A dot peen Data Matrix on polished steel or aluminum is one of the hardest codes to light. The flat face that carries the mark mirrors the lamp straight into the lens, and coaxial or direct lighting drowns the code in glare. This guide gives you the angles, distances, pixel counts and pulse times that make dot peen marks read reliably, with the arithmetic worked through. You will place bar lights at a fixed elevation above the surface, size the camera in pixels per module, strobe to your line speed, and know when dark field is the wrong tool. If the marking method itself is still open, this guide to laser, inkjet and dot peen part marking explains how each mark type catches light.
Why does dark field lighting make dot peen marks readable?
Light arriving 15° to 30° above the surface reflects off the flat metal away from the camera, while the slanted walls of each indentation scatter a fraction back to the sensor, so the Data Matrix reads as bright dots on a near-black field.
The flat face around the mark behaves like a mirror: light that arrives at a shallow angle leaves at the same angle on the far side and never reaches the lens — the principle that separates bright field versus dark field lighting. The mark breaks the mirror. Each dot is a small crater with a raised rim, and those slanted faces send light in every direction, including back to the camera. Contrast comes from geometry, not lamp power, which is why lighting geometry: camera, light and part positions decides the result before any intensity setting does. Depth helps too: a deeper dot has taller walls and scatters more, so a faint-looking mark often reads well at low angle.
Where do the bar lights sit around the part?
Put each bar face 15° to 30° above the part plane and aim it across the field: with the bar offset a distance d from the field centre, the height above the surface is h = d × tan(θ), so an 80 mm offset at 25° puts the bar about 37 mm up.
Four bars in a square arrangement give even coverage from every side; start with two opposing bars when space or budget is short and add the second pair when edge falloff or part tilt shows up. Pick each bar longer than the field's wide axis by about a quarter, so the far edge stays lit. If a verifier grades the codes downstream, DPM grading standards illuminate at 30°, 45° or 90° elevation, or through a dome, from one, two or four sides — keeping the read lighting close to the grade lighting makes a line pass predict a verification pass. The light finder filters bar lights by lit length, colour and wavelength. Bars to consider for the array, all 24 V:
- Size
- 200-400 mm
- Type
- Standard
- Size
- 200-400 mm
- Type
- Standard
- Size
- 200-400 mm
- Type
- Standard
How many pixels per module does the camera need?
Budget 3 to 5 pixels across each module: a 0.5 mm module then needs 0.10 to 0.17 mm per pixel on the part, and a 40 mm field needs only 235 to 400 pixels across, so any industrial reader has resolution to spare — the real limits are focus and contrast.
Worked example, end to end. The part carries a 12 × 12 module symbol at 0.5 mm pitch, 6 mm square, but the locating edge also has to be in frame, so set the field to 40 mm wide. Pick 4 pixels per module: pixel scale = 0.5 / 4 = 0.125 mm on the part. Pixels needed = 40 / 0.125 = 320. Take a reader with a 1/1.7-inch sensor, about 7.2 mm wide and 2560 pixels across: magnification m = 7.2 / 40 = 0.18. With an 8 mm lens, working distance = f × (1 + m) / m = 8 × 1.18 / 0.18 ≈ 52 mm. Each pixel then spans 40 / 2560 ≈ 0.016 mm on the part, so the module spans 0.5 / 0.016 ≈ 32 pixels — far above the floor, and that surplus absorbs blur and focus drift. The camera resolution calculator runs the same chain the other way, from field to minimum pixels.
The LS-CRD-0028 delivers exactly this geometry: a 2560 x 2048 sensor in 1/1.7-inch format with an 8 mm lens at 30 fps. If the code sits alone on the face, a compact reader that focuses at 70 mm — the MV-IDB005EXI-05NRP — tightens the whole cell. Browse fixed code readers by sensor format, shutter and interface. The readers below match this station:
- Interface
- Fast Ethernet
- Resolution
- 5 MP
- Interface
- Fast Ethernet
- Resolution
- Up to 3 MP
How short must the exposure be on a moving line?
Limit motion blur to a quarter of a module: at 300 mm/s belt speed with a 0.5 mm module pitch that means about 0.4 ms, and only a pulsed light holds useful brightness in a window that short.
A static station can run the bars continuously because nothing moves. A line cannot: a 0.4 ms pulse at rated current is far too dim, so a lighting controller fires the bars at higher pulse current and keeps average duty low, which also keeps the LEDs cool — the core of why a lighting controller earns its place. Watch the switching edge as well: a pulse that arrives late or ramps slowly smears the image exactly like a long exposure, which is the trap the LED strobe rise time breakdown describes.
Small marks need less hardware. The LS-SPT-0175 is a 6 mm high-luminance spot driven at 5 W:
- Size
- Under 50 mm
Aimed across the face from one side at the same elevation as a bar, a single spot light covers a mark a few modules wide and leaves room for a compact fixture.
How do you set the station up step by step?
Square the camera first, then fix the pixel scale and focus, then build the light around the part: camera axis within about 3° of the code normal, field and focus from the calculation above, bars at the chosen elevation, pulse width from the blur budget.
- Compute the field, magnification and working distance from the module size, as in the worked example above.
- Mark a sacrificial part and square the reader to the code face: if one edge of the field is sharp while the opposite edge blurs, the axis is tilted — fix that before anything else.
- Count pixels across 10 modules in the live image and divide by 10; 5 or more is comfortable, below 3 means redo the optics.
- Mount two bars in opposing positions at your elevation — 25°, offset 80 mm, height 37 mm in the example — using h = d × tan(θ) for your own dimensions.
- Wire the bars to a lighting controller and set the pulse width from the blur budget, 0.4 ms in the example.
- Raise pulse current until the flat background sits under about 10% of full scale and dot peaks near 70%, then trim per part.
- Rotate the part 90° twice and require clean reads at all three orientations before accepting the layout.
- Run a 200-part trial with your DPM grading standard as referee; if any part misses, fix lighting or mark quality before touching decoder settings.
When does dark field fail, and what do you switch to?
Dark field fails when the whole surface scatters instead of mirrors — sandblasted, cast or matte-machined faces stay bright from every angle — and on curved or deeply recessed faces no bar grazes uniformly; switch to diffuse bright field with a polarizer, or to a dome.
A dot peen mark on anodized aluminum is the classic bright-field case: the dots are bare, shiny metal inside a dark matte coat, so a diffuse bright field shows the code with the polarity flipped. Where glare still leaks in, cross polarization strips the specular component. Curved or wrapped marks call for dome lights, and the reasoning behind reading QR codes on reflective and curved surfaces carries straight over to Data Matrix. Small cylindrical parts behave differently again — the shiny metal pins guide covers dome, diffuse and polarizer choices at that scale.
For the wider DPM toolkit on flat metal — coaxial, low angle and direct combines — the DPM Data Matrix lighting guide is the companion piece. When the image already looks good but decodes still drop, the fault is usually the mark, not the light: recovering decodes on damaged, low-contrast codes separates the two.
Size the field and pixels, place the bars with h = d × tan(θ), pulse to the blur budget, and the mark reads all day. The lighting configurator turns the geometry into an order and the camera finder narrows readers by resolution and interface; for a four-bar layout around a new part, send the drawing through the quote form and we will check the numbers with you.
Questions
What light colour reads dot peen marks best?
White or blue works well on clean machined steel, and blue gives crisper dot edges because the shorter wavelength scatters strongly off the fine walls. Move to 850 or 940 nm infrared when a coolant or oil film covers the mark, since longer wavelengths pass the film with less loss, and confirm the reader's sensor keeps enough sensitivity there.
Can a code reader's built-in light replace external bar lights?
For a small mark on a dull or textured face, often yes — integrated direct illumination delivers enough contrast at close range. On polished flat metal it usually floods the field with its own reflection: the histogram shows a bright background with weak dot peaks. That signature is the cue to move the light off-axis to low-angle bars.
How do I judge whether a dot peen mark is good enough?
Grade the mark instead of eyeballing it. AS9132, a common aerospace DPM standard, accepts dot sizes from 65% to 105% of the nominal module, dot centres within 20% of the ideal grid position and grid distortion up to 6°. Bring a sample into those windows at the marking station, then watch decode confidence drift on the line.
How much quiet zone does a dot peen Data Matrix need?
Keep at least one module of clean, unmarked metal around the symbol, the ECC 200 rule, and more where a burr rises at the mark boundary. Dot peen burrs and clamp scars creep into the quiet zone, so de-burr or shift the mark before adjusting the lighting.
How steady must the camera and the part be?
Treat 6° of grid distortion as the total geometric budget a grading standard like AS9132 allows, and spend it deliberately: mark distortion takes some, so hold camera tilt to about 3° and leave the mark half the budget. Part wobble also swings the bar incidence angle and dims the dots, which is why verification stations fixture the part.