A single ring light commits you to one emission angle, and one angle hides as many defects as it shows. When the same part must be checked for missing components inside holes, print faults and fine scratches, three separate fixtures cost mounting space, extra cameras and extra triggers. Stacked ring lights solve this at one station: two or three tiers at different angles sit around one camera axis, and a controller switches them between captures. Here you get the angle-to-defect mapping, the formula that sizes the mounting height, a worked example with the arithmetic shown, and the sequencing details that keep the extra images fast.
What a stacked ring light is
A stacked ring light is two or more ring-light tiers mounted concentrically on the optical axis, each tier a full ring of LEDs with its own emission angle, and a clear window down the middle for the camera. The camera, lens and part position never move. A multi-channel light controller fires one tier per image, so every part returns a steep-angle image, a middle-angle image and a grazing image in one trigger cycle.
This is not photometric stereo. The photometric stereo ring lighting for surface inspection method captures several angles and fuses them into one shaded image. Here each image stays separate and feeds its own inspection tool, which keeps every pass simple to tune. The camera-side pattern is the same one described in sequential machine vision lighting for multi-view inspection: one trigger, several images, each with its own illumination.
What each emission angle reveals
Machine-vision ring lights state the emission angle from the part surface, the convention the category filters use: 90° shines straight down, 30-60° is oblique, and 0-15° grazes the face. The angle, more than raw brightness, decides which defects an image can show at all, because it decides which surface orientations can return light to the lens.
| Tier | Angle from surface | What the light does at the surface | Defect classes it reveals |
|---|---|---|---|
| Steep | 60-90° | Flat faces reflect straight back into the lens; walls and bottoms of holes catch it | Missing terminals in bores, presence and absence checks, print contrast |
| Middle | 30-60° | Small height differences throw shadows across the face | Hole edge shape, contamination, roughness, embossed characters |
| Low | 0-15° | Flat areas reflect the light away; tilted facets scatter it back to the lens | Scratches, particles, burrs, cracks |
A steep tier is the internal-feature pass. Light travelling nearly straight down reaches into recesses and blind holes, so a missing terminal turns into a dark hole where its neighbours shine. Flat top faces reflect it straight back, which gives a bright, even background that threshold tools like. The surface-by-surface reasoning is in lighting geometry for image contrast.
A middle tier converts height into contrast. Every edge, contamination speck or rough patch tilts the surface away from the mirror angle, so relief appears as shading while flat areas stay uniform. This is the pass that finds hole-edge burrs, print misalignment and thin contamination films.
A low tier is dark-field illumination: the flat face carries the grazing light across the camera, and only a scratch, particle or burr scatters light back up. Scratches read as bright lines on a dark field, at a contrast the steep pass cannot produce.
Sizing the stack: the height formula
h_min is the minimum mounting height above the part face, r_i is the tier's inner radius and θ is the emission angle measured from the part surface. A tier's cone converges toward the axis as it descends. Mounted at or above h_min, the cone crosses the axis before it reaches the part, so the whole face is lit at roughly the emission angle.
Mounted lower, the cone has not converged yet, and a dark disc of radius r_dark stays in the middle. A 90° tier is the special case: it lights a disc about as wide as its own inner diameter at any height, so for that tier the inner diameter, not the mounting height, decides coverage.
Worked example, with the arithmetic:
- Part: a 40 mm diameter cap with a flat top, so the face radius is 20 mm.
- Stack: inner diameter 48 mm, so r_i = 24 mm, mounted at h = 45 mm above the face.
- 60° tier: h_min = 24 × 1.732 = 41.6 mm. At 45 mm it clears by 3.4 mm, so the full face is lit at about 60°.
- 30° tier: h_min = 24 × 0.577 = 13.9 mm. Covered with room to spare.
- 15° tier: h_min = 24 × 0.268 = 6.4 mm. Covered, so even the grazing tier bathes the centre.
Now the failure check: mount the same stack at h = 25 mm. The 60° tier gives r_dark = 24 - 25 / 1.732 = 24 - 14.4 = 9.6 mm, a 19 mm dark disc in the middle of a 40 mm part. That is the right answer if you inspect an annulus, and the wrong one for a full-face bright-field pass.
The example stack is sized around a ring in exactly this class, the LS-RNG-0730: 103 mm outer diameter, 48 mm inner diameter. Recalculate h_min with your own inner radius when you adapt the example to your part.
- Size
- 100-200 mm
- Type
- Standard
- Angle
- Low angle (0-15°), 50°, 65°, 70°, 75°, 80°, 85°
- Voltage
- 24 V
Matching the camera to the smallest defect
Every tier's image still has to resolve its defects. Suppose the finest scratch that matters is 0.10 mm and the part plus placement tolerance needs a 50 mm field of view. A 5 MP sensor with 2448 pixels across gives 50 / 2448 = 0.0204 mm per pixel, so the scratch spans 0.10 / 0.0204 = 4.9 pixels, above the 3-pixel minimum for a reliable tool. At 0.05 mm the same scratch would span 2.5 pixels, and you would need a larger sensor or a tighter field of view instead.
Switching the tiers in the inspection cycle
One controller channel drives one tier. The camera's strobe output or the PLC fires channel 1, waits for the frame, then fires channel 2, then channel 3. Set the sensor exposure equal to the strobe width, so almost no ambient light accumulates between pulses; the ambient light rejection specification article shows how to put that into an acceptance test. On a moving line, lock the pulses to part position as described in moving part photometric stereo: synchronize image capture.
Timing arithmetic for three tiers: each capture costs the strobe width plus the camera frame time. With 1.0 ms strobes and a typical 45 ms frame for a 5 MP GigE camera, three captures take 3 × (1 + 45) = 138 ms, about 0.14 s, or 14% of a one-second cycle. If that is too much, drop to two tiers, reduce resolution or move to a faster interface.
The LS-CTL-0322 is a 4-channel digital LED light controller with 8 A total output across 16-24 V, so three tiers leave one channel spare. Check the total against the sum of the tier ratings.
- Output
- Constant voltage 24 V
- Channels
- 4 channels
- Control
- Digital
If the vision process should set patterns over a serial link, the LS-CTL-0070 is a 4-channel, 200 W digital control box with RS232 and a strobe feature.
- Output
- Constant voltage 24 V
- Channels
- 4 channels
- Control
- Digital
- Feature
- Strobe
Heat is rarely the limit. Average LED power equals peak power times duty, and 1 ms pulses at two parts per second across three tiers give 3 × 2 × 1 ms = 6 ms per second, a 0.6% duty cycle; a tier peaking at 10 W then averages 10 × 0.006 = 0.06 W. The real limit is each channel's peak current, which is why the LED strobe duty cycle rules matter the moment you want pulsed overdrive for short, bright flashes.
Each tier needs its own cable with matched connectors and current rating; the light controller to light cable adapters article covers the pairing. A 5 m light extension cable such as the LS-LAC-0060 reaches a stand-mounted stack from a panel-mounted controller, and longer runs add voltage drop on 24 V lights, so keep the controller near the station. Spares, diffuser plates and mounts live in the lighting accessories category.
Choosing the angles for your defect list
Map every defect class to a tier before ordering anything:
- Features inside holes, slots and recesses: steep tier at 70-90°, diffused so bore walls light evenly.
- Edges, embossing, contamination and roughness: middle tier at 30-60°.
- Scratches, particles and burrs on flat faces: low tier at 0-15°.
- Relief and internal features in one image: fire the steep and middle channels together and accept the contrast mixing.
Three 120 mm, 24 V rings at three angles form one fixture about 120 mm across:
- Size
- 100-200 mm
- Type
- Diffused
- Size
- 100-200 mm
- Type
- Standard
- Type
- Standard
- Angle
- 70°
The LS-RNG-0100 emits at 60° in 620 nm red, a wavelength that often raises contrast on warm-coloured plastics; how colour changes monochrome contrast is covered in light colour and monochrome camera contrast. The LS-RNG-0810 emits at 70°. A 70° tier over a 24 mm inner radius needs h = 24 × 2.747 = 66 mm to cover the full face, so steep tiers push the stack higher: budget stand height for the steepest tier, not the average one. Running different colours per tier also works, as multi wavelength light: sequence colours explains.
Limits: when a stacked ring is the wrong fixture
A stacked ring suits parts that are roughly flat and presented face-up. On shiny, curved parts the tiers just rearrange the glare, and a dome light is the right tool instead; see dome lights for shiny, curved and wrinkled parts. On mirror-flat faces a steep tier reflects the ring straight into the lens; a cross polarizer solves that as in cross polarization machine vision: reduce surface glare, and the saved light must be paid back in exposure or strobe current.
Deep, narrow bores also defeat the steep tier: light entering at 70-90° cannot reach a bottom deeper than the bore is wide, and that job belongs to a coaxial light that folds the illumination into the lens axis. And keep the speed cost in view: every tier is one more image per part, so a four-tier stack on a fast line needs the frame-time arithmetic above, or a fused single-image method instead.
Checks before ordering
- Measure the part's largest cross-section and add placement tolerance: the stack's inner diameter must clear it with margin, since a 90° tier lights about its own inner diameter.
- Compute h_min for the steepest tier with h_min = r_i × tan(θ), and check the stand height against the lens working distance so focus and lighting fit together.
- Count channels: one per tier plus one spare, and sum the tier currents against the controller's total output.
- Confirm the trigger path, camera strobe output or PLC, and set exposure equal to pulse width.
- Plan one cable per tier with matched connectors and current rating.
- After commissioning, re-check tier balance on a machine vision reference target for comparing illumination so a dimming channel is caught before tools fail.
To size the whole chain, start in the ring light finder, then match channels and current in the light controller finder, or step through the lighting configurator. For a stack with custom sizes or a fourth tier, describe the part and the defect list through the contact page.
Questions
What does the emission angle of a ring light measure?
The angle between the emitted beam and the part surface: 90° shines straight down onto the face, 30-60° is oblique and 0-15° grazes the surface as dark-field illumination. The angle matters more than raw brightness, because it decides which surface orientations can return light to the camera and therefore which defects are visible at all.
How much cycle time do three stacked tiers add per part?
One image per tier: the adder is tiers × (strobe width + camera frame time). With 1 ms strobes and a 45 ms frame from a 5 MP GigE camera, three tiers take 3 × 46 = 138 ms, about 14% of a one-second cycle. On faster lines cut tiers, reduce resolution or move to a faster camera interface to keep the adder small.
Can I fire two tiers in the same capture?
Yes. Driving two channels together produces a mixed pattern: a steep-plus-grazing combination keeps recesses visible while scratches stay bright. The price is contrast mixing, because features each tier reveals separately can mask one another, and you lose the clean dark field the grazing tier gives on its own. Fire the tiers separately first, then combine only if the cycle cannot afford the extra image.
How do I keep tier brightness consistent over months?
Turn every auto exposure and auto gain mode off, drive each tier from a regulated controller channel at a fixed setting, and image after a short warm-up. Compare the tiers on the same reference target at commissioning and on a schedule afterwards; a channel drifting dim shows up there before the vision tools start failing, and a fixed setting keeps the drift slow and predictable.
What inner diameter does the stack need for my part?
At least the part's largest cross-section plus placement tolerance. A 90° tier lights a disc roughly as wide as its own inner diameter, so a 40 mm part wants 48 mm or more of clear inner window. Oblique tiers converge inward, so their inner radius also sets the mounting height through h = inner radius × tan(angle).
Do the stacked tiers need to be the same colour?
No, and mixing colours is often an advantage: sequence the wavelength that raises contrast for each defect class and read every pass with one monochrome camera. Keep each tier's colour fixed per product, because the tools are tuned to that contrast, and re-check thresholds if you ever swap a tier's wavelength.





