Batch Dimensional Inspection: Organising a 100% Check That Does Not Stall the Line

A one-shot measuring system removes the per-feature programming that makes a traditional video measuring machine slow, but it does not remove the handling. The real question in batch inspection is not how fast one part is measured — it is how many parts fit in one shot, and what happens between shots.

What a Batch Check Actually Has to Deliver

A batch inspection is only useful if three things line up: the feature set is closed (you know every dimension you must judge), the acceptance limit is written down per feature, and the result is attached to a part or a lot identity. A system that measures beautifully but leaves the operator to write numbers on a sheet has moved the bottleneck, not removed it.

Before choosing any hardware, list the features in three columns: dimensions that decide pass or fail, dimensions that are recorded but do not gate the part, and dimensions that need a different instrument entirely. The third column is where batch projects usually fail — a thread pitch diameter or a surface roughness has no business being on an optical batch sheet.

Fixture Layout Decides the Economics, Not the Camera

The cost per part in a batch check is the cycle time divided by the number of parts per shot. Doubling the parts per shot halves the handling burden per part, which is almost always a larger gain than any camera upgrade.

Three layouts cover most cases. A tray with fixed pockets gives the best repeatability because every part returns to the same coordinates. A free-pallet arrangement is faster to load but needs a locate-and-report step, and any part that sits rotated still has to be recognised as the correct part. A strip or reel feed is the fastest of the three and the least flexible — it only makes sense when the part is already on a carrier for other reasons.

On the IMAGE 3 series the wide-field mode is what makes tray layout practical: IMAGE 3 covers Ø100 mm in one shot, PLUS and PRO reach 200 × 200 mm and 300 × 200 mm respectively. That is the number to divide your tray area by.

Model range: which IMAGE 3 fits this part

Model range, as stated in the IMAGE 3 series specification sheet. Repeatability here is the figure printed in the sheet; it is not a guarantee for your own part, your fixture or your operator. Treat it as the floor you start from when you run a trial, not as the number you can quote to a customer on day one.

The full specification sheet, including the camera and lighting options and the workpiece-to-model mapping, sits on the IMAGE 3 series product page. If you only want a first-pass sanity check on whether your part fits any model, the instant measurement selection tool takes the part envelope and returns the models that can cover it.

Cycle Time: Do the Arithmetic in Four Terms

Write the batch cycle as four terms and measure each one on your own parts. Nothing in a specification sheet gives you this number, because it is dominated by handling and by how many exposures the feature set needs.

• Load and unload — manual, tray swap, or robot. Usually the largest term in a manual cell and the first one worth attacking.
• Capture — one exposure in 2D mode; in height mode it scales with the number of scan positions, not with the number of features.
• Evaluate — edge finding, fitting and tolerance evaluation across all features in the shot. This runs in parallel with nothing, so keep the feature count honest.
• Judge and route — pass/fail display, NG sorting, data write. A stalled data write shows up here.

Divide the total by the parts per shot. Then compare that per-part number against the takt time of the line. If the per-part number is above takt, no amount of accuracy improvement will help — an extra station or a larger field is the answer.

The accuracy figures, with their conditions

L is the measured length in millimetres. The stitched column applies when the feature is longer than one field and the system has to join several images; the single-field column applies when the whole feature sits inside one shot. Read the sheet in this order: which mode, stitched or not, then the length — the three together decide the number.

Sampling or 100%: The Decision Is Financial, Not Technical

A one-shot system can technically inspect every part. Whether it should is a question about the cost of a missed defect against the cost of the extra handling. The comparison that holds up in practice is: parts per period multiplied by the escape rate you accept, multiplied by the cost of one escape, against the incremental cost of inspecting every part instead of a sample.

Two patterns are common. Where the process is already stable and the escape cost is a rework, sampling with a fast response to a drifting mean is usually enough. Where the escape cost is a line stop at the customer, or a safety-related feature, 100% check is the only defensible position and the batch design has to be built around it from the start.

Where a Batch Line Stops Paying Off

Three situations deserve an honest no. First, when the feature set needs more than one measurement principle — a mix of 2D contour and true thread geometry will need two instruments, and a batch cell that pretends otherwise creates a queue at the second one. Second, when parts arrive in random orientation with no practical way to present them in a known pose; the presentation cost then dominates everything. Third, when the tolerance is tighter than the repeatability floor of the chosen model, which for the series runs from ±0.25 µm on MAX and MAX2 up to ±1 µm in wide-field mode on the smaller models.

Acceptance Checklist Before the Line Starts

Run this list once, on real parts, before committing a batch line.

1. Every gating feature has a written limit and a measurement recipe.
2. A repeatability study exists on the actual fixture: same part, twenty loadings, recorded spread per gating feature.
3. The per-part cycle is measured on the real handling, not estimated.
4. NG routing has been exercised end to end, including the data record.
5. A known-bad part is in the sample set and is rejected for the right reason.
6. The operator can tell, without asking, which part the number belongs to.

FAQ

How many parts can a flash measuring machine inspect in one shot for batch inspection?

It is set by the field of view, not by a fixed count. On the IMAGE 3 series the precision-mode field is 18 × 18 mm on IMAGE 3, PLUS, PRO and PRO-H, and 6.4 × 4.8 mm on MAX, while wide-field mode reaches Ø100 mm on IMAGE 3 and 300 × 200 mm on PRO. Divide the tray area by the field and that gives the theoretical count; the practical count is lower because every part needs clearance for edge finding.

Does 100% batch inspection require a robot?

No, but it requires a presentation decision. A fixed-pocket tray gives the most repeatable result with manual loading. A robot or a strip feed only pays off when the per-part handling time is the term that pushes the cycle above takt.

How do I calculate batch inspection cycle time for a one-shot vision system?

Add four measured terms — load and unload, capture, evaluate, judge and route — then divide by the number of parts per shot. Capture is one exposure in 2D mode and scales with scan positions in height mode. None of these can be read from a specification sheet; they have to be timed on your own handling.

Can a batch inspection cell report results into an existing SPC system?

The measurement side produces per-feature values and a pass/fail state; the export format is what decides whether those values can be pushed into an existing SPC package. Agree the field list and the part identity convention before the cell is built, because retrofitting identity into a running line is the expensive version.

What is the practical tolerance floor for batch inspection with a one-shot system?

The stated repeatability runs from ±0.25 µm in precision mode on IMAGE 3 MAX and MAX2 to ±1 µm in wide-field mode on IMAGE 3, PLUS, PRO and PRO-H. A working rule is that the repeatability should be a small fraction of the tolerance band, not a comparable number, otherwise the pass/fail result is dominated by the instrument.

When should a batch line use sampling instead of a full check?

When the process is stable and the cost of an escape is rework rather than a line stop, sampling with fast reaction to mean drift is usually sufficient. When an escape reaches the customer as a stop or touches a safety feature, build for 100% from the beginning instead of adding it later.

Related reading

• What is a flash measuring machine and how does one-click measurement work
• Instant measurement vs CMM: when a one-shot system is enough
• How to read an instant measurement accuracy spec
• Flash Measurement for Connectors
• Flash Measurement for Screws
• Flash Measurement for Stamped Parts

Send Us One Part and We Will Tell You Which Route Fits

The fastest way to close this question is a real part, not a discussion. Send one sample with the drawing and the features you actually gate, and we will run it and report what the measurement returns — including the features that come back as a no. A short written feasibility note is more useful than a brochure.