Reducing Probe Lobing Error on Curved Features (Without Turning Your CMM into a Guessing Machine)

If you’ve ever measured a bore, a sphere, or a nice smooth radius and thought:

“Why does this ‘perfect’ curve look slightly… three-lobed?”

You’re not imagining things. That pattern often comes from lobing error—a direction-dependent trigger behavior common in kinematic (mechanical-switch) touch-trigger probes. In plain terms: the probe doesn’t trigger at the exact same deflection in every direction, so the measured surface can come out with a subtle “triangular / three-lobe” signature.

This blog is about reducing lobing specifically on curved features (bores, cylinders, spheres, fillets, large radii)—where the combination of surface normal changes + probe trigger direction changes makes lobing show up loud and clear.

---

What lobing really is (and why curved features expose it)

The core mechanism

Kinematic touch-trigger probes use a mechanical switching mechanism. The trigger force required can vary with probing direction, creating small form errors commonly called “lobing.”

Why curves “magnify” lobing

On a flat plane, you can often approach with a single consistent vector and get away with it.

On a curved feature:

• every touch point has a different surface normal
• your approach vector can vary unintentionally
• stylus bending and pre-travel behavior change with direction

And because pre-travel is affected by trigger direction, trigger speed, and stylus length/slenderness, your curve becomes a perfect canvas for lobing to paint on.

---

The “DeepMind” idea: lobing is not random noise—it’s a repeatable shape

Here’s the mindset shift that beats competitors’ generic advice:

Lobing is often a stable, direction-dependent signature.
That means you can either:

1. avoid changing the trigger direction, so the error stays consistent and cancels out (or at least doesn’t create form), or
2. model/compensate the probe’s directional behavior, so the system corrects it.

Most shops do neither—they just “take more points” and hope.

More points can actually make a lobing pattern look more believable.

---

Step 1: Confirm it’s lobing (not something else)

Before you “fix lobing,” do a quick sanity check:

• Re-measure the same curve with the same program and compare results. If the shape is repeatable, it’s likely systematic (lobing/pre-travel), not random.
• If your feature form looks like a stable multi-lobe pattern, that aligns with direction-dependent kinematic probe behavior described in literature.

If it’s not repeatable, suspect:

• loose stylus, mounting issues, collision damage, contamination, re-seat problems (those won’t create a consistent “lobing signature”)

---

Step 2: Reduce the root cause (best ROI fixes)

1) Slow down touch speed (yes, really)

Higher trigger speed can increase pre-travel, and pre-travel varies with direction—exactly what lobing feeds on.

Shop rule:
For critical curved features, use a slower, consistent touch speed than your general inspection speed.

2) Shorten and stiffen the stylus stack

Longer or slender styli increase pre-travel and stylus deflection effects.

Practical moves:

• use the shortest stylushttps://cnc-probe.com/cnc-probes-stylus/ that clears the part
• reduce extension bars
• avoid heavy stylus clusters unless needed

3) Consider probe technology if lobing is a recurring pain

Renishaw explicitly describes that “standard probes” using a mechanical switch can show lobing, and that strain-sensing probes (e.g., TP200) are designed to overcome that direction-variation issue. https://cnc-probe.com/cnc-touch-probes/

You don’t have to replace everything—but if your business lives on tight form of bores/spheres, this is one of the few changes that can reduce lobing at the source.

---

Step 3: Programming tactics for curved features (where most wins happen)

Tactic A: Keep the approach vector consistent whenever possible

If you always approach a surface with the same probe orientation, you reduce how much the probe’s directional behavior changes between points—meaning less “lobed” form in the fit. This is common real-world practice among CMM/CNC probing users. https://cnc-probe.com/cnc-touch-probes/

How to apply this on curves:

• For a bore/cylinder, prefer a strategy where the probe approaches points using consistent head orientation and controlled approach vectors (rather than “whatever vector the software chooses”).
• For a sphere, avoid mixing radically different approach directions unless you’re intentionally mapping the behavior.

Tactic B: Use more head angles—but only if you qualify correctly

Rotating the head to keep probing direction favorable can help, but only if your qualification matches the way you measure.

ISO-style probing performance tests explicitly involve probing in multiple orientations on a sphere (because direction matters).

Translation for the shop floor:
If you measure a bore with 3–5 different head angles, qualify those angles properly and verify with a sphere test program that resembles your measurement pattern.

Tactic C: Don’t “spray-and-pray” points around a curve

Taking many points evenly around a bore can actually reconstruct the lobing shape beautifully.

Instead:

• Use enough points for stability, but focus on consistent probing physics
• If you need dense data for form, consider scanning strategies (if available) or controlled multi-orientation probing with validation

---

Step 4: Qualification and compensation (the grown-up solution)

Qualify like you measure (not like the training slide)

Your metrology software needs accurate probe tip calibration/qualification to know tip location and diameter before measurement.

But “qualified” doesn’t mean “lobing-proof.”

What actually helps:

• Qualify the probe tips and orientations you will use for the curve
• Use qualification patterns that represent the directions you will probe in real programs

Build an “error map” if you must hit ultra-tight form

There’s established research on touch-trigger probe error compensation using generalized probe error models and even neural-network approaches—reporting significant error reduction in some cases.

In the real world, you can do a simplified version:

1. Measure a high-quality reference sphere with your actual stylus and program style
2. Look at the directional deviations (your “signature”)
3. Use software features or post-processing (where validated) to correct systematic direction bias

This is more effort than slowing down and shortening the stylus—but it’s how you win when you’re chasing form in the single-digit microns.

---

A simple shop SOP to reduce lobing on bores/radii (copy this)

When a curved feature form matters (bores, spheres, radii):

1. Touch speed: set a slower consistent speed (don’t mix speeds inside the same feature)
2. Stylus: shortest, stiffest configuration that clears the feature
3. Approach vectors: keep direction/orientation consistent whenever possible
4. Head angles: if using multiple orientations, qualify those orientations and validate with a sphere routine
5. Verification: run a reference-sphere check using the same probing style to see whether lobing shrinks or just changes shape

---

The biggest trap: “We fixed lobing” vs “We moved lobing”

It’s possible to change:

• the phase of the lobes
• the amplitude
• where the worst deviation shows up

…without actually improving truth.

That’s why ISO-style thinking treats probing performance as intertwined with the CMM system and relies on test artifacts like spheres and multi-orientation probing for verification.

In practice:
Any lobing “fix” should be confirmed by a repeatable artifact test and a before/after comparison.