SCHUNK Intelligent Luminaires Become Machine Tools Among Machine Tools

• 1
  How SCHUNK Intelligent Luminaires Become Machine Tools
• 2
  Optical Sensing Inside the Work Envelope
• 3
  Part Families That Suit In-Fixture Optical Checks
• 4
  Limits of Turning a Fixture Into a Machine Tool
• 5
  Integrating Optical Checks With 5-Axis Machining
• 6
  In-Fixture Optical Check vs Manual Gauge vs CMM

A luminaire is normally a lamp. When SCHUNK puts one inside a clamping module, it stops being a lamp and starts reporting on the workpiece. The fixture already holds the part; adding light plus a camera turns the same fixture into a gauge. That is the whole idea behind the phrase SCHUNK intelligent luminaires become machine tools.

The optical path is short. An LED ring or bar sits close to the clamping jaws, inside the work envelope. A camera reads the silhouette, edge or surface of the part as it sits in the fixture. In effect, the machine tool now carries its own inspection station, seated a few millimetres from the cut.

Why does proximity matter? Because a part measured outside the machine has already lost its clamping state. Once you unclamp, the residual stress releases and the part moves. Measuring in the fixture captures the geometry that the next operation will actually see.

This is not a replacement for a CMM. It is a fast, repeatable check at the point of cut. The luminaire supplies controlled illumination so the camera sees consistent contrast. Without stable light, edge detection drifts with every coolant splash and chip.

• 1
  Short optical pathSensor sits inside the work envelope, millimetres from the jaws.
• 2
  Clamping state preservedMeasurement happens before the part is released.
• 3
  Controlled illuminationStable contrast is what makes edge detection repeatable.

Three quantities are practical to read in-fixture: presence, edge position and surface condition. Presence answers a binary question: is the blank seated, or is a jaw empty? Edge position gives a coordinate, which feeds a wear offset. Surface condition catches a scratch, a burr or a missing chamfer before the part leaves the machine.

Edge position is the most useful of the three. On a turned shaft, a camera can locate a shoulder to within a few micrometres under good light. That number goes to the control as a delta, and the control nudges the tool offset. The operator never touches the offset page.

Surface condition is harder. Coolant, chips and vibration all corrupt a clean image. In practice, surface checks work best on dry or air-blasted features, or after a short dwell with the spindle stopped. On a busy production cell, that dwell costs cycle time.

Presence checks are the cheapest to deploy and the easiest to trust. A single threshold on a binary image survives a lot of noise. If a shop only adopts one function, presence is the one that pays back first.

• 1
  PresenceBinary check that the blank is seated in the jaws.
• 2
  Edge positionCoordinate that drives a tool wear offset.
• 3
  Surface conditionScratch, burr or missing chamfer detection.

The best candidates share three traits: a repeatable clamping datum, a visible feature and a tight tolerance that a human would otherwise gauge by hand. Shafts with shoulders, housings with a bore edge, and flat plates with a milled step all qualify. The camera needs line of sight to the feature, so a deep internal pocket is usually out of reach.

Aluminium parts in 6061, 7075 and ADC12 behave well. The material is bright and machined edges hold contrast. Stainless in 303, 304 and 17-4PH also reads cleanly, though a matte finish scatters more light and needs a longer exposure. Titanium in TC4 (Ti-6Al-4V) reflects unevenly; a diffused ring helps.

Parts that need frequent first-article checks are strong candidates. If an operator stops the cycle twice a shift to gauge a shoulder, an in-fixture check removes those stops. On a 10,000-part run, that saved time compounds.

Parts with heavy chatter marks or a rough as-machined finish are poor candidates. Ra 3.2 μm and coarser scatters light in unpredictable ways. A shop running only roughing passes will not get a stable edge reading.

• 1
  Good fitShafts, housings and plates with visible datum features.
• 2
  Good materials6061, 7075, 303, 304, 17-4PH read cleanly.
• 3
  Poor fitDeep internal pockets and rough Ra 3.2 μm surfaces.

A fixture-mounted sensor is not a metrology lab. It cannot replace a CMM for first-article inspection on a new part. The camera sees one face at a time, and its calibration drifts with temperature. A shop that treats it as a CMM will eventually ship a bad part.

Cycle time is the second limit. A presence check costs a fraction of a second. An edge measurement with a dwell and an air blast can add several seconds. On a 45-second cycle, that is a real cost, and the shop has to decide whether the scrap it prevents is worth more.

Coolant is the third limit. Flood coolant on a mill obscures the view. Most installations use an air curtain, a short pause with the spindle stopped, or a window that gets wiped by the tool change. None of these are free.

For a shop running one-off prototypes, the setup cost rarely pays back. The technology earns its keep on recurring part families where the same check runs thousands of times.

• 1
  Not a CMMUse it for recurring checks, not first-article sign-off.
• 2
  Cycle costEdge measurement with dwell can add seconds per part.
• 3
  Coolant riskFlood coolant needs an air curtain or a wiping cycle.

On a 5-axis machine, the fixture often rotates. That means the camera has to be either mounted on the fixture and rotate with it, or fixed to the table and calibrated against the rotary axis. The second option is more common. A Ø400 mm rotary table gives enough room for a fixed sensor head outside the swing.

On our 16 simultaneous 5-axis centers, the practical layout is a fixed sensor on the table, calibrated to the rotary datum. The probe head sits clear of the Ø400 mm rotary table swing. Calibration is rechecked after every tool change that touches the fixture.

For 4-axis and 3-axis work, the integration is simpler. A fixed head above the vise, aimed at a shoulder or a bore edge, is enough. On a 4,000 mm bed, a single sensor covers only part of the travel, so long parts need either multiple heads or a repositioning move.

The control side matters as much as the optics. The signal has to land in the part program as a variable, not on a separate screen. If the operator has to read a number and type it into an offset, the benefit largely disappears.

• 1
  Rotary tablesFix the sensor to the table and calibrate against the rotary datum.
• 2
  Long bedsA 4,000 mm travel may need more than one sensor head.
• 3
  Control linkFeed the reading into the program as a variable, not a screen.