Precision machine tools such as diamond turning machines for manufacture of precision optics require deliberate diagnosis to ensure that all the machine elements are properly operating, kinematically, dynamically and thermally, to produce demanded work qualities. This task could demand element-by-element machine examination, which is practically not feasible since quite a few machine elements have to be considered with complicate testing strategies. Instead, one effective way is to directly inspect topographical features of machined surfaces that have been carefully generated with prescribed machining conditions intended to exaggerate faulty consequences of any ill-operating machine elements. This technique not only provides a fast competent means of machine diagnosis but also identifies intricate environmental effects of temperature variation and external vibration.
The idea of machine diagnosis through work surface examination is in fact not new, as several pertinent attempts have been made during last decades. The consequence was however not quite advantageous, the main reason being the lack of satisfying means of surface profile metrology. Stylus type and optical profiling instruments have long been available, but their performances are still not adequate for the purpose. For appropriate machine test, machined surfaces need be measured not only in three dimensions but also over a large area to cover the whole working strokes of machine axes. In addition, the measuring resolution should be in the nanometric region in the vertical dimension, and the sampling interval in the spatial dimension should be at least in the micrometer regime being fine enough not to lose high spatial frequency surface information for necessary dynamic analysis. To fulfill the metrological requirements for surface analysis leading to machine diagnosis, presented in this paper is a very-large-scale phase measuring interferometric system. The measurement system comprises an optical probe and a precision stage. A special stitching technique is used to integrates all the patches that are separately sampled over the whole surface while moving the stage. The stitching minimizes the total sum of mismatching errors of the sampled patches, so that the motional errors of the stage are not transferred to the finally reconstructed surface profile.
Then, the measured surface profile is analyzed to identify any faulty machine elements. For this, fourier transform technique is used together with the wavelet transform, which extracts any abnormal features and classifies them into three groups, which are terms as the time based frequency, position based spatial frequency and spindle rotational speed based frequency. The abnormal frequencies are related to actual machine components through a suggested surface generation model that predicts the relationships between the cutting conditions with the actual surface profile. Experimental results prove that the method under investigation in this study is very effective in diagnosing actual diamond turning machines being used for manufacture of aspheric optical surfaces.
