Application scenario

Qualifying DPM reading on machined metal

A low-contrast Data Matrix on machined metal may be readable on a clean, stationary part. On the line, oil film, limited visibility and part position change the reflected light and therefore the ability to decode the mark.

Application: a machined part must retain its identity

This anonymised application scenario concerns Data Matrix reading on machined metal for product genealogy and process records. The machine integrator manages the trigger and result interface. A missing or incorrect identity must have an explicit outcome at the PLC.

No measured production read rate or installed-system performance is reported here. The useful decision is how to qualify the mark, illumination and result handling under the actual line conditions.

Why a clean bench sample is insufficient

A code can decode on a stationary, cleaned part and fail after machining. Oil changes reflections. Tool marks create background texture. Position and angle variation move the reflected light relative to the lens. A fixture or adjacent part may hide part of the code.

First distinguish a no-read from a correctly decoded but unexpected value. A successful decode proves that the reader recovered data; it does not prove that the part has the expected identity or that the marking meets a formal quality grade. Those checks require their own acceptance criteria.

Build a sample matrix from the process

Give every test part an identifier and a known expected code. Retain examples from the permitted marking settings and normal surface variation. Do not clean every part if oily parts are normal in production.

Variable Evidence to collect
Marking Code dimensions, smallest module, marking process and representative variation.
Surface Dry and oily states, tool marks and permitted finish changes.
Presentation Nearest and farthest positions, rotation and tilt, and any obstruction.
Movement Speed at capture, shortest trigger interval and time available for the result.
Data Expected string, allowed format and how the PLC checks the active product or order.
Failure handling No-read, unexpected string, late result, missing trigger and lost communication.

Keep an independent validation set apart from the parts used to adjust lighting and reader settings. Repeatedly reading one favourable sample does not represent the range of production parts.

Compare illumination before committing to a reader

Use the same parts to compare a low-angle light, diffuse illumination and any available multi-directional lighting. Retain images at each setting. Look for separation between code modules and their background without saturating the surface. A brighter image is not necessarily easier to decode.

Check that the chosen lens covers the complete code, including its required surrounding area, at every permitted position. Size the image from the smallest module and the reader’s documented sampling requirements. Verify focus at the distance extremes. Choose continuous or pulsed operation according to the exposure needed for motion and the supported light/controller ratings.

The general traceability and OCR guide covers the wider identification task. This scenario focuses on the surface and qualification test. A reader product label alone does not establish suitability for the mark.

Test the complete PLC decision

Assign each trigger to the physical part being inspected. Record when the camera accepts it, when a new result becomes valid and how the PLC acknowledges the result. Use a sequence identifier or the supported handshake so that an earlier code cannot be reused for the next part.

Test at least a valid expected code, no-read, unexpected code, delayed result and interrupted connection. Define the machine action for each: hold the part, route it for manual identification, reject it or stop the sequence as appropriate to the process. Confirm recovery after a restart and after changing the active product.

Agree acceptance and report the denominator

For each test condition, record the number of attempted reads, correct expected reads, no-reads, unexpected values and late results. Report the corresponding fractions with their denominators. Repeated reads of the same part should be identified separately from the number of distinct parts.

Set limits before tuning the final configuration. Include the required result deadline and the handling of every invalid outcome. If the application requires formal marking verification, specify the applicable standard, equipment and procedure separately from the production decoding test.

Acceptance should also demonstrate that maintenance can restore the approved light position, filters and recipe. Save reference images and repeat the agreed checks after restoration.

Cost inputs and the next trial

Include reader or camera, optics, lighting, controller, mounting, PLC work and validation in the estimate. Track current manual identification time, line interruptions and rework associated with unreadable marks. Estimate any improvement only after the trial; avoid valuing the same interrupted production both as downtime and lost output.

For a feasibility discussion, bring the identified sample set, code dimensions, expected data format, speed, available mounting space and failure-handling requirements. These inputs make it possible to define a test and compare suitable equipment.

Vincent van Montfoort

Talk through your application.

For DPM selection, we assess the most difficult product variants, working distance and line speed. Reader and lighting can then be selected together.

Definitions and selection criteria

Use this scenario to prepare a sample-based qualification test. For the underlying definitions and reader-selection criteria, continue with the DPM guide and code-reading overview.