Error compensation is a cost effective way to improve the accuracy for CNC machine. Repeatable error is estimated and compensated so that the error is reduced without investment on hardware with higher precision level. Nowadays, most of the compensation methods try to monitor different components of errors such as tool size, temperature, and cutting forces. These methods require installing expensive sensors and large measurement overheads. In this research, however, we focused on studying the error behavior by machining testing parts and developing an aggregated error compensation technique that aimed to reduce the needs for expensive equipments....[ Read more ]

Error compensation is a cost effective way to improve the accuracy for CNC machine. Repeatable error is estimated and compensated so that the error is reduced without investment on hardware with higher precision level. Nowadays, most of the compensation methods try to monitor different components of errors such as tool size, temperature, and cutting forces. These methods require installing expensive sensors and large measurement overheads. In this research, however, we focused on studying the error behavior by machining testing parts and developing an aggregated error compensation technique that aimed to reduce the needs for expensive equipments.

Previous work showed that a partial aggregate error approach using the tool path information from the encoders in CNC machine could be use to compensate the error in 2D repeated milling. And the similar methodology could apply to an aggregated error approach on pockets with simple shape by offline measurement. In this research, the focus had been extended to general polygonal and circular shape pockets. Experiments were conducted to characterize the error predictability and the result showed improvement on accuracy compared with the partial aggregated approach.

Further work on economization of this technique found that there were some alternations of test pattern and compensation process that could gain productivity while losing limited amount of accuracy under different situations. From the results we found that not only could we predict the errors of general polygonal and circular shape pockets by test pattern with similar geometry and different scales, we could also predict a complex part by a simple rectangle if the part had been produced for number of pieces in a certain way. This finding provided us a technique to easily update the error information during the production when tool was changed or tool’s accuracy was degraded over time.

Previous work had studied the relationship between the angles and the dimensional errors of the hypotenuse of right trapezium in order to predict the error of pocket with different angles without machining the test pattern with the same shape. In this research, a model named “Effective Tool Error Model” was developed, from which the errors of a general polygonal shape pocket with arbitrary angles could be predicted with the test pattern of a specially designed five-sided pocket. This model further simplified the initial error estimation process and reduced the time and cost significantly especially for the polygonal pockets with large number of edges and different angles. In the verification test, most of the dimensional errors of a new pocket could be controlled within 30μm after error compensation, and 2 times of improvement in accuracy compared to anticipated error was realized.