In the summer of 2010, I was selected to participate in the National Institute of Standards and Technology’s Summer Undergraduate Research Fellowship program (NIST SURF). I was assigned to the Machine Tool Metrology Group within the Manufacturing Engineering Laboratory at NIST (now renamed), under Dr. Ronnie Fesperman. My project focused primarily on a novel micro-scale, 3-axis machine tool that had been designed and built at NIST for research purposes. While this machine tool was theoretically capable of positioning accuracy at submicron levels, it had never been formally examined or calibrated. My task was to characterize and compensate for the errors extant in the machine.
My first task was to characterize the machine’s current performance. I used the ISO 230-9 standard (a standard for evaluating uncertainty in measurement when measuring the performance of machine tools) as a reference point for examining the machine’s current performance and characteristics. My initial tests included determination of thermal variation (both due to ambient temperature shift and machine operation), and measurement sensitivity analysis to determine what sources of error would most profoundly affect the results of future tests. Other tasks included developing both kinematic and CAD models of the machine (in MATLAB and SolidWorks, respectively).
My second task was to address the machine’s lack of a consistent reference point for the X and Y axes, a critical feature for successful error compensation. The machine used a high-resolution optical grid plate to determine the position of the X and Y axes, which would lose counts if the table was moved rapidly (for instance, when power to the electromagnetic actuators was cut). I solved this problem by designing a pair of “hard stops” (see photo above). These stops sat outside the work envelope of the XY table, and used magnets to make contact between precision ground balls and kinematic flats with a precise, consistent force, ensuring repeatable positioning.
Finally, I was able to calibrate the machine. Due to time constraints, I was only able to address linear errors in the X and Y directions. However, my calibration of these axes was extremely effective, with positioning accuracy improving from 10 microns to less than 200 nanometers in both axes. I presented my results to the SURF program at the SURF Student Colloquium in August 2010.