Shape Measurements of Fast Rotating Objects with Enhanced Speckle Correlation Coefficient

Abstract

Shape measurements of fast rotating rough objects are important for several applications such as in-situ monitoring of workpieces in lathes or roundness determination of shafts. For such tasks the laser Doppler distance sensor technique was invented. This technique is based on two mutually tilted interference fringe systems, where the distance from the sensor to the object surface is coded in the phase difference between the generated interference signals. Large tilting angles of the interference fringe systems are necessary for a high sensitivity of the measurement technique.

However, the correlation coefficient of the interference signals decreases with increasing tilting angles. The signals become more and more dissimilar due to the speckle effect, since different angles of the laser beams generate different speckle pattern. A trade-off between sensitivity and correlation coefficient occurs. It limits the tilting angle between the interference fringe systems to values below 1.5°, which on the other hand results in a low sensitivity of the measurement technique. In order to overcome this restriction a physical modelling and computer-aided simulation of the speckle patterns have been conducted. The outcome of these investigations is a new approach for the detection of the interference signal. The scattered light is received from two different directions matched to the illumination optics, i. e. the tilting angle between the fringe systems is equal to the sum of the main receiving angles of the two photo detectors. By this way, high correlation coefficients have been achieved also for high fringe tilting angles of 7.5°.

Based on the idea of matching of illumination and receiving optics, the measurement uncertainty can be reduced by more than one magnitude. For displacement measurements of a recurring rough surface a total standard deviation of 110 nm were attained at high lateral velocities of 5 m/s. The measurement uncertainty is independent of the surface velocity with a good approximation. Thus, the technique is particularly suitable for measuring fast moving rough objects for example at turning and grinding processes. Due to the simultaneous distance and velocity measurements the diameter, ellipticity, eccentricity and 3D-shape of rotating objects are determined by only one single sensor.