The call
In March 2024, at 1:57 in the morning, my phone lit up with a text from the night operator: "Reactor pressure reading is dead. We need you." I've worked in industrial instrumentation long enough to know that "need you" at 2 a.m. isn't optional. I got dressed, skipped coffee (that's a lie—I grabbed the travel mug), and headed to the plant.
I'm an electrical and instrumentation lead at a mid-size specialty chemical operation. In my role coordinating emergency repairs, I've handled roughly 200 unplanned callouts in the past ten years. That experience has taught me a simple truth: instrument failures are rarely the real problem. The real problem is the time around the failure.
The component in question appeared to be an Ametek - US Gauge transducers/sensors unit on a reactor vessel. It wasn't new—it was industrial-old—and we'd seen intermittent readings in the weeks before. I still kick myself for not replacing it during the last planned shutdown. A scheduled swap would have taken an hour. Instead, we had about eight hours before a batch tied to a customer delivery.
The cross-check
When I reached the control room, the operator had already checked the basics. Reactor temperature was stable. Tank level made sense. And the plant's Coriolis flow meter was still reading flow normally. A Coriolis flow meter measures mass flow directly, so it doesn't depend on a pressure transmitter to calculate flow. That made it a useful cross-check. The process was fine. The instrument loop was not.
Out at the instrument rack, the transducer indicator showed no output. No mV across the sensor, no loop current. I did the standard things: unhooked the loop, tested the wiring with a handheld meter, looked for broken wire. Nothing. The supply side had power, but the 4-20 mA output was dead. I was already building a case for rush replacement, and a new part from the Ametek - US Gauge transducers/sensors line would normally take a couple of weeks through standard channels. We didn't have a couple of weeks.
What is a FLIR thermal camera?
Then a young electrician asked, "Did you check the supply terminals?" I hadn't. It was an older panel, and the connections were tight and crowded. That's when I grabbed the FLIR thermal camera. What is a FLIR thermal camera? It's essentially a handheld infrared camera that shows heat patterns. In electrical work, it's a safe way to find a loose connection before you put a finger anywhere.
I scanned the 24 V terminal strip and saw one hot spot immediately. The terminal had high contact resistance and was running warm. We powered down, reseated the wire, torqued the terminal properly, and powered back up. The "dead" transmitter woke up and started sending a clean 4-20 mA signal.
We still replaced that sensor during the next planned outage. Not because it failed—it didn't, that night, at least. I replaced it because when an old sensor gets you that close to a missed deadline, the cost of certainty is much lower than the cost of a second callout. In March, the customer shipment was worth about $70,000, and missing it would have cost far more than the sensor. That kind of consequence changes how you think about spare parts.
The replacement
For the replacement, I spent an hour before ordering. I checked the Ametek official site first. If you're not used to buying process instruments, that might sound obvious, but in an emergency people tend to take the first equivalent part a distributor offers. The Ametek official site gave me the current model, datasheet, pressure range, output options, and a list of local distributors. According to the Ametek official site, the US Gauge transducers/sensors line includes pressure transducers and transmitters for industrial applications, with options designed for severe service. More important than the specs: every distributor I called could reference the same part number and the same lead time. That shared documentation is a form of quality. It's what makes a supply chain predictable.
I still remember an old forum thread titled "Klein vs multimeter" that used to come up in our shop. The underlying question was simple: can a hand-tool brand with a multimeter be trusted next to brands that only make test instruments? My answer after years of emergency callouts isn't about logos. It's about documentation and support. If you're stuck at 2 a.m., you want a vendor who can answer the phone and a part number that exists. Quality perception is built by those small, unexciting things, not by brand hype.
Lessons from a 2 a.m. outage
First, don't rely on one measurement. The Coriolis flow meter gave us confidence that the process was still safe while we worked on the loop. A Coriolis flow meter measures mass flow by sensing the effect of flow on vibrating tubes, so it doesn't share the same failure mode as a pressure-based reading. That independence is exactly what you want in a cross-check.
Second, quality is not how expensive a component looks. It's whether you can get the right spare fast, prove it's correct, and install it before a deadline. A cheap spare that's unavailable is worse than a trusted name with a real support chain. When I look back at our 2 a.m. outage, the deciding factor was certainty: the certainty that the next part would be the right part.
Third, your customers judge your company by what they don't see. The batch shipped on time. The customer never knew anything happened. That's the real meaning of quality perception: in industrial B2B, your reputation is mostly built by avoiding failures that never show up on an invoice.
I'm not 100% sure why that terminal worked loose—thermal cycling is my best guess—but now every panel inspection includes a thermal scan and a torque check. Honestly, I should have done that long before the call. That said, my experience is based on one plant, one night, and one particular mix of older equipment. If you're in a different industry, your cross-checks and vendor list will look different. The principle stays the same: buy from a company you can verify, keep a good spare, and check the measurements that still work when one instrument goes quiet.
