Application of a Linear Center Identification Scheme to Deterministic Polar Positioning

In a number of manufacturing applications, parts of circular cross-section must be centered for optimal processing or measurement. However, part form is never perfect, making accurate determination of the “centered” state of a part difficult. Imperfect inputs to the manufacturing process such as roughprocessed parts, deformation due to heat treatment, or raw formed materials present difficulty in centering by the traditional manual method. This paper presents a filtering and quantification technique for identifying the true center of an imperfect round part through isolation of the lowest polar frequency component. A low-cost device is presented that centers parts based on this frequency domain identification of center. INTRODUCTION Precision convex product shapes occur in all facets of manufacturing industry. A primary example is the manufacture of antifriction bearings. Ball and roller bearing manufacturing comprised $5.1B of US manufacturing gross output in 2003 [BEA 2005]. Due to requirements of center-based or radial manufacturing processes, precision parts such as bearing rings can have a centering requirement before the processing or measurement cycle. Such centering ensures both minimization of singlelobe effects in measurement and form matching of outer to inner diameters in material removal processes (minimization of wall thickness variation). A number of techniques are currently employed to center manufactured parts. For low precision applications, hard stops or V-blocks with dimensional offset based on nominal radius are employed, and roundness derived from methods such as the inverse matrix method of Okuyama [2001]. Such methods are subject to tool wear and centering error due to part diameter variations. For precision measurement, an axis adjustment based on the measured offset is used. These adjustments can be a standard manual handwheel type or an automatic system such as the microstage of Liu [2004]. These center compensation methods require a larger capital tooling investment or cycle time burdened at a skilled operator rate. In high-volume measurement or processing applications on the shop floor, manual centering by a skilled operator tapping with a hammer is typically used. A fallacy with manual centering is in the operator’s inability to interpret highfrequency signals from surface imperfections and separate the true off-center value. As centering preprocesses are required not only for finished parts, but also for rough cut or even direct raw materials, automating the centering process in an efficient and cost-effective manner can have a profound implication for high-volume manufacturing.