Calibrations of an analogue probe and error mapping

An analogue probe having a stylus with a spherical tip of radius (r) is calibrated using a sphere of known radius (R) mounted on a machine. The stylus tip is driven into the sphere from a plurality of directions (at least 9), each nominally normal to the sphere surface, until the stylus has deflected a predetermined amount. The machine movement is then reversed, and probe (a, b, c) deflection outputs are recorded simultaneously with machine (X, Y, Z) axis positions until the stylus tip leaves the surface. The readings are extrapolated to obtain the (X, Y, Z) readings when the probe radial deflection is zero. The value of (R+r) is determined from these readings along with the position of the sphere centre giving a value with zero probe errors. Values of (R+r) are also determined using a pre-selected radial deflection for each of the directions, by converting probe (a, b, c) outputs at that deflection to incremental machine (X, Y, Z) axis values using a trial probe transformation matrix. The differences in (R+r) values from the value obtained by extrapolation are noted as an error in each case, and the trial probe matrix is then optimised until the errors are minimised. Using the transformation obtained by the calibration, the probe can be error mapped. Relatively immune to errors caused by slippage of the stylus on the surface of the sphere as calibration artefact.