Using the end discharge of micro-rod-shaped electrode to scan layer by layer, micro-electrical discharge machining (EDM) can fabricate complex 3D micro-structures. During the machining process, the discharge state is broken frequently due to the wear of the tool electrode and the relative scanning motion. To keep a favorable discharge gap, the feed spindle of the tool electrode needs the characteristics of high-frequency response and high resolution. In this study, an experimental system with a macro/micro-dual-feed spindle was designed to improve the machining performance of servo scanning 3D micro-EDM (3D SSMEDM), which integrates an ultrasonic linear motor as the macro-drive and a piezoelectric (PZT) actuator as micro-feeding mechanism. Based on LabVIEW and Visual C++ software platform, a real-time control system was developed to control coordinately the dual-feed spindle to drive the tool electrode. The micro-feed motor controls the tool electrode to keep the favorable discharge gap, and the macro-drive motor realizes long working range by a macro/micro-feed conversion. The emphasis is paid on the process control of the 3D SSMEDM based on macro/micro-dual-feed spindle for higher machining accuracy and efficiency. A number of experiments were carried out to study the machining performance. According to the numerical control (NC) code, several typical 3D micro-structures have been machined on the P-doped silicon chips. Our study results show that the machining process is stable and the regular discharge ratio is higher. Based on our fundamental machining experiments, some better-machined effects have been gained as follows. By machining a micro-rectangle cavity (960 μm×660 μm), the machined depth error can be controlled within 2%, the XY dimensional error is within 1%, the surface roughness R a reaches 0.37 μm, and the material removal rate is about 1.58×10 4 μm 3/s by using a tool electrode of Φ=100 μm in diameter. By machining multi-micro-triangle cavities (side length 700 μm), it is known that the machining repeatability error is <0.7%.
Read full abstract