Kinematic Modeling and Analysis of a Planar Micro-Positioner | NIST

1 Fig. 1 Picture of a PCBMP stage with dimensions. 1. Introduction The tremendous growth of opto-electronic devices manufacturing has raised the need for high performance micro-positioners, which are used in their assembly and alignment. The static and dynamic performance of micro-positioners depends in part on the quality of operation of their controllers, which in turn depends to a significant degree on the accuracy of the kinematic and dynamic mathematical models upon which they are based. The objective of this work was to evaluate the accuracy of two kinematic mathematical models of a planar X-Y micro-positioner under a variety of testing conditions. A fully instrumented X-Y Parallel Cantilever Bi-axial Micro-Positioner 1 (PCBMP), was used for the tests (Fig. 1). The performance of the stage was evaluated by measuring the output motions as a function of input displacements over a two-dimensional grid of test positions. Second-order and linear kinematic models were developed through an examination of the stage geometry. The performance test data was used to identify the unknown parameters of the kinematic models using least-squares fitting algorithms. The accuracy of the fits was examined under the conditions of varying the number of input test data or adding white noise. 2. Stage Performance The static performance of the stage was evaluated by e xamining its input and output displacements while moving it throughout its motion range of about 130 input actuators are servo-controlled based on built-in capacitance gage sensors, thus eliminating the typical piezoceramic effects of hysteresis, non-linearity, and creep. 2 At each test position, along with the input position commands and capacitance gage data, we also recorded six output data: the X and Y position of the output stage, measured by separate capacitance gages 3 ; and the angular orientation of the output stage in all three angular axes, with a redundancy in yaw (rotation about the Z-axis). The angles were measured using two orthogonally placed, two-dimensional, high-resolution autocollimators, which tracked a cube reflector on the stage. A LabView 4 program was written to scan the stages and read the data. The entire array of data was swept multiple times over a period of several hours to verify repeatability and correct for drift. In initial tests, the motion actuators were coupled to the stage through a single universal joint (SUJ) on one end, the other end being hard mounted. The universal joint is a flexure device made from cross-bored pairs …