Confocal Displacement Sensor

What is a Confocal Displacement Sensor

A confocal displacement sensor is a non-contact optical gauge that reports the distance to a target surface, the displacement of that surface over time, or the thickness of a transparent layer, by determining which part of a focused light cone is in sharp focus on the surface. The defining feature is the confocal arrangement: light passes through a pinhole on its way out and through a conjugate pinhole on its way back, so only rays that return precisely in focus reach the detector while out-of-focus scatter is rejected. This optical filtering is what gives the instrument its high axial selectivity and its tolerance for surfaces that defeat other techniques.

The confocal principle traces to Marvin Minsky, who patented confocal microscopy in 1957 to image thick biological tissue by rejecting stray light. The displacement-sensor branch industrialised that idea in two ways. The first, reflectance or laser confocal, scans the focus mechanically with monochromatic light and finds the position of peak return. The second, chromatic confocal, replaces the mechanical scan with a dispersive lens and broadband white light: each wavelength focuses at a different distance, so distance is encoded as colour and read instantly by a spectrometer with no moving parts. The French company STIL, founded in 1993 and now part of the Marposs Group, was an early commercial pioneer of the chromatic approach, and today the chromatic variant dominates industrial point sensors.

Structurally, a modern chromatic confocal system has three blocks. First, a broadband light source, historically a white LED covering roughly 380 to 760 nm, and in research systems a supercontinuum source spanning several hundred to a few thousand nanometres. Second, the sensor head or probe, which carries the dispersive objective that spreads the focal points of different wavelengths along the optical axis and is connected to the controller by an optical fibre. Third, the controller, which houses the spectrometer that identifies the in-focus wavelength, the peak-extraction algorithm, the calibration table that maps wavelength to distance, and the industrial interfaces. Splitting the head from the controller lets the small passive probe sit in tight or hot fixtures while the electronics stay remote.

Four engineering metrics decide whether a given confocal sensor fits an application: measuring range, spot diameter, resolution, and linearity. They are coupled through the objective, so they cannot be optimised independently. A short-range probe yields a tiny spot, fine resolution, and tight linearity but a small working window; a long-range probe relaxes all three to gain reach. There is no universal confocal head, and the heart of selection is matching the range and spot to the geometry and roughness of the part rather than chasing a single headline accuracy number.

In application scale, confocal displacement sensors span an enormous dynamic range of precision. Short-range laboratory and metrology heads resolve single-digit nanometre steps over a fraction of a millimetre, supporting wafer flatness, lens form, and roughness work governed by the ISO 25178 areal surface texture series. Production heads trade resolution for reach and speed, measuring runout, glue-bead height, and the same coating thickness a dedicated coating thickness gauge reports, plus glass and film thickness on fast lines. The same physical principle therefore serves both a metrology laboratory and an inline gauging station, which is unusual among displacement technologies such as the short-range capacitive sensor and the contact-based LVDT sensor.

Confocal Sensor Types

Confocal displacement sensors split first by how they encode distance, then by probe geometry. The encoding split, chromatic versus reflectance, is the decision that matters most for procurement, because it determines whether the head has moving parts, how fast it samples, and how it is calibrated. The geometry split, point versus line and fibre-coupled versus integrated, then tailors the chosen principle to the part. The table below contrasts the two encoding families against the laser triangulation sensor most engineers already know.

Chromatic confocal is the mainstream industrial type. A dispersive objective spreads the focal points of a continuous spectrum along the optical axis, short wavelengths focusing nearest the lens and long wavelengths farthest, so a single distance corresponds to a single in-focus colour. A spectrometer reads that colour, and a factory calibration table converts wavelength to distance. With no moving parts the head can be small, passive, and fast, and it can resolve two surfaces of a transparent layer at once. Keyence CL-3000, Micro-Epsilon confocalDT, Omron ZW, and Precitec CHRocodile are all chromatic confocal systems.

Reflectance, or laser, confocal uses monochromatic light and a fixed pinhole, then physically sweeps the focus through the range and records the position at which the return signal peaks. Because it concentrates all energy at one wavelength it can deliver extremely high axial resolution, but the mechanical scan adds a moving element, slows the measurement, and complicates inline use. It survives mainly in laboratory microscopy and specialised metrology rather than fast production gauging.

Within the chromatic family, geometry creates further variants. Point sensors measure one spot at a time and are the workhorse for distance, displacement, and single-track thickness. Line sensors spread many confocal points across a line, so Precitec reports its white-light line arrays acquiring on the order of hundreds of thousands of points per second for areal topography and roughness. Fibre-coupled heads, like any fiber optic sensor and used by Omron ZW and Micro-Epsilon, separate a tiny passive probe from the controller for tight or harsh mounting, whereas integrated heads keep optics and electronics together for simpler installation. Most catalogues also divide heads into focused-spot variants for the smallest spot and finest resolution and quad or wide variants for larger spots that average over rough texture.

The Dispersive Measurement Chain

To read a confocal spec sheet you need to understand the physical chain that turns reflected light into a distance value. In a chromatic confocal sensor that chain has four stages, and each stage sets one or more of the headline parameters. The table below maps the four stages to the specification each one drives, so you can trace any number on a datasheet back to the part of the instrument that produces it.

The broadband source must emit a continuous, stable spectrum because the instrument encodes distance as colour. White LEDs covering roughly 380 to 760 nm are the production standard; their stability and lifetime suit factory duty. Research and extreme-range systems use supercontinuum sources spanning hundreds to a few thousand nanometres for a wider encoded range. Source brightness sets how much signal returns from dark or steeply tilted surfaces, which in turn caps the usable measuring rate on difficult targets, because the controller integrates longer when light is scarce.

The dispersive objective is the defining optic. Its controlled axial chromatic aberration assigns each wavelength a focal distance, and the span between the nearest and farthest focus is the measuring range. The same lens design fixes the spot diameter and the tilt-angle tolerance through its numerical aperture: a high-aperture, short-range objective produces the smallest spot, the finest resolution, and the tightest linearity but the narrowest range, while a longer-range objective relaxes all of these. This coupling is why range, spot, resolution, and linearity always appear together on a head datasheet and must be read as a set.

The confocal pinhole is what makes the instrument confocal rather than merely chromatic. By passing only light that returns in sharp focus and blocking everything else, it gives the sensor sharp axial selectivity and its remarkable tolerance for reflectivity, so the same head reads matte black and polished metal. The pinhole also enables transparent-thickness measurement, because the front and rear surfaces of a clear layer each produce their own in-focus return at their own wavelength, appearing as two clean peaks instead of a blur.

The spectrometer and peak-extraction algorithm close the chain. The spectrometer disperses the returning light onto a detector array, and software locates the wavelength peak, then the calibration table converts that wavelength to a distance. Algorithm quality directly sets resolution and linearity: a simple maximum-intensity search is fast but coarse, a centroid method is more robust, and Gaussian peak fitting generally gives the best accuracy, which is why high-end controllers advertise their peak algorithm. The detector readout and integration time then set the measuring rate, which ranges from a few hundred microseconds on production heads down to tens of microseconds and tens of kilohertz on fast controllers such as the Omron ZW-7000, which the datasheet rates at a 20 microsecond sampling capability.

Surfaces, Materials, and Thickness

The reason engineers reach for a confocal sensor is its tolerance for surfaces that defeat laser triangulation. Because the confocal pinhole accepts only the in-focus return and discards scatter, one head can read targets across a huge reflectivity range without re-aiming. Keyence states that a confocal head measures stably on materials from dark rubber to clear film without adjusting mounting or settings, which is the single largest practical advantage of the technology. The two limits that remain are signal level and surface tilt, and both are managed through head choice and controller settings rather than mechanical fixturing.

Dark and low-reflectivity surfaces return little light, so the controller compensates by raising exposure and lowering the measuring rate to integrate longer. The instrument still reads, but at reduced speed, so a line that must gauge black rubber at high throughput needs both a bright source and headroom in the rate budget. Mirror and specular surfaces are read well as long as the head sits near normal incidence; outside the tilt-angle window the specular return walks off the pinhole and signal collapses, so the datasheet tilt tolerance, which Micro-Epsilon describes as the angle the sensor tolerates by design, defines the mounting envelope.

Rough surfaces interact with spot diameter. When roughness features are larger than the spot, each measurement samples a different micro-facet and the wavelength peak broadens, inflating noise. The fix is a larger-spot head that averages over the texture, which is exactly why catalogues offer both focused-spot heads for smooth precision targets and wider quad-type heads for coarse machined parts. For dedicated roughness work, where the confocal head competes with a stylus surface roughness tester, the relevant rules come from the ISO 25178 areal surface texture series, whose Part 602 sets the design and metrological characteristics of confocal chromatic probe instruments and Part 607 covers confocal microscopy instruments.

Transparent-thickness measurement is the confocal sensor's signature trick. White light reflects from both the front and rear face of a clear layer, returning two wavelength peaks; the controller measures the optical gap between them and divides by the material refractive index to recover physical thickness, all from one side and with no backing fixture. The method is largely insensitive to tilt. Multi-peak controllers extend it to laminated stacks: Micro-Epsilon confocalDT evaluates up to six peaks to gauge up to five layers, retrieving a wavelength-corrected refractive index for each layer from a material database, and Omron ZW-8000 separates the reflections from the top and bottom of a thin transparent sheet that conventional laser sensors cannot resolve. The table below maps common surfaces to the recommended head choice and the parameter that limits performance.

Key Specification Parameters

A confocal head datasheet typically lists a dozen or more parameters, but seven drive the selection decision: measuring range, reference distance, spot diameter, resolution, linearity, measuring rate, and tilt-angle tolerance. The Key Specifications comparison below pulls real, datasheet-traceable values from three established product families so the trade-offs are concrete rather than abstract.

Measuring range is the axial window between the nearest and farthest in-focus distance, and reference distance (standoff) is the gap from the head face to the centre of that window. Keyence CL-3000 heads, for example, are named by reference distance from 7 mm up to 150 mm, with corresponding ranges from a few millimetres to tens of millimetres, while Micro-Epsilon IFS2405 ranges run from 0.3 mm to 30 mm. Choose the smallest range that still clears your standoff and the full travel of the part, because a smaller range buys finer everything.

Spot diameter sets the smallest feature you can resolve laterally and how the head behaves on rough surfaces. Focused-spot heads reach the tens-of-microns class (Keyence focused heads to 25 micron, Micro-Epsilon IFS2405 down to a 6 micron spot), ideal for edges, grooves, and small targets, whereas quad and wide heads relax to hundreds of microns up to about 1000 micron to average over texture. Resolution is the smallest detectable change: Micro-Epsilon quotes 0.01 micron (10 nm) on its finest range and Keyence lists 0.015 micron typical and 0.003 micron on its best configuration.

Linearity is the worst-case deviation from a straight reference across the range, and it is the number that most honestly describes accuracy. It scales with range: Keyence CL-3000 spans plus-or-minus 0.28 micron on short heads to plus-or-minus 5.5 micron on long ones, and Micro-Epsilon IFS2405 reaches better than plus-or-minus 0.15 micron on its smallest range. Read resolution, linearity, and repeatability as three separate specifications, because a sensor can resolve a 10 nm step yet still deviate by a micron across the full range; never collapse them into one accuracy figure.

Measuring rate and tilt-angle tolerance finish the set. Rate ranges from the kilohertz region on standard heads to the Omron ZW-7000 capability of a 20 microsecond sample, and it drops on dark or tilted targets where the controller must integrate longer. Tilt tolerance, the angle of incidence the objective accepts before the return walks off the pinhole, defines the mounting envelope on specular parts and is generous on diffuse surfaces. Confirm both at your real surface and reflectivity, because vendor numbers are quoted on a flat, cooperative reference target.

Selection Decision Factors

To turn the preceding chapters into a specific model, follow the decision sequence below. Most selection errors come not from a single wrong value but from deciding parameters in the wrong order, for instance fixing on a resolution before checking that the range clears the part travel. These seven steps work as a fixed RFQ template for any confocal displacement sensor enquiry.

1. Measurand and standoff: First decide whether you are measuring distance, displacement, profile, or transparent thickness, then fix the reference distance from the available mounting space and the worst-case travel of the part. The standoff and travel together set the minimum measuring range you can accept.
2. Range, spot, and resolution as a set: Pick the smallest range that still clears standoff plus travel, because that choice simultaneously buys the smallest spot, finest resolution, and tightest linearity. Confirm the spot is smaller than the feature you must resolve and, for rough parts, large enough to average texture.
3. Surface and reflectivity: Map your target to Chapter 4. Dark or tilted surfaces need a bright source and rate headroom; specular parts need near-normal mounting inside the tilt-angle tolerance; transparent layers need a multi-peak controller and the correct refractive index per layer.
4. Speed budget: Set the required measuring rate from line throughput and part dynamics, then derate it for dark or tilted targets that force longer integration. A 20 microsecond capability on a cooperative target may fall well below that on black rubber.
5. Probe geometry and mounting: Choose point versus line, fibre-coupled versus integrated, and focused versus quad. Fibre-coupled passive probes suit tight, hot, or vibrating fixtures and keep the spectrometer remote; line sensors suit areal topography and roughness.
6. Interfaces and integration: Confirm the controller offers the analog and digital outputs and fieldbus or Ethernet protocols your PLC or machine vision system needs, plus the I/O for triggering, and that multi-peak or thickness firmware is included where required.
7. Standards and traceability: For surface texture work, require conformance to the relevant ISO 25178 parts, namely Part 602 for confocal chromatic probe instruments and Part 607 for confocal microscopy, with calibration guidance from VDI/VDE 2655. For displacement work, ask for a calibration certificate stating linearity and resolution test conditions.

One last commonly overlooked dimension is manufacturer serviceability: local calibration laboratories, spare-probe lead time, firmware and refractive-index database updates, and application engineering for difficult surfaces. These matter little at purchase but determine downtime years into production. Keyence (CL-3000 series), Micro-Epsilon (confocalDT IFS2405 and IFS2407), Precitec Optronik (CHRocodile), STIL (a Marposs company), Omron (ZW-8000 / 7000 / 5000), Nanovea, LMI Technologies, and Cyberoptics are the established suppliers, with Keyence, Precitec, and Micro-Epsilon holding the largest combined market share. Keyence and Omron lead fast turnkey inline gauging, Micro-Epsilon and Precitec lead high-end metrology and multi-layer thickness, and STIL and Nanovea anchor laboratory profilometry.