One-Shot Multi-Dimension Measurement: What 100 Items in a Single Capture Really Means

The POMEAS HM series online image measuring heads are stated to output up to 100 items from a single capture — outer diameter, width, height, corner radii and other drawing-indicated dimensions, plus swing during rotation and workpiece position coordinates. It is a genuine capability and it is also one of the most frequently over-read numbers in this product class. This page sets out what the claim covers, what it does not, and how accuracy actually varies across those items.

What the Number Refers To

It is a statement about how many dimensions the software can evaluate from one silhouette, not a statement that any hundred tolerances can be held to the same accuracy. If your drawing has forty dimensions that are all defined by the outline of the part — distances between edges, diameters, widths, radii on the profile, angles between edges — then one capture can genuinely report all forty. That is the useful reading of the capability, and it is why a projection head can replace a sequence of separate gauging operations.

The Constraint: Everything Must Be on the Silhouette

A transmitted-light head measures the shadow. Any dimension that is defined by the outline is available. Any dimension that is not — a bore diameter, a wall thickness, a step height measured normal to the view, a depth, flatness across a face — is not, regardless of how many items the software can count. This is the single most important boundary, and it is a property of the principle, not of any particular model.

Note the consequence for defect detection, which the same datasheet lists. Defects are detectable when they change the outline: a burr standing off an edge, a chip out of a corner, foreign matter breaking the silhouette. A scratch, a stain or a discolouration entirely inside the outline does not change the shadow and will not be reported. Capability and limit come from the same fact.

Accuracy Is Not Uniform Across the Items

The HM heads publish repeatability and position measurement accuracy per model — for example the HM-1040 at φ40 mm with ±0.08 µm repeatability and ±1.0 µm position measurement accuracy, and the HM-1120 at φ120 mm with ±0.2 µm and ±2.5 µm. Those figures describe the head. Individual dimension results vary around them because of what each dimension depends on:

• A diameter measured across a large, clean span of the field is the best-conditioned measurement available.
• A small radius in a corner is resolved by relatively few pixels and is correspondingly less certain.
• An angle between two short edges amplifies whatever uncertainty sits on each edge.
• A dimension taken near the edge of the field is more sensitive to residual distortion and to pass-line position than one at centre.

So the honest way to use the capability is to identify the two or three dimensions that actually govern function, confirm the head's figures cover those, and treat the rest as informational. Reporting forty numbers where three carry the tolerance creates the illusion of control without the substance. The HM series model selection page gives the per-model figures.

Repeatability Versus Position Accuracy, Again

The two published figures answer different questions and must not be swapped. Repeatability — how tightly repeated readings cluster — governs whether you can detect a small change over time, such as tool wear. Position measurement accuracy — agreement with true size — governs whether a dimension is in tolerance. A dimension that gates a pass/fail decision depends on the accuracy figure.

One further caution on the smallest published figure: the ±0.03 µm repeatability given for the HM-1006 is a manufacturer's specification-table value published without test conditions in the material we hold. Read it as instrument capability under defined conditions, not as a production tolerance. On a line, what you can hold is set by presentation, vibration, temperature and optical cleanliness as much as by the head. Build the full budget rather than quoting the instrument figure — the method is in building an inline diameter measurement error budget.

Where One Shot Beats a Sequence of Gauges

The alternative to a single multi-dimension capture is a sequence of single-purpose checks: a micrometer for one diameter, a height gauge for a length, a ring gauge for a groove. Each is accurate, and each costs handling time, and the results sit in separate places. One capture collapses that into a single event with all the dimensions in one record, which is what makes trend analysis possible at all. The trade-off is that every dimension inherits the conditioning of the silhouette — which is why the governing tolerances have to be checked individually rather than assumed from the head's headline figures. If the job is a bench inspection rather than a line-rate one, an instant one-click system such as IMAGE 3 is usually the better-matched tool; the comparison is set out in inline versus offline optical measurement.

In Transit, Without Stopping

The reason all of this is available at line rate is the 35 µs exposure, which freezes the product so the capture happens in transit. The trade-off is illumination: a very short exposure needs a strong, dedicated source, which is why these heads are supplied as a transmitter-and-receiver pair with a collimated LED rather than as a camera aimed at the line. The mechanics of that are set out in measuring parts in motion at 35 µs.

FAQ

Does one capture really give 100 dimensions?

The manufacturer states up to 100 items from a single capture. It means the software can evaluate that many dimensions from one silhouette — all of which must be defined by the outline. It is not a claim that any hundred tolerances can be held to the same accuracy.

Can it measure a bore or a wall thickness?

No. Those are not on the silhouette. A transmitted-light head measures the outline; internal features need a different principle.

Why do some dimensions come out noisier than others?

Because each depends on different features. A diameter across a long clean span is well conditioned; a small corner radius is resolved by fewer pixels; an angle between short edges amplifies edge uncertainty; dimensions near the field edge are more sensitive to position.

Will it find surface scratches?

No, unless they change the outline. Burrs, chips and foreign matter on an edge do; scratches and staining inside the silhouette do not.

Which published figure governs a pass/fail decision?

Position measurement accuracy, not repeatability. Repeatability tells you whether readings cluster tightly; accuracy tells you whether the dimension agrees with true size.

Is ±0.03 µm achievable on my line?

It is the HM-1006 specification-table repeatability, published without test conditions. Treat it as instrument capability under defined conditions and build your own error budget before writing a control plan.

Does the part have to stop for the capture?

No. The 35 µs exposure freezes the product so the measurement happens in transit.

Next Step

Send the drawing with the dimensions marked, the largest feature that must appear in one field, the tolerances and the line speed. We can tell you which of those dimensions a single capture will return, which will not be available from a silhouette at all, and which HM head covers the field. See also inline diameter measurement methods and online gauge, confocal and vision compared. Start with the HM series selection page.