A wide inspection image is useful only when the smallest required feature remains visible everywhere within it. A bright centre cannot compensate for dark corners, and a visually even background does not guarantee reliable defect contrast. For engineers specifying a conveyor or stationary inspection station, the practical task is to balance coverage, exposure time and lighting direction. Start with measurable image requirements, then position and test bar lights against those requirements before choosing hardware.
Define the inspection area and acceptance limits
Large field of view lighting has no universal size threshold. Treat the field as large when available mounting positions make adequate illumination difficult across the complete inspection area.
Record the field width and height at the actual inspection plane. Include placement tolerance, part rotation and any features near the image boundary. Distinguish camera working distance from light-to-part distance: they are separate dimensions, and changing one does not necessarily change the other.
For a hypothetical inspection, the initial requirements might be:
- Inspection area: 800 × 600 mm, including placement allowance.
- Smallest relevant feature: 2 mm across.
- Required sampling: at least 5 pixels across that feature.
- Maximum conveyor speed: 800 mm/s.
- Allowed motion blur: no more than 1 pixel.
- Initial illumination-uniformity target: minimum patch intensity divided by maximum patch intensity of at least 0.80.
The uniformity target is an application starting point, not an industry acceptance standard. Your inspection algorithm may tolerate more background variation or demand substantially less.
Begin with bar lights when adjustable direction and distributed placement are useful. The lighting finder helps narrow physical formats, but dimensions alone do not establish coverage.
Calculate sampling and exposure before demanding brightness
Lighting must provide enough signal within the exposure that motion permits. Increasing exposure until the image looks bright can hide an unsuitable illumination design.
For the example above, calculate the minimum horizontal sampling requirement:
This establishes sampling only. Lens resolution, focus, contrast and the inspection algorithm still determine whether the feature is distinguishable.
Suppose the selected image has 2500 horizontal pixels across the inspection area. The corresponding object-space sampling and exposure limit are:
An exposure of 300 microseconds would produce approximately 0.24 mm of travel, or 0.75 pixel of blur. These calculations assume motion parallel to the inspection plane and uniform image sampling for the estimate.
Use the exposure and moving-part blur explanation when setting the final acquisition limit. Keep exposure and gain fixed during lighting comparisons; otherwise automatic adjustments can disguise declining edge illumination.
Estimate coverage without assuming a universal distance law
A beam must reach the inspection boundary before brightness there can be evaluated. For an idealized source aimed perpendicular to a flat plane, a preliminary beam-width estimate is:
For an assumed full beam angle of 70 degrees and a perpendicular distance of 600 mm:
That estimate only suggests geometric reach. It does not establish useful brightness at the boundary, and the beam-angle definition may describe a substantial intensity drop. Tilted mounting changes the footprint, while an extended bar requires considering its emitting length and beam distribution together.
Do not interpret an estimated 840 mm footprint as proof of uniform illumination across an 800 mm field. Request an irradiance or illuminance map at the intended distance, ideally including the measurement conditions and beam-width definition.
For an ideal point source illuminating a surface at normal incidence, irradiance follows an inverse-square distance relationship. Doubling the distance would reduce irradiance to one quarter. Nearby extended bars, lenses and diffusers need not follow that simple relationship over your working range.
Position bars to fill weak regions, not reinforce the centre
Start with an adjustable mounting arrangement. Move and aim each light independently while watching both the reference surface and representative parts.
A single bar can be sufficient when its measured distribution covers the full area. Opposing bars are useful when one side is weak or lighting direction makes features disappear. More lights are not automatically better: overlapping beams can create an even brighter central region.
| Arrangement | Useful when | Main limitation to test |
|---|---|---|
| One bar aimed across the field | Access is limited and one direction gives adequate contrast | Far-edge brightness and directional shadows |
| Opposing bars aimed toward weaker regions | One-sided falloff or shadows limit inspection | Central overlap and changing reflections |
| Multiple bars with overlapping coverage | A broad field exceeds one useful beam footprint | Bright bands and channel-to-channel imbalance |
| Diffused illumination | Small hotspots or harsh angular changes dominate | Reduced signal and possible loss of defect contrast |
Aim each beam so its stronger region contributes where the combined image is weakest. Adjust position before using unequal controller settings to compensate for geometry.
For example, record separate images with each bar active. If both produce their brightest patch in the centre, adding their outputs will probably preserve that central bias. Re-aiming may help more than increasing total power.
The LS-BAR-0567 is listed as a standard bar light and is a candidate for a format review.
- Size
- 400 mm +
- Type
- Standard
- Voltage
- 24 V
The LS-BAR-1168 is listed as a diffused red bar light and provides a different format to assess.
- Size
- 400 mm +
- Type
- Diffused
- Voltage
- 24 V
These are not a matched photometric comparison. Confirm the required colour, beam distribution and mounting envelope separately. The LED colour and inspection contrast guide explains why a brighter-looking image is not always the more useful inspection image.
Measure uniformity and inspection contrast separately
Use a flat, matte reference that fills the inspection plane. It should have reasonably consistent reflectance at the chosen wavelength. A glossy sheet can measure reflection geometry rather than illumination uniformity.
Capture images with fixed exposure, gain, aperture and processing settings. Avoid clipping. Prefer linear image data, and account for the camera's black-level offset before comparing intensity ratios.
Divide the reference image into a 5 × 5 grid. Measure the mean or median within each patch, excluding borders and contamination. This gives 25 regional measurements rather than allowing a single noisy pixel to define the outcome.
Define a simple uniformity metric:
Suppose the initial patch measurements range from 88 to 160 image levels after offset correction:
After re-aiming the bars and reducing the central overlap, suppose the range becomes 124 to 148:
These hypothetical measurements pass the example's initial uniformity target. They do not yet demonstrate successful inspection.
Move the same representative good and defective parts through the centre, edges and corners. For positive, offset-corrected measurements, one useful local contrast measure is:
A centre background of 148 and a feature intensity of 110 give 38 / 258, or approximately 0.15. At an edge, a background of 124 and a feature intensity of 112 give 12 / 236, or approximately 0.05.
The reference looks reasonably uniform, yet the feature contrast is much weaker at the edge. Reflection angle, local shading or part geometry may explain the difference. Repeat measurements across multiple frames to check noise and stability before setting a pass threshold.
The lighting and dynamic-thresholding guide is relevant when background variation affects detection. Threshold adaptation should not substitute for a feature that is optically absent.
Use diffusion and accessories only for a measured problem
Diffusion can spread angular illumination and soften localized bright regions. It also redistributes light, so the camera may receive less useful signal at the permitted exposure. Some surface features depend on directional illumination and become less visible after diffusion.
Compare the original and diffused arrangements at the same exposure and gain. Record reference uniformity, minimum feature contrast and the brightest pixel region. Then decide whether any improvement justifies the signal loss.
Do not assume that a diffuser plate fits a light because their nominal dimensions seem similar. Confirm its intended mounting method, optical material, temperature limits and compatibility.
Use lighting accessories and the accessories finder to identify candidates for a compatibility check. Extension cables also need confirmation of connector, pin assignment, electrical rating and permitted length. Mechanical mating alone does not establish electrical compatibility.
Add synchronized strobes when exposure is the constraint
A synchronized pulse can concentrate illumination into the acquisition window and reduce unnecessary operation between images. Pulsing does not automatically increase peak brightness. Higher peak output requires documented permission from both the light and controller.
For the earlier exposure example, suppose there is one 300-microsecond pulse per image at 50 images per second:
This is an operating example, not a permitted rating for any listed component. Check pulse width, repetition rate, peak current, thermal limits and total output limits against the selected hardware documentation.
The LS-CTL-0070 is listed as a digital control box with a strobe feature.
- Output
- Constant voltage 24 V
- Channels
- 4 channels
- Control
- Digital
- Feature
- Strobe
That feature alone does not establish support for the example timing, overdrive or a particular light. Review light controllers through the controller finder and confirm the complete electrical pairing.
Verify that the pulse falls within the camera's valid exposure interval. Rolling-shutter acquisition can require different timing from simultaneous exposure across the image. A mistimed pulse may create partial illumination or inconsistent brightness.
Use the strobe timing and duty-cycle guide for the operating checks, and the hardware trigger configuration guide when the acquisition event is not detected.
Recognize when the usual answer is wrong
A wider field does not automatically require stopping down the lens. A flat inspection plane perpendicular to the optical axis does not need extra depth of field merely because its corners have longer geometric ray paths. Part height variation, plane tilt and lens performance are the relevant checks.
If edges remain blurred under adequate illumination, review lens selection for inspection before purchasing more lighting. Check focus, resolution across the field and aperture performance independently.
Other cases also need a different response:
- Bright reference, weak defect: change illumination direction or wavelength before increasing intensity.
- Good uniformity, clipped reflections: control reflection geometry; global exposure reduction may hide darker features.
- Dark band appearing only during motion: investigate shutter and pulse timing before adding another bar.
- Stable static image, unstable production image: check ambient-light changes, vibration, part presentation and automatic camera adjustments.
- Corners respond differently despite equal reference brightness: test representative surface orientations rather than relying on the flat target alone.
Measure ambient contribution with the inspection lights off and on, using identical camera settings. Keep both images below clipping. Their difference estimates the added inspection-light signal when the camera response is linear and the scene is unchanged.
Complete the pre-order and acceptance checklist
Before committing to a lighting arrangement, collect a short evidence package:
- Coverage: field dimensions, part-height range, mounting positions and light-to-plane distances.
- Sampling: smallest feature size and required pixels across it.
- Exposure: maximum speed, permitted blur and fixed acquisition settings.
- Uniformity: patch measurements across the full reference plane, including corners.
- Contrast: representative good and defective parts at every critical location and orientation.
- Timing: verified exposure overlap, pulse frequency and documented operating limits.
- Compatibility: controller output type, channel loading, connectors, cables and diffuser mounting.
- Repeatability: measurements after warm-up and during expected ambient-light changes.
Save the images alongside settings and light positions so another adjustment can be compared against the same baseline. Accept the arrangement only when the weakest inspection location meets the task requirement without clipping elsewhere.
Start the lighting configuration with those requirements, or send the measurement package through applications support. A defined exposure budget, field map and minimum defect contrast make the selection more reliable than choosing by light length or headline brightness alone.
Questions
Does a bar light need to be longer than the field of view?
Not necessarily. Emitting length, beam distribution, mounting distance and aiming all affect useful coverage. A longer bar can still leave weak corners or create a central bright region. Request distribution data at the intended distance, then test representative parts across the complete inspection area.
Should I move the light farther away or add another bar?
Move the light farther away when coverage or relative uniformity improves enough to justify the reduced signal. Add or reposition a bar when a specific region needs illumination or directional shadows limit detection. Compare both options at the same exposure, gain and inspection conditions.
What edge-to-centre lighting uniformity is acceptable?
There is no universal acceptable ratio. Begin with a measurable reference-plane target, then validate the inspection at the weakest location. Feature contrast, noise, clipping and false decisions matter more than background uniformity alone. Some algorithms tolerate gradual shading but fail when a reflection removes the required feature.
Can any bar light run brighter when strobed?
No. Pulsing can synchronize illumination with acquisition without increasing peak output. Higher peak current is permissible only when the light and controller documentation explicitly supports it at the required pulse width and repetition rate. Verify electrical compatibility, timing and thermal limits before operation.
Why do defects disappear at the edges of an evenly lit image?
The reference target measures illumination on one surface, while actual parts can reflect light differently with position and orientation. Repeat the same good and defective parts across the image. If contrast changes, adjust direction, diffusion or wavelength rather than assuming that a flatter background will solve detection.