Rice Lake Resources

The Metrology Toolkit Has Changed. Your Quality Process Probably Hasn't.

Posted on 2026-11-17 by Anika Petrova

Most quality teams are still running a measurement process designed for 2015. The tools changed. The process didn't.

I've been in quality for eleven years. For the first six, my job was straightforward: grab a micrometer, check the dimension, write it down. If it was out of spec, reject the batch. If it was in spec, move on.

That approach worked when we were measuring 200 parts a week and the biggest thing on the floor was a 40-pound machined bracket. It doesn't work now. We're measuring assemblies that weigh two tons, we're doing it on the shop floor instead of a climate-controlled room, and we're being asked to catch problems that don't show up on a caliper at all—like a spindle running 12°C hotter than it should.

Here's my argument: the measurement tools have evolved faster than the quality processes around them. And most teams are paying for it in rework, missed defects, and unnecessary calibration headaches.

Argument 1: The operator variable is bigger than we admit

In 2022, I ran a blind test with three of our most experienced inspectors. Same part, same Starrett micrometer, same reference temperature. I asked each of them to measure the same ten features and write down the readings.

The spread was 0.008 mm on the tightest feature. Not because anyone was careless—because micrometer reading is a skill, and skill has variance. Thimble pressure, parallax, the way you hold the frame. It adds up.

That's not a knock on micrometers. Learning how to read a Starrett micrometer properly is still a foundational skill, and I don't want anyone on my team who can't do it. But foundational doesn't mean sufficient. When you're checking 40 features on a part and the tolerance is ±0.02 mm, operator variance becomes a real cost—not in scrap, but in false rejections and wasted rework.

We started using a FARO CMM arm for the complex geometries about three years ago. The operator variable dropped significantly because the arm captures a point cloud, not a single reading. It's not magic—you still need a competent operator—but the repeatability is better. And repeatability is what lets you trust a pass/fail decision at 4:30 on a Friday.

Argument 2: The measurement has to go to the part, not the other way around

Our stationary CMM is accurate. It's also in a room that stays at 20°C ± 1°C, and the largest part it can accept is about 800 mm on the diagonal.

Half of what we build now is bigger than that. We were cutting parts off the fixture, moving them to the metrology lab, waiting for them to stabilize, measuring, and moving them back. The logistics alone were eating four hours per batch. And every move introduces the risk of handling damage or a fixture shift that invalidates the measurement.

The faro arm cmm changed that workflow. We can measure in place—on the fixture, on the machine, on the truck before it ships. The trade-off is real: a portable arm is not as accurate as a stationary CMM in a controlled room. But for the 80% of checks that are go/no-go against a ±0.05 mm tolerance, it's more than enough. And it catches problems that a lab measurement would miss because the part is already off the machine and the setup is gone.

I'll be honest about the calibration side: portable arms need more frequent verification than a stationary CMM, and the verification process is more involved. We run a monthly check against a certified sphere and a full calibration annually. That's a cost. But the cost of a missed feature on a $40,000 weldment is higher.

Argument 3 (the counterintuitive one): Thermal imaging taught us that measurement isn't only dimensional

This one surprised me.

In early 2024, we started evaluating thermal cameras for predictive maintenance—not quality inspection. The idea was to catch overheating bearings and electrical panels before they failed. We benchmarked a few units, including the e54 advanced thermal imaging camera, mostly on resolution and temperature accuracy.

The unexpected benefit was on the quality side. We pointed the camera at a machining center that had been producing parts with a slight taper—within tolerance, but drifting. The spindle was running hotter than normal. Not alarm-hot, but hot enough to cause thermal growth. We adjusted the cooling, and the taper went away.

No dimensional measurement would have caught that. The parts were in spec. They were just trending. A thermal camera gave us a leading indicator instead of a lagging one.

I'm not saying every quality team needs a thermal camera. But if your measurement process only looks at dimensions, you're missing a whole category of process drift that shows up as heat before it shows up as scrap.

"We didn't have a formal cross-verification process between the portable arm and the stationary CMM. The first time a batch passed on the arm and failed on the CMM, we found out the hard way. Now we run a monthly correlation check on a known artifact. Should have done it from day one."

The pushback I hear—and my response

"Portable arms aren't as accurate as a stationary CMM." True. Per ISO 10360-2, a stationary CMM in a controlled environment will outperform a portable arm on volumetric accuracy. But accuracy without access is useless. If the part can't come to the CMM, the CMM's accuracy doesn't help you. The right question isn't "which is more accurate?" It's "which measurement can I actually perform, reliably, at the point where the decision needs to be made?"

"Thermal cameras are a maintenance tool, not a quality tool." That was my assumption too. I was wrong. Thermal data is process data. If your process is drifting thermally, your dimensions will drift eventually. Catching it earlier is cheaper.

"This sounds expensive." It is. A portable CMM arm and a decent thermal camera are not impulse purchases. But price the rework on a single rejected weldment, or the cost of a missed taper that turns into a field failure, and the math changes. I've also seen teams rent a portable arm for a specific project before committing to a purchase—that's a reasonable way to test the workflow without the capital outlay.

"What about the basics? Calibration, traceability, documentation?" Those don't go away. If anything, they get more important. A portable arm that hasn't been verified against a certified artifact is just a very expensive pointer. The fundamentals haven't changed. The execution has.

What I'd tell a quality manager starting today

Don't throw out your micrometers. Don't sell your stationary CMM. But don't assume that the measurement process you built five years ago still fits the parts you're building now.

Start with one question: where does your measurement process force the part to travel to the tool, instead of the tool traveling to the part? That's your first bottleneck. Then ask: what process drift are you not seeing because you're only measuring dimensions? That's your second.

The tools are better than they were in 2020. The principles—calibration, verification, traceability—are the same. The gap is in the process. Close it deliberately, and you'll spend less time arguing about rejections and more time shipping parts that stay in spec.

I have mixed feelings about how long it took us to figure this out. Part of me wishes we'd started the cross-verification process years earlier. Another part knows we needed the failures to make the case internally. Either way, the lesson is the same: the measurement toolkit evolved. The quality process has to evolve with it.

Anika Petrova

Anika Petrova

Anika Petrova is a nondestructive testing and inspection systems analyst focused on ultrasonic flaw detectors, digital X-ray inspection, industrial metal detectors, needle detectors, and machine-vision inspection. She draws on ASTM E317 and ISO 17636-2 methods when assessing sensitivity, spatial resolution, signal-to-noise ratio, calibration stability, probability of detection, and false-reject rates. Her work helps quality engineers and production managers compare inspection technologies, define validation samples, and set detection thresholds for specific materials and defects.

Leave a Reply