Narrow Results

The two prerequisites for the referential-based image comparison method, perfect registration and stable lighting, are hard to achieve under real PCB inspection environment. In this paper, an objection description based on learning for PCB inspection is proposed to compensate for allowable minor variations of object placement and lighting variations, and the threshold image approach is provided to effective avoid false defects. The experimental results show that the developed method is successful in locating defects occurring on PCB, and the result can be used to quality control.

The dynamical model for a drawer structure with multi-layer printed circuit boards (PCBs) was established by the finite element method and the random vibration responses were analyzed in this paper. Firstly, the natural frequencies and their corresponding modes were obtained in terms of the finite element model of the drawer structure with PCBs. Secondly, the random vibration responses were obtained numerically for the drawer structure under random excitations. Finally, sensitive points on each layer of the drawer structure were selected to calculate the vibration responses and the acceleration power spectral density (PSD). The stress power spectral density of typical points on the PCB with high stress was also analyzed. The simulation results reveal that the largest vibration energy of the PCBs usually occurs in the middle layer of the structure and thus components in the middle of the PCBs are most likely to be destroyed.