Two weeks ago, our maintenance tech walked into my office with a screenshot from a FLIR One. "This breaker is overheating," he said. Then the contractor arrived with a testo 883 thermal imaging camera and said the same panel was fine. Two thermal images. Two different answers. A replacement breaker already waiting in the stockroom (unfortunately).
I manage purchasing for a small lab and light-manufacturing site—about 60 to 80 orders a year, many of them under $1,000. I'm not an engineer; I report to operations and finance. My job is to make sure the right tool arrives before we need it, and that the invoice matches the purchase order.
The immediate question was: which reading was wrong? After five years of managing this kind of stuff, I've learned that's usually the wrong question. The instruments weren't fighting each other. They were answering different questions.
Why the Readings Didn't Match
Here's the thing: a thermal camera doesn't directly measure temperature. It measures infrared radiation arriving at the detector and then converts that signal into a colorized map. To do the conversion, it has to know emissivity, reflected temperature, distance, and field of view. Change those settings and the same camera can produce two very different numbers.
That's why thermal imaging standards like ASTM E1934 treat thermography as a process, not a photography exercise. The operator is supposed to account for the things that affect the reading. A capable instrument like the testo 883 thermal imaging camera gives you control over those settings. A phone attachment like the FLIR One gives you a lot less.
We assumed they'd agree because they're both "thermal cameras." That was my assumption, and it was wrong. I'm not a thermographer, so I can't walk you through emissivity tables. What I can tell you from a purchasing perspective is that the FLIR One is a quick screening tool and the testo 883 is a decision tool. That doesn't mean one is bad and the other is good. It means they have different jobs.
Take a polished metal panel. Its emissivity can be below 0.3, while a painted wall is often close to 0.90. If the camera assumes the wrong value, the temperature it shows can be off by 10°C or more. The camera also sees reflected energy. The operator standing too close can show up as a hot spot. That's not a camera conspiracy. It's physics.
The same logic follows for other tools. The testo 925 thermometer measures one point in the probe's immediate area. A thermal camera measures a surface map. They can disagree because they're not looking at the same thing. The Fluke 1587 FC insulation multimeter, in insulation test mode, applies a high voltage and measures leakage current. Its IEC 61010-1 rating matters as much as the number on the screen. If the test leads are wrong or the surface is dirty, you get a number that doesn't represent the insulation you meant to test. Even the 5702 centrifuge fits the pattern: it doesn't measure temperature, but if the rotor speed drifts, the samples it produces are wrong and every downstream test looks wrong.
The biggest cause was the one I didn't see coming: a calibration certificate is a snapshot. Here's something vendors won't put on the invoice: the accuracy statement on a new thermometer is a claim about the day it was measured at the factory. It becomes less trustworthy as probes get bent, cables get flexed, and electronics age. If you're not checking your own reference, you're running on faith.
What a Wrong Reading Actually Costs
In the breaker case, the cost was $200 plus a half-day of confusion. In another case, a false insulation reading made us suspect a motor. We replaced it, and the new motor failed the same test. The problem wasn't the motor—it was the test setup. That mistake cost more than the motor. It cost trust, because operations stopped believing our readings.
There's also a hidden cost in the purchasing office. When a measurement argument comes up, I'm the one who has to figure out who pays for the revisit. I used to spend hours trying to decide which supplier's product was to blame. Now I know the product wasn't the issue.
Whenever I see guides titled "how to use FLIR One thermal camera," the focus is on focus and image capture. That's useful, but the bigger lesson is knowing when not to trust it. A quick look is fine. A decision that affects a $5,000 component is not. The same goes for any measurement: the first step is deciding what question the instrument is allowed to answer.
On the supplier side, I know what it's like to be a small order. Early on, I tried to buy a thermometer and one supplier didn't even reply. Another spent fifteen minutes on the phone helping me choose the right probe for the testo 925. That supplier got the order, and they've gotten orders ever since. Small doesn't mean unimportant—it means a chance to earn a customer who will remember you.
We didn't have a formal calibration process. The third time a probe came back from the field with a damaged cable, I added a simple benchmark check to the calendar. I should have done that after the first time.
A Simpler Way to Keep Instruments Honest
You don't need a metrology department. You need a one-page rule set and the discipline to follow it. At our site, the rule set looks like this:
- Agree on a tiebreaker before an argument starts. For thermography, that's the testo 883. For spot temperatures, the testo 925 is the reference. For insulation testing, the Fluke 1587 FC is the tool. The 5702 centrifuge gets its speed checked annually.
- Set a calendar reminder to verify, not just file away the certificate that came in the box. If you can't send the instrument out to an ISO 17025 accredited lab, compare it against a known good unit.
- Match the tool to the decision. The FLIR One is allowed to find a maybe. It is not allowed to make the final call.
Look, I'm not telling you to buy a cart full of expensive instruments. I'm telling you to stop guessing. If two good instruments disagree, define which one is the source of truth before you need it. Then the argument gets a lot shorter.
A measurement you can't trace to a decision is just a number with an opinion.
If you're the one who buys these tools, the best thing you can buy is a clear answer to a simple question: what reading will everyone accept before I spend the money? Once that's settled, the thermal camera, thermometer, multimeter, and centrifuge stop acting like rivals and start acting like a team.
