Dark Object Detection: Sensor Options for Low-Reflective Targets
For opaque dark objects, try beam interruption first if mounting allows. If all hardware must stay on one side, compare distance-based optical sensing with a suitable non-optical method. Choose by the actual material, geometry and required result—not a “black target” label or white-card range.
Which sensor should you try first for a dark object?
Start with the required result and available mounting space. For simple presence or counting, an opaque target can interrupt a beam without returning light. Distance measurement, registration marks and feature inspection require different information; a presence switch alone cannot perform all four jobs.
| Available condition | First option to evaluate | Decisive check |
|---|---|---|
| Emitter and receiver can face each other | Through-beam; a fork or fiber arrangement may suit small parts. | The smallest part must attenuate enough of the effective beam throughout its allowed path. |
| Only one side can be wired, but a reflector fits opposite | Retro-reflective sensing with the specified reflector. | For glossy parts, test polarized or mirror-rejection designs so the part cannot imitate the reflector. |
| No hardware can be fitted opposite | Background suppression (BGS) or time-of-flight (ToF) when distances differ; diffuse when return intensity clearly separates the states. | Check actual dark-target return, target-to-background separation and valid operating range. |
| The target is metal and close to the sensor | Inductive proximity sensing. | Metal type, target face, total gap and surrounding mounting metal matter more than paint color. |
| Optical contrast remains unreliable, or more than presence is needed | Ultrasonic, capacitive or vision, according to material and task. | Confirm a usable echo, dielectric change or image feature. Each alternative has its own failure modes. |
One-sided wiring is not one-sided access. A retro-reflective sensor still needs space and a clear path to a reflector on the far side. If that space does not exist, move to a genuinely one-sided method.
Beam interruption also assumes sufficient opacity at the sensor’s wavelength. Dark-looking film or a thin wall may transmit enough light to need a different setup. For a dedicated optical setup guide, see detecting black objects with photoelectric sensors.
Why can two black parts give different signals?
A sensor sees the light returned to its receiver at a particular wavelength and angle. Human-visible color is only one clue. Finish, curvature, pigment, distance and the portion of the spot covered by the target all change that return.
Matte, glossy and textured surfaces fail differently
Matte dark material may send too little light back to an intensity-based diffuse sensor. Glossy or curved material may return a strong reflection at one orientation and direct it away at another. A groove, hole or label can make a small spot see a different surface as the part moves.
That explains why “the same black rubber” is not a complete test specification. Identify the actual detection face and include its permitted rotation, surface texture and position. A larger spot may average texture but can also include the background.
Infrared or blue light is a test option, not a guarantee
OMRON’s wavelength FAQ gives a useful counterexample: a black dye mark on white paper can reflect enough infrared light that the mark becomes difficult to distinguish. A higher return may help detect the object’s presence while reducing contrast between the mark and its background.
Selected blue-light sensors can help with certain dark materials, but there is no universal best wavelength for every black formulation. Compare candidate sensors on the real parts. OMRON’s engineering curves also show why range on a standard reference target cannot simply be transferred to a smaller, darker or glossier one.
Can BGS or ToF separate a black part from a black background?
Yes, if the target and background occupy distinguishable distance regions and the sensor receives enough usable light. Background suppression (BGS) and time-of-flight (ToF) reduce reliance on intensity alone; neither removes the need to verify the actual target and geometry.
Check distance separation before maximum range
Triangulation BGS judges distance from the position or angle of the returned light. ToF designs use light travel timing or phase information. These terms are not competing categories in every catalog: a ToF sensor may also provide background suppression. Compare the sensing method, selected mode and limits rather than the label alone.
For a dark part above a dark belt, document the farthest allowed target face and nearest allowed background. Then check the model’s dark-target range, switching transition, hysteresis and spot size. A nominal gap in a CAD drawing is not automatically a usable separation after vibration and mounting tolerance.
Black foam on black plastic: teach both conditions
Banner’s Q4X tutorial demonstrates black foam in front of a black plastic door panel using two-point static background-suppression teach. The demonstration teaches foam present and foam absent, then places the switching point between the two taught values.
The practical lesson is to compare both application states, not teach only the easiest target. This demonstrates that a distance-based method can solve that particular black-on-black arrangement. It does not establish a minimum foam thickness, production speed or lot tolerance for another model or installation, and it is not an xsz sensor test.
A monitored background is a different strategy
If the target gives an unreliable return but a stable background is available, a documented reference-background or dual-teach mode may detect a change from that reference. Check its truth table: missing background, contamination and a real object can all disturb the return. Ordinary BGS should not be assumed to report a completely non-returning target as “present.”
If neither a usable target return nor a dependable background reference exists, evaluate another physical principle or change the mounting arrangement. Increasing sensitivity cannot supply missing distance information.
When should you use a non-optical sensor or vision?
Change technology when the material or required inspection offers a more dependable signal than reflected light. A black-painted bracket, foam pad and molded plastic cap may look similar while requiring different tests.
Inductive for metal; capacitive for a stable short-range material change
An inductive sensor responds to metal through electromagnetic interaction, so the visible paint color is not the selection variable. Confirm metal grade, target size, operating gap and nearby metal. Paint or another nonmetal coating still contributes to the distance between the sensing face and the metal underneath.
Capacitive sensing can be useful for nearby nonmetal targets, including plastic, paper and liquids. It responds to a change in capacitance, which depends on the material and surrounding arrangement. Compare target-present and empty conditions with permitted moisture, buildup and material variation; do not treat a teach button as immunity to environmental changes.
Ultrasonic for a usable echo—not simply for anything black
Large, solid targets can provide an acoustic echo regardless of visible color. Check target size and angle, minimum distance or blind zone, response time and neighboring sensors. Soft porous foam or fabric can weaken the echo, so an ordinary target-echo sensor may be a poor substitute for an optical one.
Some ultrasonic designs instead monitor a sound-reflecting background. That can help with poor-echo targets, but it requires the appropriate mode and a suitable reference surface. Validate reference loss and obstruction behavior rather than applying a target-echo rule to every ultrasonic sensor.
Vision when the result is shape, orientation or several features
A camera is justified when one switching point cannot answer the inspection question. For an opaque outline, backlighting creates a silhouette and reduces dependence on front-surface color. For surface features, choose lighting and exposure that reveal the required detail. Verify motion blur, occlusion and part variation; 3D systems that rely on returned light also need a valid signal from dark surfaces.
How should you set the distance, spot and threshold?
Use the final geometry and compare the weakest acceptable target with every condition that must remain “no target.” Select a setting that separates those states across their allowed variation—not merely at one nominal position.
Keep the spot on the intended feature
Verify spot dimensions at the closest and farthest positions. Include lateral wander, rotation, holes and shiny edges. A tiny laser spot can isolate a small feature but may fall into a recess; a wider spot can mix a low-return part with a bright rail behind it. The beam spot-size guide covers this geometry in more detail.
With a reflector, check the effective beam where the part actually crosses it, not just the visible spot at the sensor. With BGS or ToF, check the distance transition as well as the nominal switching point. Do not apply a universal mounting angle to every glossy curve.
Overlapping signals cannot be fixed by moving one threshold
Assume an intensity-based diffuse sensor reports 40–70 units with the part present and 25–45 units with it absent. The ranges overlap from 40 to 45. At a 50-unit threshold, the weakest real part can be missed; at 35 units, the strongest background can cause a false detection.
No single intensity threshold separates all these assumed states. Change distance, background, spot placement or sensing method, then repeat the comparison. If the ranges become distinct, still allow for noise, drift and both switching/reset thresholds. The example is not a recommended signal level or acceptance margin.
A sensor without numerical diagnostics can still be evaluated by mapping switching and reset positions over repeated samples. Record the permitted operating region rather than treating one steady indicator LED as proof of reserve. Save the original settings before adjustment, and stop and secure the machine before repositioning hardware or handling samples near hazards.
Is the sensor missing the part, or is the PLC missing the pulse?
Compare the sensing state, electrical output and PLC input before changing sensitivity. A good stationary result followed by missed counts at speed can come from moving geometry, insufficient sensor response or downstream pulse capture. The symptom alone does not identify which one failed.
| Observation | Check next | What the result can distinguish |
|---|---|---|
| White parts work; dark parts do not | Compare samples at the same face, position, angle and speed. | A return-signal problem versus a size, orientation or path difference between the samples. |
| Detection changes with rotation or lateral position | Map the spot against curvature, holes, edges and background. | Angle-dependent reflection or mixed target/background return before blaming response time. |
| Output is correct at the sensor but not at the PLC terminal | Have qualified personnel check the circuit under load. | PNP/NPN compatibility, input common, wiring, supply or signal-level problems. |
| A valid pulse reaches the input but the count is missed | Measure HIGH and LOW times; review input filtering, capture mode and program logic. | A controller-capture problem that more optical gain will not correct. |
| Performance drifts after use | Compare clean and service-limit conditions, including optics, reflector and bracket. | Loss of optical reserve, mechanical drift or changed material rather than an automatically defective sensor. |
A modest count rate can still create a short event
Assume a part is 12 mm long in the travel direction, the effective beam is 2 mm wide along that direction, the clear gap is 20 mm and speed is 2 m/s = 2,000 mm/s. The part covers the full beam for approximately:
(12 − 2) ÷ 2,000 = 0.005 s = 5 ms.
The ideal pitch is 12 + 20 = 32 mm, giving only 62.5 parts/s. That average rate does not describe the shortest usable detection interval. Full-beam coverage is a geometry estimate, not the electrical output width; partial blockage, ON/OFF delays and the active mode affect the actual waveform.
Separately, if a capture shows a 3 ms input pulse and the configured input requires 5 ms of continuous HIGH before accepting it, that pulse will be rejected. These are assumed values, not a PLC specification. Check the real input manual and consider an appropriate capture input or supported pulse-stretch mode; stretching must not merge adjacent parts.
OMRON’s response-time definition separates optical input changes from the sensor’s electrical output. Its CJ2M documentation also distinguishes normal, quick-response and high-speed input behavior. Use the actual module and channel configuration; a fast sensor does not automatically make the entire input chain fast. See response time versus switching frequency for the parameter distinction.
Use qualified personnel and the machine’s approved test mode for electrical captures and running trials. Do not reach into moving equipment or bypass safeguards to present a target.
What should you test before choosing the production model?
Confirm repeatable target/no-target separation and correct controller detection using the real material, installation and operating limits. The useful result is a documented configuration and tested operating region, not a claim that one sensor “detects black.”
- Define the required decision.State presence, count, distance or inspection result, including acceptable misses/false detections and how the trial will count opportunities and failures.
- Test representative limits together.Use relevant production lots, finishes, minimum feature size, allowed angles, distance tolerance and backgrounds. Include permitted service-limit contamination, ambient light and fastest speed without creating unsafe test conditions.
- Record the whole detection chain.Capture available signal or quality data, switching/reset behavior, electrical pulse widths and PLC results. Record sample coverage and trial duration; zero failures in a small trial do not establish a guaranteed failure rate.
- Save the configuration and replacement check.Record exact order code, manual revision, firmware if relevant, response mode, teach settings, optics/reflector, bracket drawing and input settings. For IO-Link, verify the actual available parameters and restoration procedure.
If target and no-target states still overlap, valid readings drop out, or the PLC misses pulses under required conditions, keep the selection open and resolve that specific gap. Repeating an easy demonstration is not a substitute for the failed check.
For a discussion with xsz sensor, provide representative parts, a dimensioned mounting sketch, the background, maximum speed and required output/interface. Ask the supplier to identify the proposed model’s tested limits for those conditions and the evidence still needed before ordering.
This guide concerns process sensing. Ordinary object-detection sensors are not personnel-protection devices; machine-safety functions require suitable protective equipment and validation.