Tube laser forming on a small diameter tube can be used as a high precision actuator to permanently align small (optical)components. Applications, such as the alignment of optical fibers to photonic integrated circuits, often require sub-micron alignment accuracy. Although the process causes significant scattering in bending angle and direction, this accuracy can be achieved by multiple bending steps of decreasing size once the target location is in proximity. In this paper an algorithm is proposed which, for each bending step, determines the best values of the main driving parameters; the axial laser spot position on the tube and the laser power. This algorithm is self-learning by using a statistical analysis of all previous bending steps. The algorithm is therefore robust for differences in for example the laser absorption between the tubes. A fully automated experimental micro tube bending setup has been developed using the proposed method in an algorithm to iteratively align an optical fiber to a virtual target with 0.1 μm accuracy, using tubes with a diameter between 450 μm and 700 μm. For the best performing tube geometry, this required an average of 14.5 steps, where each step uses a single laser pulse with a spot position determined by the algorithm.
|Number of pages||8|
|Publication status||Published - 2016|