LightingTroubleshooting

Coaxial light hotspot on a specular surface: replace the point source

Why a telecentric lens with built-in coaxial light shows a hotspot on mirror-like parts, the tests that confirm it, and the lighting swaps that fix it.

LumaSpec Vision12 min read Share on LinkedIn

On this page 9 sections
  1. 01Why does a coaxial light make a hotspot with no part in view?
  2. 02Is the hotspot a lens or sensor fault?
  3. 03Why do matte parts image evenly under the same lens?
  4. 04Which light keeps defect contrast on a flat specular part?
  5. 05Which light gives a uniform bright field on a specular part?
  6. 06Can software remove the hotspot without changing the light?
  7. 07Which light should each part finish get?
  8. 08How do you prevent the hotspot on the next station?
  9. 09Questions

A telecentric lens with a built-in coaxial light gives one part a clean image and the next part a bright blob in the middle of the frame — sometimes with nothing at all in the field of view. The blob is not dirt on the sensor and not a failed camera: it is the light source imaging itself through the beamsplitter, and a mirror-like part returns it at full strength. This page gives you three bench tests that pin the cause in minutes, the numbers that size the replacement light (emission area against field of view, ring angle against surface finish), and a decision table for matte, flat specular and curved specular parts. The physics is the same whether the coaxial channel is built into the lens or built from a separate light and splitter.

Why does a coaxial light make a hotspot with no part in view?

You are imaging the light source itself: the coaxial channel in a telecentric lens feeds one small LED through a beamsplitter, and the splitter's glass surfaces return a fraction of that light straight to the sensor as a compact bright spot.

The architecture explains it. One LED feeds a beamsplitter inside or in front of the lens; the splitter turns part of the light down the optical axis onto the part and passes the returning light to the sensor. No splitter is perfect: a few percent of the emitter's own light reflects at every glass surface along the way, and the lens images those ghost reflections. With the field empty, the spot you see is the emitter and its ghosts. Put a matte part in focus and its scattered return swamps them; put a polished part there and the mirror returns the emitter's image almost undiminished, focused exactly where the geometry puts it.

A point source therefore suits diffuse-reflective surfaces. On specular material the emitter's own image becomes part of the picture, and no camera setting removes it. The class is easy to recognize: the LS-TCL-2194 is a 0.5x telecentric with a built-in coaxial light, 110 mm working distance, for a 1" sensor.

Telecentric lenses0.5x, WD 110 mm, 1", C-mount telecentric lens with coaxial light (LS-TCL-2194)LS-TCL-2194
Sensor format
1"
Mount
C-mount
Image circle
11-16 mm
Type
Object-side telecentric
$1,539.00View partBuild a system with it

For the jobs where on-axis light is genuinely the right tool — recessed features, bore floors — see In-line illumination in machine vision: when specular parts need it.

Is the hotspot a lens or sensor fault?

Usually neither — three five-minute tests separate illumination geometry from hardware dirt, and a hotspot that vanishes when the light switches off, or when a matte part enters focus, is a lighting problem, not a defect.

Run them in this order.

  1. Kill the light. Switch the coaxial channel off and grab a frame. Spot gone: the image problem is illumination-driven. Spot still there: clean the sensor window and the front element per the lens manual before touching the lighting.
  2. Show it a matte witness. Focus a piece of paper or anodized aluminum at the working distance. If the spot melts into an even gray field, the specular return of the source is confirmed.
  3. Tilt the part. Tilt a polished witness by 2°: the reflected ray direction changes by twice the tilt, so the spot walks across the frame. Tilt 4° and it walks twice as far. A dust speck on the glass stays where it is.

All three pointing at the source means a lighting-geometry problem, and the fixes below are the decision.

Why do matte parts image evenly under the same lens?

A diffuse surface scatters the coaxial light in every direction, so each pixel integrates light from the whole emitter and the uneven angular spread averages out, while a mirror-like surface preserves the emitter's small size and its exact position in the image.

The law of reflection decides it: angle in equals angle out. A small emitter sits at one position, so a mirror-like surface sends its image to one place on the sensor — the hotspot. A rough surface breaks that single ray into thousands of scattered rays, every pixel receives light from the whole emission area, and the unevenness averages away.

Surface finish sets which case you are in. A surface reflects specularly once its RMS roughness drops below roughly λ/8 — about 0.07 µm at 550 nm — and scatters as roughness climbs. Turned and milled finishes at Ra 0.8 µm and above scatter strongly enough to image evenly under a point source; polished, rolled or lapped faces below about 0.1 µm Ra act like mirrors. Bright field vs dark field lighting: the law of reflection decides works the same reflection arithmetic for lighting-angle choice, and choosing a surface finish method for machined parts maps common finishes to the roughness each leaves.

Which light keeps defect contrast on a flat specular part?

Use dark field: a high-angle ring light striking the surface at roughly 60° to 75° from vertical, because the flat mirror throws that light away from the lens and only slope changes — scratches, dents, edges — scatter light back onto a dark background.

Dark field works because flat regions obey the reflection law and throw those steep rays well outside the lens's acceptance cone, while a scratch wall, a dent rim or a machined edge tilts locally and those facets send light back. The background sits at black level and the defect is the only bright thing in the frame, so contrast is set by the defect alone.

Size the ring with two rules: light leaving at 60° or steeper from vertical for polished metal, and an inner clear opening at least as wide as the field-of-view diagonal, so the lens looks through the hole rather than into the LEDs. The center bore stays visible here too: its walls intercept the grazing ring light and scatter it back, which is exactly the signal a recessed edge should give. Dark field answers one question — where does the surface slope? It does not answer where the reflectance differs; print, film or coating contrast on a shiny face needs the next fix.

Bright-field and dark-field lighting geometry High-angle light reflects off the flat surface straight into the lens, so the surface looks bright. Low-angle light reflects away from the lens, so the surface looks dark and only edges or scratches scatter light upward.flat part with a scratchCamerabright field(high angle)dark field (low angle)flat areas reflect away;only the scratch sends light up
Bright-field and dark-field lighting geometry. High-angle light reflects off the flat surface straight into the lens, so the surface looks bright. Low-angle light reflects away from the lens, so the surface looks dark and only edges or scratches scatter light upward.

Why slot angle changes defect detection under coaxial light shows the same slope selectivity at work under an axial light.

Which light gives a uniform bright field on a specular part?

Replace the point source with a large diffuse coaxial light whose emission width is at least about 1.5 times the field-of-view diagonal, so a mirror-like surface images a uniformly lit aperture across the whole frame instead of one small LED. The 1.5 factor covers the light's own edge roll-off.

FOV width = sensor width / magnification FOV diagonal = sqrt(width² + height²) emission width ≥ 1.5 × FOV diagonal

Worked example: the LS-TCL-0980 is a port-type telecentric — 1.5x magnification, coaxial port, 65 mm working distance — on a 1/3" sensor. A standard 1/3" camera has a 4.8 × 3.6 mm active area, so the field is 4.8 / 1.5 = 3.2 mm wide and 3.6 / 1.5 = 2.4 mm high. The diagonal is sqrt(3.2² + 2.4²) = sqrt(10.24 + 5.76) = 4.0 mm, and the emission rule asks for 1.5 × 4.0 = 6 mm across. A 50 x 50 mm coaxial light such as the LS-CXL-0227 clears that eightfold: the mirror now images the flat, diffused center of a 50 mm aperture instead of one LED chip, and the field reads even.

Pixel check: object-space pixel size = pixel pitch / magnification. At a 3.45 µm pitch and 1.5x, that is 2.3 µm on the part, so a 0.05 mm engraved line spans about 22 pixels — comfortable for width and position checks on that mark.

Camera resolution for the smallest feature
Units
Pixels needed
–
Sensor size
– MP
Smallest sensor that fits
–
Pixel on the part
– mm
Pixels across the feature, that sensor
–

Pixels needed = field / feature × pixels per feature on each axis (field incl. margin). The lens must resolve the pixels too: Nyquist = 1000 / (2 × pixel µm) lp/mm.

Find cameras with enough pixels Open the full calculator

Telecentric lenses1.5x telecentric lens, coaxial port, WD 65 mm (LS-TCL-0980)LS-TCL-0980
Sensor format
1/3"
Mount
C-mount
Image circle
Up to 8 mm
Type
Object-side telecentric
$220.00View partBuild a system with it

For larger fields the same rule scales. The LS-CXL-0092 is a 70 x 70 mm coaxial light on 24 V. Against the 1.5 factor it covers field diagonals up to about 47 mm.

Coaxial lights70 x 70 mm coaxial light (LS-CXL-0092)LS-CXL-0092
Size
50-100 mm
Voltage
24 V
$309.00View partBuild a system with it

The port matters because a built-in emitter cannot be enlarged; for this fix choose a lens with a coaxial port and add the large light. Telecentric lens selection: match magnification and image circle sorts the port and magnification options, and Telecentric lens field of view calculation from magnification derives the field arithmetic in full.

A splitter path is lossy: a 50/50 splitter turns about half the emitted light down toward the part and passes half of the return into the lens, so a mirror-like part sends roughly a quarter of the emitted light back to the sensor and a matte part far less. Size the drive current on the return you measure, and keep it stable — constant-current output holds the light fixed as the electronics warm, which is what keeps a shading reference valid. The LS-CTL-0180 is a 4-channel constant-current controller rated 120 W for 24 V lights.

Light controllersConstant current controller, 4 ch, 120 W, 89 x 147 mm, for 24 V lights (LS-CTL-0180)LS-CTL-0180
Output
Constant current
Channels
4 channels
Control
Digital
Feature
Constant current
$359.00View partBuild a system with it

Can software remove the hotspot without changing the light?

Partly: a flat-field correction computed from an averaged empty-field reference divides the fixed unevenness out of every frame, but it cannot recover pixels the glare has saturated, so hold the hotspot's peak below about 90 % of full scale — for example 920 of 1023 levels on a 10-bit camera.

Capture the reference with the field empty at the exact working distance, average ten or more frames to cut sensor noise, and apply the map to every frame after. Re-shoot the reference whenever the light current, the lens, the focus or the part height changes, or the map corrects the wrong thing.

corrected(x, y) = raw(x, y) × mean(reference) / reference(x, y)

The headroom exists because division cannot undo a clipped pixel: where the hotspot pegged the sensor at full scale, the true detail is gone and the corrected image keeps a flat white scar. Set exposure so the spot peaks just under the cap, then apply the map. Shading correction is the fix for an existing station this week; for a new station, size the light by the rule in the previous section.

Which light should each part finish get?

Match the light to the finish and the task: matte parts keep the built-in coaxial channel, flat specular parts take a high-angle ring for defect contrast or a large diffuse coaxial for even bright field, and curved specular parts take a dome.

Part and task Light Why it works
Matte metal or plastic, marks and fine detail Built-in coaxial channel Scattered return hides the emitter's unevenness
Flat specular face, scratches, dents, edges High-angle ring light, dark field Mirror throws the light away; slope changes scatter back
Flat specular face, even bright field, marks under coatings Large diffuse coaxial light covering the field Camera images a uniform aperture, not a point
Curved specular part, overall even view Dome light Wrap-around diffuse source follows the curvature
Deep small bore, floor and lower walls Large diffuse coaxial light, axial path Grazing angles cannot reach down the axis

For curved parts, Dome lights for shiny, curved and wrinkled parts covers dome sizing. For measurement on turned, shiny parts, Telecentric illumination: measuring round shiny parts accurately is the companion read, and Coaxial lighting for engraved characters in recessed metal parts shows the axial-reach case done deliberately.

The decisions, per case: matte part — keep the built-in channel, nothing to fix; flat specular face with a defect call — high-angle ring for dark field; flat specular face needing even bright field — port-type telecentric plus a large diffuse coaxial sized by the 1.5 rule; curved specular part — dome; existing station that cannot be re-lit soon — shading map with the spot held below saturation.

How do you prevent the hotspot on the next station?

Spec the light against the part's finish at quote stage: write the finish class, the lighting geometry and the emission-area rule into the station specification, and prove both with a matte and a polished witness part before the bracket is machined.

Carry these onto the specification sheet:

At quote stage

  • Record the finish class with the part drawing: polished faces (Ra below about 0.1 µm) get dark field or large-area coaxial by default; machined faces (Ra 0.8 µm and up) tolerate the point-source coaxial.
  • Image a matte and a polished witness part under the proposed light before the bracket is machined.

At build stage

  • Size any coaxial light on a specular part by emission width ≥ 1.5 × FOV diagonal, and confirm the splitter inlet clears the lens barrel.
  • Spec defect-call rings at 60° or steeper with an inner opening at least the field diagonal.
  • Hold hotspot peaks below 90 % of full scale wherever a shading map will be applied.
  • Drive lights from constant-current channels and re-shoot the shading reference after any current or optics change.

When you are replacing the light rather than the lens, start from the coaxial lights finder or the lighting configurator, match emission area to the field you computed above, and pick constant-current light controllers so the result holds from shift to shift. The telecentric lenses category filters by magnification, working distance and coaxial port, and the lighting topic page collects the related guides.

Questions

Does dimming the coaxial LED make the hotspot disappear?

No. Dimming scales the spot and the surrounding background by the same factor, so the contrast between them stays constant; the spot is the geometric image of a small emitter, and only a larger emission area, a different lighting angle or a shading map changes it. Dim or strobe for heat and motion blur, not for uniformity.

Will crossed polarizers remove the hotspot?

Not usefully on an on-axis coaxial path. A mirror-like return preserves the light's polarization, so a crossed pair blocks the specular return completely — darkening the whole field, signal included. Crossed polarizers earn their keep with ring or bar lights at an angle, where specular glare leaves at a different polarization mix than the diffuse signal you keep.

Can I keep the built-in coaxial light and add a ring light?

Yes, and the pairing is common on polished parts. The high-angle ring carries defect contrast on the flat face; the built-in coaxial, driven at low current on its own controller channel, reaches a bore floor or recessed base that grazing light cannot. Drive them on separate channels so each can be tuned, strobed or sequenced independently.

Does the hotspot affect measurement accuracy or only image appearance?

It can reach the measurement. Silhouette and edge gauging survives a hotspot parked away from the edge, but surface features — engraved characters, marks, shallow pits — fail when the spot crosses them, because local contrast collapses and saturated pixels carry no gradient to threshold. Measure feature contrast inside the spot area before accepting the image.

Can I stick a diffuser sheet over the coaxial window instead?

Only as a partial measure. A diffuser enlarges the effective emitter and softens the spot's edge, but a diffuser smaller than the field still cannot present a mirror-like part with a uniformly lit aperture the size of the frame, and it eats light you then have to replace with drive current. If the part stays specular, move to the large-area coaxial.