Proceedings Article | 27 July 2019
KEYWORDS: Sensors, Doppler effect, Error analysis, Manufacturing, Metrology, Optical sensors, Laser optics, Spindles, Stochastic processes, Analytical research
Between 50 to 60% of tasks in manufacturing metrology are measurements of cylindrical parts. The tasks are accomplished by roundness testers, which feature high-precision displacement sensors and rotary tables. This paper describes a high-precision method for evaluating roundness and diameter of the cylindrical parts, where a novel laser Doppler sensor and error separation techniques are employed.
Nowadays, the most commonly employed sensors in the roundness testers are still contact stylus, which might damage the machined surface. Non-contact capacitive sensors offer sub-nanometer accuracy, but suffer from low lateral resolution. Therefore, we employ a multi-functional optical sensor, the laser Doppler distance sensor with phase evaluation (P-LDD sensor), for the roundness measurement. The P-LDD sensor offers a high lateral resolution, low uncertainty, and also, can determine the diameter simultaneously.
Apart from the distance sensor, the rotary table also plays a critical role in the roundness measurement. Its error motion, always leads to systematic deviations. Error separation technique (EST) can separate the spindle error motion from the roundness, thus, cancelling the systematic deviation. Regarding this technique, substantial research effort has been paid, especially into the harmonic suppression problem, which has long been regarded as the dominant factor affecting the measurement accuracy. Nevertheless, even today the ESTs are only sparsely represented in industry and still under research and development. We suspect that a shift of the research focus from the harmonic suppression problem to the measurement uncertainty propagation will yield the foundation for an eventual solution to the measurement accuracy problem, and thus, bring a new paradigm for the EST. Therefore, by means of the stochastic spectral method, we analytically derive the propagation law of the measurement uncertainty within the two-step error separation method (TSM), which is subsequently validated by Monte Carlo simulation. Based on the propagation law, three improved TSMs are further put forward for reducing the uncertainty propagation: the angle-optimized TSM, the hybrid TSM, and the fusion TSM. In the angle-optimized TSM, the angle is optimized to minimize the measurement uncertainty. In the hybrid and the fusion TSMs, two measurements are performed first under different angles; then, the two obtained estimations are hybridized or fused in the harmonic domain, which decreases the measurement uncertainty significantly. Finally, by using the P-LDD sensor and the improved TSMs, test measurements are performed and the results are discussed.
