The doctoral thesis presents a custom-designed laser triangulation based metrology system which enables high precision surface displacement measurement of various material types with a single sensor configuration. Laser structuring enables rapid and precise prototype manufacturing which is widely used in the field of electronics and photovoltaics. The process utilizes a tightly focused laser beam with an appropriate wavelength and structuring is achieved by deflecting the laser beam over the material surface. Laser beam divergence, which is governed by the used optical elements, defines the offset from the laser focus plane with an appropriate beam diameter and intensity which satisfy the manufacturing tolerances. Thus, there exists a direct connection between the quality of the manufactured sample and the precision of positioning the material surface regarding the laser focus plane which requires a robust displacement measurement system.
Commercially available contact, as well as contactless displacement sensors are not suitable for usage in the laser structuring application. This is because a fixed vacuum-safety chamber is used during the laser structuring process which removes the excess material and protects from laser radiation. In addition, the displacement measurement needs to be conducted in the center of the laser processing field without occluding the laser beam. In order to satisfy the aforementioned requirements we developed a custom contactless displacement sensor which utilizes an appropriately intensity regulated built-in laser source and an additional camera for acquiring the reflected light. We mounted the camera outside the vacuum-safety chamber, thus not limiting the laser processing filed. The use of a high-quality UV laser beam, which is primarily used for structuring, presents a novelty. In addition to its high pointing stability and homogeneous beam intensity distribution it enables measurement of transparent materials which present the biggest limitation for commercial laser triangulation sensors. The usage of a UV laser beam is therefore appropriate as the majority of transparent materials show high absorption of UV light. The precision of displacement calculation is strongly affected by optical properties and various irregularities of the measured surface as the measurement result is obtained indirectly from acquired signal centroid detection. We decreased the sensitivity of the measurement to the mentioned problems by projecting a symmetrical pattern which increases the active measurement area, thus reducing the effects of local surface irregularities. We implemented symmetrical pattern projection by scanning a laser dot around a closed trajectory where the scanning speed needs to be adequately high. This step introduces and averaging effect of the light intensity distribution across the pattern, as well as reduces the measurement uncertainty due to speckle noise.
In order to robustly measure various material types the intensity of the laser beam needs to be regulated appropriately regarding the optical properties of the measured surface and ambient factors. This process must show high repeatability and objectiveness with no external interaction, thus it has to be automated. The algorithm for setting the optimum laser beam intensity is based on analysis of the acquired symmetrical pattern light distribution.
Even though symmetrical pattern scanning increases robustness to surface irregularities and optical properties of the measured material, repeatable centroid detection of the acquired pattern is required for achieving high measurement precision. This information, equally as in projecting a dot, defines the object surface displacement regarding a reference point. The acquired pattern symmetry offers adequate information for centroid calculation despite potential deformations and even partial interruptions in the pattern. We achieved robust centroid calculation by implementing the double curve fitting algorithm (DCF algorithm). Gaussian curves are fitted to radial cross sections of the acquired pattern in the first step, while in the second step, an ellipse is fitted to positions of the fitted Gaussian curve maximums. This process noticeably decreases the measurement dependence on variations in acquired pattern quality, thus increasing the measurement robustness.
Subsurface scattering, which can be noticed while measuring translucent materials, has a noticeable effect on the measurement result. Therefore, we implemented a compensation method which is based on the pattern symmetry property. The method was designed in reference to verification results of the effect of subsurface scattering on light intensity distribution of the acquired pattern.
A commercial laser structuring system, the LPKF ProtoLaser U3, was used as a test platform for designing and validating the measurement system. We verified the measurement method by conducting various experiments and comparing them to a reference contact displacement sensor. In addition, we implemented and verified the measurement system on a related laser structuring device, the LPKF ProtoLaser S v1.3, which utilizes an IR laser beam instead of a UV laser beam as a working tool. The IR laser beam provides limited possibilities for metrology purposes, hence a built-in laser pointer is used instead. Results of the conducted experiments clearly show that the measurement system proves robust to laser beam intensity variation and enables repeatable displacement measurement of objects with various optical properties with measurement bias lower than 50 μm for all materials.
Additionally, we used a Matlab based simulation to conduct a study of the effects of optical distortions in the acquired pattern on measurement uncertainty. Using this approach we decoupled the measurement uncertainty which arises from interferences in signal acquisition from fundamental uncertainty of the measurement algorithm. We generated three distinct optimal input patterns and used the simulation of various irregularities in pattern acquisition to analyze the relationship between the measurement error and the position of the initial centroid. The simulation results clearly indicate that the measurement results are highly affected by the quality of the acquired pattern.
|
