American Optoelectronics Laser Interferometer Data for Large 5-Axis CNC Work

A laser interferometer does not measure a part. It measures the machine. A heterodyne laser head is mounted on the bed and a reflector rides the moving axis, and the system counts interference fringes to resolve displacement along the beam path. On a large 5-axis machining center the useful work is comparing commanded position with actual position across the whole travel, not just at one point near the home position.

For a machine with 4,000 mm of X travel, straightness and angular errors accumulate. A 10 μm pitch error at the rail becomes a 40 μm error at the tool tip once the column leans. Laser measurement separates those contributors: linear positioning, straightness in two planes, yaw, pitch, roll, and squareness between axes. You get a number for each, which is what lets a process engineer decide whether to compensate in the control or to re-level the machine.

The second half of the job is rotary. A 5-axis machine adds two rotational axes, and their centerlines have to sit in a known relationship to the linear axes. Interferometry with an indexed rotary attachment measures angular positioning of the table or the head, then feeds those values into the kinematic model. If that model is wrong by 20 arc-seconds, a part with a 300 mm swing will drift roughly 0.03 mm at the outside diameter.

• 1
  Linear axesPositioning, straightness, pitch, yaw and roll over full travel.
• 2
  SquarenessPerpendicularity between X, Y and Z, measured as an angle.
• 3
  Rotary axesAngular positioning and centerline location for the 4th and 5th axis.
• 4
  Volumetric checkCombined error at the tool tip, not just per-axis numbers.

We keep an american optoelectronics laser interferometer in the metrology room and run it on any machine that will carry a large 5-axis job. The reason is simple: the compensation file in the control drifts as a machine wears, and a machine that passed at installation may be 30 μm out two years later. Measuring first means the quote is based on what the machine can hold today.

This matters most for parts that combine a long linear feature with a compound angle. An aerospace bracket with a 900 mm bolting face and a 30° flange has to satisfy two tolerances at once, and the error at the far end of the face depends on both the linear accuracy and the rotary centerline. Without measurement data we would have to quote a wider tolerance band or add a secondary operation.

For smaller work the picture changes. A part that fits inside a 500 × 500 × 450 mm envelope can often be finished on a machine with a tighter, shorter travel, where straightness errors have less room to grow. Matching the part envelope to the machine travel is a cheaper way to hold tolerance than chasing the last few microns on a large machine.

When a customer's drawing calls for a true position of 0.05 mm across a 2 m weldment, we check whether the machine can do it before accepting the order. If the measured volumetric error is larger than the tolerance budget, we say so and propose either a different machine or a different datum scheme. That conversation is cheaper than a rejected lot.

• 1
  Measure firstRun the interferometer before quoting, not after the first article.
• 2
  Match envelopePut the part on the smallest machine that can hold it.
• 3
  Decline earlyIf the machine cannot hold the tolerance, say so before the order.

Laser interferometry describes machine geometry. It says nothing about thermal drift, tool wear, or spindle growth. A machine can measure perfectly cold and cut a 1,200 mm aluminum part 0.08 mm out of flat after four hours of roughing. We handle that with in-process checks and a warm-up cycle, not with the laser.

It also cannot measure a feature that is already cut. If you want to know whether the finished part is good, that is a CMM or an on-machine probing job. The interferometer tells us whether the machine is capable; inspection tells us whether the part is conforming. Both reports exist, and they answer different questions.

Environmental conditions matter. Air temperature and the part's thermal state change the reading, so we let the machine idle to a stable temperature before a volumetric run. On a large machine this soak can take several hours, which is why we schedule measurement around production rather than inside a job.

Finally, the data has a resolution and a stated uncertainty. A reading of 8 μm is not a promise that every part will land within 8 μm. It sets the floor for what the process can hold after fixturing, tool deflection and material springback are added on top.

• 1
  Not a part checkFinished geometry is verified by probing or CMM, not the laser.
• 2
  Thermal blind spotDrift during long roughing cycles needs separate control.
• 3
  Soak timeStable machine temperature is a prerequisite for a valid run.

GreatLight runs 127 high-precision CNC machines across 3 wholly-owned plants in Dongguan and Singapore, including 16 simultaneous 5-axis machining centers with travel up to 4,000 × 400 × 150 mm and a Ø400 mm rotary table. Large 5-axis work is scheduled onto the machines whose latest interferometer run fits the tolerance budget of the part.

Tolerance capability is ±0.005 mm on features that sit inside the verified working volume, with surface finish options from Ra 0.2–0.8 μm for fine work up to Ra 1.6–3.2 μm as-machined. Materials range from 6061-T6 and 7075 aluminum to 17-4PH stainless, Ti-6Al-4V, Inconel and engineering plastics. The measurement data is what decides which of those combinations is realistic on a given machine.

Inspection is 100% before shipment and covers raw material check, in-process monitoring and final inspection, with reports available on request. The plants hold ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022 certification. Quotation and free DFM analysis come back within 12 hours, production can start within 24 hours, and parts typically ship in 3–5 days.

There is no minimum order quantity. A single prototype and a 10,000-part run go through the same measurement gate, because the machine capability question does not change with batch size. Uploads are handled as confidential, and an NDA is available on request.

• 1
  16 five-axis centersScheduled by measured capability, not by availability alone.
• 2
  ±0.005 mmCommitted on features inside the verified working volume.
• 3
  12-hour quoteQuotation and free DFM analysis, with production start in 24 hours.