Laser triangulation
Laser triangulation — Laser triangulation projects a line or spot onto the target at an angle and measures where the reflected light lands on a position-sensitive detector, so distance is solved from the known baseline and the observed image position.
Time-of-flight
Time-of-flight — A time-of-flight sensor modulates its emitted light, compares the phase of the returned signal with the emitted one and converts that phase shift into a distance, which means it needs enough reflected energy but not a sharp image of the spot.
| Mechanism | Limit at long range | Typical failure on a reflective target | Typical failure on a dark target |
|---|---|---|---|
| Laser triangulation | Optical magnification falls with the square of distance, so the spot image blurs | Specular glare throws the spot off the detector and produces spikes | Too little returning energy, the peak is lost in noise |
| Phase time-of-flight | Ambient light and the emitted power budget, not the optics | Multi-path echoes create a phantom target further away | Low contrast, so the modulation amplitude falls below the usable threshold |
| Pulsed time-of-flight | Detector timing jitter and the pulse edge | Reflections from a shiny wall behind the target win because they are stronger | Long dead time, so close and far targets merge |
| Confocal chromatic | Very small measurement range by construction | Almost none, it rejects most off-axis light | Needs a defined return, so a matte black surface still works |
| Required resolution | decides the mechanism first |
|---|---|
| Standoff available | decides the second |
| Target optical behaviour | decides the model within the mechanism |
| Machine cycle time | decides sampling and averaging |
| Neighbouring sensors | decides cross-talk handling |
- Illustrative resolution of a triangulation head as range grows
- 8x worseat 4 times the distance
- Source: Illustrative sample dataset, not a measured value.
Why does a laser triangulation sensor lose resolution so quickly with distance?
Because the geometry is a triangle: at a fixed baseline the image position moves less and less for each millimetre of target movement as the target moves away, so the same detector pixel now represents a larger slice of distance, and the spot also defocuses. Resolution therefore degrades roughly with the square of the working distance, which is why a head specified at a few micrometres at 50 mm is a millimetre-class device at 400 mm, no matter how good its optics are.
Can two time-of-flight sensors watch the same conveyor without interfering?
Only if you manage the coupling. Two phase-based heads emitting modulated light at the same frequency and seeing each other's beam will report a distance somewhere between the two targets, which looks like random long jumps rather than a stable error; separating them optically with apertures, alternating the emission with a sync line, or choosing units that hop modulation frequency are the practical fixes, and retro-reflective variants with polarised optics reduce the coupling further.
Which technology reads a transparent bottle reliably?
None reads it the way they read steel, because at a smooth transparent surface part of the beam reflects and part refracts into the wall and then into the contents, so the return is several signals at once. Triangulation sees a double or triple peak and ToF averages them into a meaningless value; the reliable approaches are to break the specular path with a 45 degree incidence and a matte or black backing behind the bottle, or to measure the wall from the inside by putting the reflector in the liquid path.
Every figure on this page, including the stat block and the comparison table, is an illustrative sample value for a demonstration site. Replace them with your own measurements before publishing and do not quote this page as a specification.
The mechanism, because it predicts the failure
A triangulation head is a camera looking at a laser line through a known angle. The electronics find the centre of that line in the captured image and convert its vertical position into a distance using the baseline and the angle. Everything the sensor can do follows from that: it needs a detectable, localisable bright line. A specular surface mirrors the line away from the lens, so the line either disappears or jumps to a different part of the field, and the sensor reports a spike rather than an error. This is why a polished bracket can produce readings outside the measurement range while a painted one is stable.
A phase time-of-flight head is not an imaging system in the same sense. It floods the target with light whose amplitude oscillates, then measures how far the returning wave has been delayed by comparing the phase. That delay is a genuine distance measurement but only if the reflected modulation is strong enough to be trusted, so the useful indicator on a ToF head is its signal amplitude or modulation margin, not a picture of the spot. Dark, matte, low-albedo surfaces eat the amplitude and the reading becomes noisy long before the nominal range limit is reached.
The consequence for selection is that the two technologies fail in different directions: a triangulation head fails when the surface optical behaviour changes, and a time-of-flight head fails when the returning energy drops. Which one you should tolerate depends on your process, and a shiny machine interior with dark pallets is the case where both are marginal.
Range, aperture and the square law
Resolution of a triangulation head is usually quoted at the centre of its range, at its optimum distance, with a matte white target. Two effects then erode it as the geometry grows. The image displacement per unit of target motion shrinks because the triangle flattens, and the depth of field does not grow with it, so the spot edge softens. The practical rule is that resolution degrades faster than linearly with distance - roughly quadratically in the region beyond the design point - which is why heads are sold in hundreds of narrowly ranged variants. Specifying a longer range “for margin” therefore buys you robustness in one dimension and quietly spends resolution in the other.
Aperture buys depth of field by reducing light and increasing diffraction, so a head that has been dialed for a wide depth of field loses signal margin on dark targets. Where the drawing demands the tolerance, choose the shortest range that clears the mechanical envelope, then verify resolution on the actual parts rather than the datasheet target.
Multi-path, ambient light and neighbouring heads
In a phase ToF measurement, any path that is longer than the direct path is a lie. Inside a stainless steel enclosure the emitted beam can bounce off a curved wall and return after the target echo, and the receiver mixes both waves into a single phase that corresponds to neither. The classic signature is a reading that jumps to a fixed longer distance whenever a particular door or bucket is present. The countermeasures are geometric: make the first reflection land outside the receiver’s field, dull a shiny nearby surface with matte tape or anodised black, and never let the sensor see a corner that faces it.
Ambient light adds a direct-current pedestal that reduces modulation contrast, and pulsed light sources such as strobe lights or daylight through a skylight are much worse than steady incandescent light, because they modulate the background at frequencies the sensor cannot reject. Sunlight contains broad infrared that raises the pedestal far more than any lamp, which is why an outdoor or rooftop-truck application often needs a hood, a narrower field of view, or a modulated laser head with an optical bandpass filter.
Neighbouring heads are the third coupling. The rule of thumb: heads that can see each other’s emitter, or see the same shiny surface, need synchronisation or separation. This is one of the most common commissioning faults on a multi-lane line, and it presents as a sensor that works alone and misbehaves only when the adjacent lane runs.
Choosing between the two
Reduce it to three questions. First, what resolution does the tolerance in the drawing require, and at what distance? If the answer is micrometres, the working distance must be short and triangulation or confocal is the only family left. Second, is the target’s optical behaviour constant? If the process passes painted, machined and oily surfaces on the same line, triangulation needs either a fixed reference geometry, an amplitude-based target-detection check, or a technology that tolerates reflectivity variation. Third, what does the machine do around the sensor - strobe lighting, moving steel, other heads? Each of those pushes the choice one way or the other, and the answer is usually visible on the signal amplitude trace before it is visible in the output errors.