Measurement Machine to Three Coordinates: The Concept of Classification

• 1
  Why three coordinates define every measurement machine
• 2
  Bridge, gantry and horizontal arm classes
• 3
  Contact, scanning and optical probing classes
• 4
  Accuracy classes and the uncertainty budget
• 5
  Matching the class to the part and the print
• 6
  CMM classes compared

A coordinate measuring machine does not measure a diameter or a length directly. It touches or scans a surface, records the position of that point in a Cartesian frame, and repeats the process until enough points exist to define a feature mathematically. Three orthogonal axes are the minimum for that job, which is why the term three coordinates is not a marketing label but a geometric requirement.

The classification of any measurement machine to three coordinates starts from one question: how does the machine carry the probe to the point? Everything else follows from that. The moving mass, the length of the arm, the number of joints and the way the part is supported decide the uncertainty budget before a single measurement is taken.

Points are the raw data. A circle needs at least three points, a plane needs three, a cylinder needs five or six, and a freeform surface may need thousands. Software then fits the nominal geometry to those points and reports the deviation. Accuracy claims without a stated probe, temperature and point count mean very little.

This is also why two machines with the same stated accuracy can behave differently on the same part. One may be faster to set up; the other may hold tighter form on a tall component. Classification exists to separate those cases. The pages that follow describe the classes most often quoted in RFQs and drawings.

The bridge machine is the default in most inspection rooms. A granite table carries the part, and a bridge moves along one axis while the ram moves vertically. Because the bridge is supported at both ends, the moving mass is balanced and the geometry stays stable over a moderate volume. Typical volumes reach roughly 600 × 600 × 600 mm up to 1,200 × 2,000 × 800 mm.

The gantry machine flips that layout. The part sits on the table and the gantry legs travel along rails on the floor, so the measuring volume can be several meters long. This class suits large weldments, die castings, airframe ribs and automotive fixtures. The trade-off is thermal sensitivity: long rails change length with room temperature, so a controlled 20 °C environment is not optional.

The horizontal arm machine reaches into the side of a workpiece rather than down onto it. It is common for car bodies and long panels where a vertical ram would collide with the part. Arm deflection grows with extension, so accuracy falls off as the arm reaches out. Keep heavy work close to the column.

There are also articulated arm CMMs, which use rotary joints instead of stacked linear axes. They are portable and good for in-process checks on large assemblies, but they trade repeatability for reach. For a first article inspection report, a bridge or gantry machine is still the safer instrument.

A touch trigger probe fires a signal when the stylus contacts the surface. Each point is discrete, so the machine moves, stops, measures, then moves again. It is the workhorse for prismatic parts, hole positions and datum features. Point density is low, and a typical inspection of a machined bracket takes minutes rather than seconds.

A scanning probe stays in contact and drags across the surface, collecting points continuously. Form error, roundness and profile become measurable in one pass because the data set is dense. Scanning is the right choice for cams, turbine blades, seal grooves and any feature where the tolerance is form rather than size.

Optical and laser sensors work without contact. They are fast on soft materials, thin walls, sheet metal and parts that cannot be touched. The limits are real: surface finish changes the return signal, steep walls can drop data, and transparent or highly reflective surfaces need preparation. A single sensor rarely covers a whole drawing.

Most modern inspection cells combine two or three sensor types on one machine. A scanning head may carry a touch trigger stylus and a laser line scanner on the same quill, with the controller switching between them. When you write a CMM requirement, name the sensor, not just the machine class.

Machine makers publish an MPE value, usually written as a formula such as MPE_E = A + L/B, where L is the measured length in millimeters. A single number quoted without the formula is not comparable between builders. Ask for the formula and the temperature range it applies to.

Accuracy grades split into roughly three bands. Standard shop floor machines sit in the middle, suitable for general machining inspection. High accuracy machines reach the low micrometer range and are used for gauge work and optical components. Ultra high accuracy machines are lab instruments with active thermal compensation, and they demand a controlled environment.

The stated machine accuracy is only one term in the budget. Probe stylus bending, part temperature, clamping distortion, surface roughness and the number of points all add error. A part measured straight off a machine tool can be 5 °C above room temperature; on aluminium that alone shifts a 200 mm dimension by tens of micrometers.

So the practical question is not which machine is most accurate. It is which machine, in which room, with which probe, can demonstrate the tolerance on the drawing. That is the only classification that survives an audit.

Start with the largest dimension on the drawing and add clearance for the probe and stylus. If the part is 400 mm long, a 500 mm class machine is already tight once you allow for approach moves and probe length. Most programmers want at least 30 percent headroom over the part envelope.

Then look at the tightest tolerance. A general machining callout of ±0.05 mm is comfortable on a standard bridge machine. A ±0.005 mm position callout on a bore pattern is not; it needs a high accuracy machine, a temperature soak, and a probe with a short, stiff stylus.

Weight and fixturing decide the rest. A heavy casting on a granite table is stable. A thin-walled housing clamped in a vise can move more than the feature being checked. Sometimes the correct answer is a dedicated fixture, not a better machine.

Finally, consider throughput. If the inspection is a production gate rather than a one-off first article, cycle time matters as much as accuracy. A scanning head that sweeps a profile in one pass often beats a touch probe that fires a hundred points, even if the touch probe is nominally more accurate.