Renishaw Laser Interferometer Detailed Introduction

• 1
  What a Renishaw laser interferometer detailed introduction actually covers
• 2
  Linear, angular and straightness setups in practice
• 3
  Why the air in the beam path decides the result
• 4
  Reading the numbers: ppm, resolution and real error
• 5
  Where the laser method stops being the right tool
• 6
  Using laser results to tighten machined tolerances
• 7
  Which measurement setup fits which job

A Renishaw laser interferometer does not measure a part. It measures how far a reflector moved along the beam path. A helium-neon source emits a stable wavelength, and the beam splits into a reference path and a measurement path. When the moving reflector shifts, the two beams recombine with a phase difference.

The system counts interference fringes. Each fringe equals one half wavelength of travel, so a 633 nm source resolves motion in roughly 316 nm steps before electronic interpolation. Interpolation inside the detector splits each fringe further, which is where nanometer resolution comes from. Resolution is not accuracy. The two numbers come from different parts of the chain.

Because the count depends on wavelength, anything that changes the wavelength changes the reading. Air does exactly that. Temperature, pressure and humidity shift the refractive index of the air in the beam path, and over a 1 m axis a few degrees of temperature drift can move the reading by tens of microns if nothing corrects it.

This is why a laser head alone is not a measurement system. The compensator unit sits beside it, reading local air conditions many times per second and scaling the fringe count to a standard condition. Skip that, and you have a very precise instrument reporting the wrong distance.

Linear measurement is the common case: the laser sits at one end of the axis, a linear reflector rides the moving element, and the displayed value is the distance traveled. The beam must stay inside the aperture of the reflector over the full stroke. On a 4,000 mm axis, a 1 mm misalignment at the source can walk the beam off the target by the time it reaches the far end.

Angular measurement swaps the linear reflector for a pair of angular optics. The beam splits into two parallel paths a known distance apart, and any pitch or yaw between them shows up as a path length difference. This is how you find out that an axis is square at the near end and 40 μm out at the far end.

Straightness uses a different optic again, usually a Wollaston prism that shears the beam into two paths and compares their lateral displacement. The reading is relative to the beam, not to a granite square, so the setup itself defines the reference line. Move the laser and you have a new reference.

Each optic set has a working range. Linear reflectors tolerate more misalignment than angular ones. Straightness optics are the most sensitive to beam height and tilt. If a setup looks marginal on paper, it will be marginal on the machine.

The compensator reads air temperature, pressure and humidity and applies the Edlen equation to get the refractive index. That correction is only as good as the sensor placement. A sensor sitting in the exhaust of a spindle will report a different index than the air 300 mm away in the beam path.

Thermal gradients are the real enemy. A machine that has run for two hours has a warm bed, a warm column and a cool floor. The beam crosses all three. Even with compensation, a gradient along the axis produces a length error that no single-point sensor can remove, because the sensor samples one location and the beam averages many.

The practical response is to let the machine soak. Run it through its warm-up cycle, then measure. Take a repeat run at the end of the session. If the two runs disagree by more than the tolerance you care about, the numbers from the middle of the session are not trustworthy either.

Air turbulence from open doors, fans or passing traffic adds noise to the fringe count. Short measurement runs hide it. Long runs do not. For a 4,000 mm axis, plan the run so the beam is not crossing a doorway or a cooling vent.

Laser specifications usually quote accuracy in parts per million. A figure of ±0.5 ppm means ±0.5 μm per meter under ideal conditions. Over a 2 m axis that is ±1 μm from the instrument alone. Add compensation uncertainty, reflector misalignment and thermal gradient, and the real uncertainty is larger than the headline number.

Resolution is a separate figure and often quoted in nanometers. It tells you the smallest change the system can display, not the smallest change it can measure correctly. A system that resolves 1 nm in a room drifting 0.5 °C per hour cannot measure 1 nm. Resolution sells instruments. Uncertainty decides whether the result holds up.

For machine acceptance work, the relevant standard is usually ISO 230-2, which defines how to run the test and how to report the result. The standard fixes the number of target points, the approach direction and the number of runs. Deviating from it produces a number that looks like an acceptance result but is not comparable to anyone else's.

When we verify a machined part against a drawing tolerance of ±0.005 mm, the laser is not the tool that decides it. A CMM or a gauge does that. The laser checks the machine that made the part. Keeping those two roles separate prevents a lot of confusion in a first-article meeting.

The method needs line of sight. If the beam cannot reach the far end of the axis without clipping a way cover, a fixture or a cable carrier, you either remove the obstruction or you pick another method. Removing a way cover to get a clean path changes the thermal behavior of the machine, which changes the result.

It measures motion, not geometry. A laser will tell you an axis is accurate to 3 μm over its stroke. It will not tell you the spindle is out of square to the table, or that the guideways are twisted. Those checks need a square, a level or a ballbar, depending on the error you are chasing.

Very short travels are awkward. Below roughly 100 mm, the setup overhead and the uncertainty from the optics can exceed the error you are trying to find. A comparator stand and a gauge block often answer the question faster.

The method also assumes the machine is in a state worth measuring. If the guideways are worn or the ballscrew has backlash, a laser run will report the error accurately and repeatedly. It will not fix it. Measurement comes after the mechanical repair, not before.

For a shop holding ±0.005 mm on production parts, the laser run is a maintenance input. It tells you which axis has drifted and by how much, so the compensation table can be updated instead of guessing at the control. That keeps a machine inside tolerance for years longer than running it until parts fail inspection.

The link to the part is indirect but real. If a 5-axis machine has 20 μm of pitch error over its working envelope, a part with tight features at the extremes of that envelope will show the error even when the cut itself was perfect. The laser finds the machine error before the part does.

We run 127 high-precision CNC machines across three plants, including 16 simultaneous 5-axis machining centers, and hold ±0.005 mm with a 99.99% qualification rate. That does not come from measuring parts harder. It comes from keeping the machines calibrated and the thermal environment stable.

When a customer asks why a part moved after a design revision, the first question is often about the machine, not the program. A recent laser record answers it in minutes. Without one, the investigation starts from zero.