DPM Direct Part Marking Reading Guide: Reliable Reads on Metal and Plastic Parts

Automotive parts, machined hardware, PCBs and medical instruments increasingly carry their identifier on the part itself — no label to peel. Reading a DPM (Direct Part Marking) code reliably is a system problem: marking method, illumination, optics and decode software all have to line up. This guide follows that chain.

An industrial fixed-mount reader at a production station · a fixed station and steady takt come first

1. What DPM Is: Four Marking Methods

What the surface looks like depends on how the mark was made. Four methods cover most production:

Laser marking

The laser ablates or alters the surface, leaving pits or colour-changed dots; consistent and fast.

Dot peen

A hard stylus punches pits tens to a hundred microns deep; tough, corrosion-resistant, common on castings and forgings.

Inkjet

Ink is jetted onto the surface and reads by ink-to-substrate contrast; cheap, but solvent or abrasion removes it.

Electrochemical etching

An electrolyte and stencil etch a conductive surface; flat edges, even depth, typical on bearings and round parts.

2. Why DPM Is Hard to Read

A DPM code is part of the material, and the problems stack:

• Low contrast: laser and dot-peen marks image by geometry, not colour, so pits and surface look almost identical.
• Curved surfaces: on cylinders and cones the cells stretch, turning the Data Matrix grid trapezoidal or arced.
• Specular glare: polished stainless and chrome bounce light into the lens and blow out part of the code.
• Depth coupled to light direction: the same dot-peen code reads well under grazing light from one side and vanishes when the light moves.
• DPM-specific distortion: uneven cells, burrs and crowded quiet zones make general-purpose decoders fail.

An AI-decoding industrial module built for low-contrast and damaged codes

3. Hardware: Illumination Comes First

With DPM, fix the light before the camera:

• Low-angle / dark-field: light skims the surface so pits cast shadows, manufacturing the geometric contrast — first choice for dot peen.
• Coaxial: suits flat, colour-changed laser marks; on mirror-like parts it returns as a bright patch.
• Multi-angle / dome: several directions even out shadows on curved faces, the practical option for complex shapes.
• Polarisation: cuts the specular component and keeps the diffuse signal.

Balance the optics too:

• Depth of field and working distance: part height varies under the takt, so it must cover the whole marking zone.
• Resolution and field of view: each smallest cell needs several pixels, sized from the marking plus location tolerance.
• Exposure and fusion: short exposures stop glare blowing out the code; fusing frames recovers weak signal.

4. Software: Decoding Built for DPM

General decoders assume clean cells and plenty of contrast — what DPM withholds. A DPM pipeline usually does three things: image enhancement (denoising, local contrast stretching, adaptive thresholding), geometric correction (grid fitting, distortion correction, quiet-zone location), and decode algorithms tuned for dot patterns and low contrast, with an AI model on some options to rescue poor images. Enhancement serves the decoder: stretching noise along with the signal smears the code just as badly.

5. Validation: Grades Plus First-Time Read Rate

“It scans” is not enough; production needs repeatability:

• ANSI/ISO grade: printed codes are graded under ISO/IEC 15415, DPM under ISO/IEC 29158 (the DPM grade), scoring contrast, cell modulation and axial uniformity.
• First-time read rate: reading at a fixed station and takt and counting the share that succeed on the first attempt — how well the system works at that station.

Sample different batches, orientations and surface conditions, not one golden part.

6. Troubleshooting Order When a Code Will Not Read

1. Light first: move to low-angle or multi-angle illumination, add polarisation on reflective parts.
2. Then the mark: magnify it — enough contrast, intact quiet zone, no missing cells.
3. Then the optics: working distance, depth of field and field of view against the real mark size.
4. Then exposure: fix exposure and add frame fusion so auto-exposure stops drifting.
5. Only then the decoder: enhancement strength, threshold, algorithm profile.
6. If it still fails, revisit the process: marking depth, cell size, surface finish, mark position.

Solve the physical problem first instead of asking software to compensate for the optics.

Embedded engines and fixed-mount readers side by side on a production station

7. Marking Method, Material, Difficulty and Recommended Setup

Marking method	Typical materials	Main reading difficulty	Recommended setup
Laser marking	Stainless steel, aluminium, hard plastics	Low contrast, uneven colour change, specular glare	Coaxial or multi-angle light + polarisation, short exposure
Dot peen	Castings, forgings, carbon steel	Varying depth, burrs, directional shadows	Low-angle dark-field light, multi-directional fill
Inkjet	Plastics, anodised surfaces, packaging	Fading, solvent removal, dot spread	Diffuse or ring light; guard against contamination
Electrochemical etching	Bearing steel, tooling, round metal parts	Curved-surface distortion, low contrast	Multi-angle light + grid distortion correction