Your CMM Calibration Isn't Working: A 5-Step Verification Checklist

Look, I'm not saying your CMM calibration is bad. I'm saying the sticker on the machine doesn't tell you what you need to know. As a quality manager who signs off on roughly 200+ measurement deliverables every year, I've learned that the gap between 'calibrated' and 'delivering correct results' is bigger than most people assume.

This checklist is for the person who needs to trust their Hexagon CMM data—whether you're running first-article inspections, PPAP samples, or daily production checks. It's not a full calibration manual. It's the verification list I wish I'd had before my team wasted 80 hours on bad data that looked fine.

Six steps. Do them in order. Don't skip step three.

Step 1: Verify the Machine Geometry (Not Just the Sticker)

From the outside, a calibration sticker looks like a guarantee. The reality is calibration covers a specific range at a specific moment. A machine that passed last week might be drifting today—especially if the shop floor temperature fluctuates.

What to do: Run a quick geometry check using a certified reference sphere or a step gauge. Measure it in five different positions across the volume. If the variation exceeds 50% of your required tolerance, stop and call in a service.

In our Q1 2024 audit, we caught a 12-micron drift on a Hexagon CMM that was 'in calibration.' The variance was within the calibration spec but way too large for the parts we were measuring. (Note to self: always check against actual part tolerance, not the machine spec.)

Step 2: Check Your Hexagon CMM Probes—All of Them

People assume the probe is the least likely thing to fail. What they don't see is how easily a probe tip gets nicked or bent—especially with a heavy stylus configuration. I've rejected entire probe setups because of a 0.5-micron scratch that looked fine to the operator.

What to do: Qualify each probe and stylus combination. Compare the results to the original qualification data. If any deviation exceeds 0.002mm, replace the tip. Then requalify.

This sounds obvious. But here's the thing: most teams requalify only when they see bad results. By then, you've already measured 20 parts wrong. Better to do it before you start.

Step 3: The Thermal Drift Test (90% of People Skip This)

This is the step nobody talks about. I went back and forth between including it here and keeping the list short. The truth is: skipping this is the #1 reason calibration data doesn't match production data.

What to do: Measure a stable reference artifact (like a gauge block) at the start of your shift, after lunch, and at the end of the day. Record the results. If the variation correlates with room temperature changes (say, a 2-degree swing), your thermal compensation settings need updating.

Calculated the worst case: re-measuring 150 parts because thermal drift caused a 0.01mm shift. Best case: you catch it early and adjust the compensation. The expected value says it's worth checking once a quarter.

Step 4: Verify Your Software Settings and Filtering

Honestly, this is where most failures hide. The machine can be perfectly calibrated, but if your software has the wrong filter settings (like outlier removal or stylus radius compensation), the data will be wrong. Period.

What to do: Open your measurement plan. Check the filter settings. Are they appropriate for the part geometry? (Think: a tight filter on a rough surface will give you a false high reading). Then run a known part and compare to a reference measurement (like from a CMM with different software).

In our Q3 audit, we found a 0.008mm difference between two machines on the same part. The cause? The filter was set to 'standard' on one and 'high precision' on the other. A simple setting change fixed it.

Step 5: Acceptance Criteria—Know What 'Good Enough' Means

The upside of a thorough check is confidence. The risk is spending too much time chasing perfection. I kept asking myself: is getting to 0.001mm worth potentially delaying the production line? Sometimes the answer is no.

What to do: Define clear acceptance criteria for each check. For a 0.1mm tolerance part, a 0.005mm probe deviation is fine. For a 0.02mm tolerance part, that's a deal-breaker. Document your criteria and stick to them.

The vendor who lists all fees upfront—even if the total looks higher—usually costs less in the end. Same principle here: define the criteria upfront, even if it seems conservative. It saves rework later.

Common Pitfalls to Avoid

• Don't trust the sticker. A calibration certificate is a snapshot. Use it as a baseline, not a guarantee.
• Don't skip probe qualification. Even if you're using the same probe from yesterday. Temperature changes affect it.
• Don't assume software settings carry over. If you update software (like PC-DMIS or QUINDOS), verify settings. I've seen migration scripts drop filter parameters.

That's it. Six steps. From the outside, it looks like extra work. The reality is it's less work than remeasuring a batch of parts you thought were good.

And between you and me, the teams that follow this checklist rarely need rework (note to self: track our rework rate this quarter).