| dc.description.abstract | The intent of this work is to improve the machining speed of an existing 3 axis CNC
wood working lathe. This lathe is unique in that it is a modi ed manual lathe that is
capable of machining complex sculptured surfaces. The current machining is too slow for the lathe to be considered useful in an industrial setting. To improve the machining speed of the lathe, several modi cations are made to the mechanical, electrical and software aspects of the system.
It was found that the x-axis of the system, the axis that controls the depth of cut of
the tool, is the limiting axis. A servo motor is used to replace the existing stepper motor, providing the x-axis with more torque and faster response times, which should improve the performance of the system. To control the servo motor, a 1st-order linear transfer function model is selected and identi ed. Then, an adaptive sliding mode controller is applied to make the x-axis a robust and accurate positioning system. A new trajectory generator is implemented to create a smooth motion for all three axes of the lathe. This trajectory uses a 5th-order polynomial to describe the position curve of the feed pro le, giving the system continuous jerk motion. This type of pro le is much easier for motors to follow, as
discontinuous motion will always result in errors. These modi cations to the lathe system
are then evaluated experimentally using a test case. Three test pieces are designed to
represent three of the common shapes that are typically machined on the wood turning
lathe. These test cases indicated a minimum reduction in machining time of 52:91% over
the previous lathe system.
An algorithm is also developed that attempts to sacri ce work piece model geometry to achieve speed gains. The algorithm is used when a certain feedrate is desired for a
model, but machining at that speed will cause toolpath following errors, leaving surface
defects in the work piece. The algorithm will attempt to solve this problem by sacri cing
model geometry. A simulation tool is used to detect where surface defects will occur
during machining and a then the work piece model is modi ed in the corresponding area.
This will create a smoother part, which allows each axis of the system to follow the new toolpath more easily, as the dynamic requirements are reduced. The potential of this algorithm is demonstrated in an experimental test case. A test piece is created that has features of varying di culty to machine. When the algorithm is run, Matlab/Simulink is used simulate the output of the lathe and locate the areas in the part geometry that will cause defects. Once located, the geometry features are smoothed in SolidWorks using the fi llet feature. The algorithm produces a work piece with smoothed geometry that can be machined at a feedrate approximately 42:8% faster than before. Although it is only the first implementation of the algorithm, the experimental results con rm the potential of the
method. Machining speed gains are successfully achieved through the sacrifice of model geometry. | en |
