The plant had installed a massive grid of buried copper to provide a ground plane for the entire plant. Even though the grid was monstrous, it was still possible to have the grid “tilt” 100 volts during a lightning strike. The guys working with me put their heads together and tried to figure out what was happening. Lightning seemed out since the failure occurred at one IMUX and was not correlated with lightning strikes. Swapping boards and modems did not help. This one particular IMUX persisted in failing.
One day I happened to walk into the area where the quixotic IMUX was located. I shielded my eyes because welders were busily creating arcs. Then I noticed that one welder had attached his grounding clamp to the IMUX stainless steel enclosure. WHOA! I started to laugh. When the welder stopped for a moment, I asked him if he had clamped to the IMUX box before. “Yes,” he said. Every few days, he had to weld new items in place on the steel walkway next to the IMUX. He felt safer making his ground to the convenient IMUX box. Apparently the IMUX modems were not up to dealing with the current spikes the welder produced. After arranging for the welder to use a different grounding point, the IMUX quit failing.
Gus S. Calabrese works with WFT Electronics.