Replacing a Flash Measurement or VMM Station: What to Verify First

The decision is about the feature, not the machine

Across 2,602 machine vision companies in the POMEAS global intelligence database, Keyence appears in 54 company profiles and Mitutoyo in 7. Both are named in public descriptions of installed dimensional measurement equipment.

That tells you one-press image measurement and conventional coordinate measuring machines are common in production quality control. It does not tell you that one substitutes for the other, and it certainly does not tell you that any specific model replaces another.

Before comparing machines, list what the station actually measures. The list, not the instrument, drives the decision:

- 2D features — lengths, diameters, hole positions, angles, pitch - Height, step, or flatness — anything requiring a Z axis or a non-imaging sensor - 3D profile or roughness — topography, not dimensions

A one-press image measuring machine handles the first group well and cannot address the second or third from the image alone. This article is a verification checklist, not a cross-reference. Nothing here states or implies that a POMEAS machine replaces any third-party model.

1. Field size against part size — the constraint that decides everything else

IMAGE 3 Series is specified with a 300 × 200 mm field. A part larger than the field cannot be measured in one shot: it must be stitched or moved, and both add cycle time and accumulate error.

What to verify: the largest dimension of the part, including any fixture or clamping that must appear in the same image. Then check whether the features you care about fit within the field at the resolution you need — which is the next check and often the binding one.

2. Tolerance against pixel pitch — the number people skip

This is where most measurement projects go wrong. The sampling limit of an image-based system is:

` pixel pitch across the field = field width / horizontal pixel count `

Worked example: a 300 mm field on a 20 MP camera. A 20 MP sensor with a 4:3 aspect is roughly 5,472 × 3,648 pixels. Across the 300 mm width:

` 300 mm / 5,472 ≈ 0.055 mm (55 µm per pixel) `

A tolerance of ±0.02 mm cannot be resolved from that image, regardless of sub-pixel processing. Sub-pixel edge detection improves repeatability between measurements; it does not create resolution that the sampling does not support.

What to verify: the tightest tolerance on the part, then compute the pixel pitch at the field size required. If the tolerance is smaller than roughly three times the pixel pitch, an image-based machine is the wrong tool for that feature — move it to a different station or a different sensor.

3. Whether the part needs height at all

This single question separates the machine classes.

- 2D only — a one-press image measuring machine is the fast option. - Height, step, flatness, or 3D profile — needs a Z axis, a confocal sensor, or an interferometer.

MetX Series is specified as a 2D/3D automatic image measuring machine covering dimensions and contour. Where the station must output height as well as in-plane dimensions, that is the relevant class.

For height specifically, POMEAS publishes a spectral confocal line:

The pattern is the one that matters: range and resolution trade against each other. A 3 nm repeatability figure comes with a ±0.05 mm range. Specifying 3 nm over a 2 mm travel is not available in this line and is not a realistic request of any single confocal sensor.

What to verify: the height range you must cover, then the resolution available at that range — not the best resolution in the catalogue.

4. Throughput: parts per hour, not seconds per part

A one-press machine measures many features in a single shot. A conventional VMM moves between features. For a part with many 2D features, the one-press advantage is large; for a part with two features and a height requirement, it largely disappears.

What to verify: the full cycle including loading, fixturing, measurement and unloading. The measurement step is often not the bottleneck.

5. Measurement on the line instead of in the lab

If the part must be measured while it moves, a benchtop machine is the wrong class entirely.

LM-1030 is specified as a non-contact inline diameter gauge covering 0.3–30 mm. The relevant check for an inline sensor is different: sampling rate against line speed.

` required samples per second = line speed / spacing between samples `

A line running at 10 m/s sampled every 1 mm needs 10,000 samples per second. Verify the sensor's stated rate against that number before anything else.

What to verify: line speed, required sample spacing, and the sensor's published sampling rate.

6. Output format and SPC integration

The machine must produce what the quality system consumes: pass/fail per feature, actual values, Cp/Cpk, or a formatted report. A machine that measures correctly but exports in the wrong format creates manual work and transcription errors.

What to verify: the report format the quality system requires, and whether the machine produces it directly.

Where the POMEAS measurement line sits on these six

Real application scenarios

A stamped part with 40 dimensions. All features are 2D, the part fits a 300 × 200 mm field, and tolerances are ±0.1 mm. At roughly 55 µm per pixel, the sampling supports the tolerance with margin. A one-press machine measures all 40 features in one exposure.

A connector housing with a coplanarity requirement. The in-plane dimensions are easy; the coplanarity of the terminations is a height measurement. The check is the height range and the resolution needed — a confocal sensor at the appropriate range, added to or replacing the vision-only station.

A cable or tube diameter on the production line. The part is moving. The check is sampling rate against line speed, and the diameter range against the gauge specification — 0.3–30 mm for LM-1030.

What this checklist does not tell you

- It does not state that any POMEAS machine replaces any third-party machine. That requires both sets of specifications and a measurement on the actual part. - The pixel pitch calculation is a sampling limit, not a measurement accuracy. Actual accuracy depends on calibration, illumination, edge quality and thermal conditions. Do not quote 55 µm as an accuracy figure. - Third-party specifications are not reproduced. Nothing in this article states a specification for Keyence or Mitutoyo.

---

FAQ

Is one-press measurement always faster?

For parts that fit the field and features that are 2D, yes, and the advantage grows with the number of features. For a part with two features and a height requirement, the advantage largely disappears.

Can a one-press machine measure height?

Not from the image alone. Height, step and flatness require a confocal sensor, a probe, or a Z axis with an appropriate sensor.

What limits the smallest feature a one-press machine can measure?

The pixel pitch across the full field. A 300 mm field on a 20 MP camera gives roughly 55 µm per pixel — a useful sampling limit, not an accuracy figure.

Why would I still buy a VMM?

When the part needs height, step, flatness, or 3D profile, or when the tolerance is tighter than the pixel pitch supports at the required field size.

What about measuring parts on the line rather than in the lab?

Use a dedicated inline gauge. Verify the sampling rate against line speed first; the measurement range second.

Does sub-pixel processing make the pixel pitch irrelevant?

No. Sub-pixel edge detection improves repeatability between repeated measurements. It does not create resolution the sampling does not support.

---

Next Step

• Flash Measuring Machine vs Profile Projector
• Evaluating a Keyence IM Series Alternative: A Parameter Checklist

Send the part dimensions, the feature types (2D dimension, height, flatness), the tolerance and the required throughput. POMEAS will recommend the configuration and state the sampling limit for your field and camera combination.