Specify Machine Vision Camera Lighting for Mixed-Product Inspection Lines

17/11/2026

A scratch that is obvious on a matte part can disappear in glare after a changeover to a polished finish. In Deloitte’s 2025 Smart Manufacturing and Operations Survey, 28% of respondents at large manufacturers with US headquarters or operations ranked vision systems among their top two investment priorities [1]. For APAC plants evaluating that investment, the purchase decision starts with a line trial that includes real product variation.

Machine vision camera lighting is the controlled illumination used to make inspection features visible to a camera. On a mixed-product line, specify it alongside the lens, capture geometry and trigger timing so defects stay visible across approved products and line speeds.

Key Takeaways

Start with the smallest rejectable defect and the hardest surface to image. Set the field of view and focus range, then test lighting and exposure on moving parts from representative lots. Approve the combinations of product, position, finish and speed that meet defect detection, false reject and throughput limits.

1. Start from the defect requirement

Specify each rejectable defect by type, size, location and decision rule before selecting a camera or light. Give quality and engineering a shared set of accepted, rejected and borderline parts so they evaluate the same inspection decision.

For a molded housing, define where a scratch matters, which surface marks are acceptable and how inspectors judge a borderline sample. Record the smallest rejectable feature in physical units. Add examples from approved finishes and suppliers, including normal texture or tooling marks that can resemble defects.

Build the initial test set around the hardest legitimate conditions. A shallow scratch near a curved edge may need a different view from a missing clip or an out-of-tolerance hole. Use the defect list to decide which features can share a view and which need a second camera or illumination angle. That decision shapes both station layout and cycle time.

2. Specify the capture geometry

Set the field of view, working distance and allowable part position together. The camera must cover the full inspection area and resolve the smallest required defect at the limits of the part’s position and height [2] [3].

Estimate object-space sampling by dividing the field width by the number of active image pixels across that width. Divide the defect width by that sampling value to estimate how many pixels span the feature. Treat this calculation as a starting point. Lens performance, lighting and focus determine whether those pixels show enough contrast to distinguish the defect [3].

Check the tallest and shortest products, maximum lateral offset and far corners of the image. Use a calibrated target to verify scale and a real part to confirm visibility. If the widest product forces a field of view that loses detail on a smaller one, test a tighter second view or a dedicated station. Record the mount position and fixture tolerances so technicians can reproduce the setup after maintenance.

3. Test machine vision camera lighting and material variation

Choose lighting by how it renders the defect and the surrounding surface. Compare candidate setups using representative finishes, part orientations and supplier lots, then retain the images that explain why each setup passed or failed [4] [5].

Lighting approachStart withCheck during trials
BacklightSilhouettes, holes and outer dimensionsEdge spread on translucent parts
Low-angle dark fieldScratches and raised featuresScratch orientation and distracting texture
Directional bright fieldPrinting and surface contrastGlare as the part tilts
Diffuse domeCurved, glossy surfacesWhether diffusion hides shallow flaws
Coaxial bright fieldFlat, reflective surfacesLoss of contrast when surfaces tilt or curve

Hold exposure and gain steady during each comparison so camera adjustments cannot mask lighting differences. Test ambient light from doors and nearby equipment, then repeat with shielding in place. If one light cannot reveal both the surface defect and the silhouette, test sequential captures and include the extra exposure in the cycle-time budget.

4. Account for product movement

Set exposure from the blur the defect can tolerate at maximum line speed, then verify trigger position and image-to-part tracking. A sharp frame still fails the inspection if the system associates it with the wrong part [6] [7].

Estimate blur distance by multiplying the part’s speed across the object plane by the effective illuminated exposure time. Divide by the object-space distance per pixel to express blur in pixels. Validate that budget on the smallest rejectable feature. A short synchronized strobe can help freeze motion if it delivers sufficient light within the exposure window [6] [7].

Use a part sensor or encoder to trigger capture. Where speed varies, test position-based timing and confirm the reject mechanism tracks the correct part downstream. Run at maximum speed and minimum part spacing, then test acceleration, stop and restart, missed triggers and late decisions. Confirm how the line handles a part with no valid image or decision.

5. Approve an imaging envelope

An imaging envelope defines the products and operating conditions under which inspection meets agreed acceptance criteria. Record the approved ranges for finish, position, working distance and speed, along with the recipe revision and the action required when a condition falls outside those limits.

Deployment pipeline

  1. Define acceptance. Use the quality defect catalog and boundary samples to set limits for missed defects, false rejects and throughput.
  2. Qualify optics. Use a calibrated target and production samples to verify focus, scale and defect contrast at the extremes of each recipe.
  3. Prove line timing. Use controller event logs and captured frames to reconcile triggers, decisions and physical rejects.
  4. Release the recipe. Use role-based access and version control to approve changes, retain evidence and restore an approved configuration.

TMA’s company history records its achievement of CMMI Level 5 and ISO/IEC 27001 [8]. Apply engineering and security controls by keeping test evidence traceable to each released recipe and restricting access to recipe changes and stored images.

Lessons learned from the field

TMA Innovation’s flexible printed circuit board example addresses microscopic defects that require fine image detail. Its undated Defect Detection On Electronic Board page describes high-resolution capture and reports an inspection time of six seconds per board [9]. The page gives a system-level result. The contribution of the lighting setup remains unquantified.

For mixed-product procurement, ask vendors to show which images and recipe settings produced each acceptance result. A handoff with unversioned exposure settings or screenshots from a single product can leave plant engineers unable to reproduce a failure after a changeover.

FAQ

How much camera resolution does the inspection need?
Start with the smallest rejectable defect and the full field width. Estimate how many pixels span the feature, then verify contrast and focus on real parts at the least favorable position [3].

Can one lighting setup cover every SKU?
Yes, if the same setup reveals every required defect across approved finishes and positions. Use separate recipes, illumination angles or views when the trial shows that one setup cannot meet those criteria.

When should I use strobed lighting?
Use a synchronized strobe when continuous light cannot provide enough signal within the exposure needed to control blur. Verify pulse timing, thermal limits and image consistency at the highest trigger rate [6] [7].

How should we handle a new finish or supplier?
Treat it as a change to the imaging envelope. Test accepted and defective samples under the approved recipes and line conditions before releasing it to automatic inspection.

How do I evaluate an inspection vendor?
Request a witnessed trial with borderline defects, multiple finishes and full-speed changeovers. Require raw images, per-SKU results and a demonstration that plant staff can restore the approved recipe.

Conclusion

Approve the station against the products and conditions it will actually see. Start with the hardest defect, least forgiving finish and fastest approved line speed. Share representative parts, defect samples and line-speed data with TMA Innovation to scope a camera and lighting feasibility trial [9].

References

[1] Deloitte Insights, 2025 Smart Manufacturing and Operations Survey: Navigating challenges to implementation, 2025. Survey

[2] Edmund Optics, Imaging Fundamentals, undated. Application note

[3] Edmund Optics, Resolution, undated. Application note

[4] Edmund Optics, The W of Illumination Geometry, undated. Imaging Lab

[5] Edmund Optics, Silhouetting Illumination in Machine Vision, undated. Application note

[6] Basler, Optimizing Image Quality, undated. Product documentation

[7] Basler, Triggered Image Acquisition, undated. Product documentation

[8] TMA Tech Group, Giới thiệu TMA Tech Group, undated. Company history

[9] TMA Innovation, Defect Detection On Electronic Board, undated. Solution page

 

TMA Innovation
Author: TMA Innovation
Table Of Content
1. Start from the defect requirement
2. Specify the capture geometry
3. Test machine vision camera lighting and material variation
4. Account for product movement
5. Approve an imaging envelope
Deployment pipeline
Lessons learned from the field
FAQ
Conclusion
References
Start your projectRight now!
Contact Us
Share:
Machine Vision Camera Lighting for Mixed-Product Lines