| The thin film disk head is an important component in a computer hard disk drive. Its critical dimension, stripe height, needs tight control during fabrication. This is achieved by the bar lapping process. An array of heads along with the resistive sensors form a bar. They are lapped simultaneously. The objective is to achieve minimum final stripe height variation.;The bar lapping is a multi-input and multi-output control problem. The existing control method assumes there is no coupling among the inputs. Proportional-integral (PI) controls are applied independently. As technology advances, tighter tolerance is required. The existing control method becomes inadequate. Therefore, it is imperative to establish a fundamental understanding of the lapping process, and define the lapping problem in a broader sense. This is the objective of the present thesis. The resulting methodology can serve as a tool for designing a lapping system, re-assessing an existing lapping system or suggesting new methods of control.;The theoretical development of this thesis can be divided into three stages: process simulation, control model construction and controller design. By design of experiment, the Preston equation is identified to be a reasonable model relating lapping rate to pressure and relative velocity. This constitutive relationship is used in the simulation stage. A pressure estimation method is established based on the modified beam on elastic foundation model to incorporate the effect of surface shape. Computer simulation of the existing control method shows consistency to production observations. In the second stage, the simulation model is used to identify the structure of the lapping process model for the purpose of control. The open-loop solution is compared with the simulation results to validate the process model. Finally, a controller is designed based on trajectory following.;The proposed control method is implemented on an existing Seagate machine. Experiments are conducted to validate the methodology. Results show that the proposed method is less sensitive to the quality of the preparation processes. However, the gains should be appropriately adjusted for it to perform optimally. With good preparation processes and appropriate gains, performance of the two methods are statistically comparable. Otherwise, the proposed method performs better. The proposed method is inadequate only when its gains are improperly adjusted but the preparation processes of the existing method are good. |
