Model-free tool path modification in ultra-precision diamond turning of freeform surfaces using iterative learning control

With the aid of the commercial ultra-precision machine tool, the tool servo diamond turning process provides stable and deterministic material removal, meeting the demands for mass production of high-end freeform surface optics. However, in relatively high-speed applications, the machining accuracy is limited by the heavy servo axes, even within the working bandwidth of −3 dB. Therefore, to facilitate the industrial adoption of diamond turning, a cost-effective and user-friendly programming strategy is essential for enhanced motion accuracy in commercial machine tools. This work proposes a model-free tool path modification strategy using iterative learning control (ILC), which adjusts tool path amplitude iteratively based on the error data of servo axes. By aligning the geometry-based tool path with the dynamic properties of the servo axes, such adjustments reduce tracking errors caused by frequency-based phase lag and amplitude variation effects in high-speed applications. Additionally, this strategy eliminates the need for additional complex equipment or model identification, making it well-suited for industrial applications. The fundamental principle of the proposed method is first presented, followed by a demonstration of its convergence. A series of validation experiments are conducted through trajectory tracking and diamond turning. Experimental results indicate that trajectory tracking achieves a reduction of approximately 60 % in peak-to-valley error and about 80 % in root-mean-square error with the proposed strategy. For diamond turning experiments on sinusoidal grid surfaces, the form error is significantly reduced from 903 nm to 527 nm. Further experiments confirm the long-term effectiveness of the ILC-based tool path modification strategy in high-speed applications, offering valuable insights for industrial use.