A genetic algorithm for probe testing problem on high-density PCB

Printed circuit boards (PCBs) are designed for packing two or more semiconductor chips. On these PCBs, there are various types of faults in the wiring. We must therefore test the PCBs to detect these faults, and it is essential to establish an efficient testing method. One type of test method uses two probes. Two probes, each touching one edge (end) of an inter-chip wiring, are used to check for the presence of faults. Testing is complete when we have confirmed that no faults exist on the PCB. The objective is to minimize the completion time for the testing, that is, our aim is to design efficient routes for the two probes; this is a two-probe routing problem. Several previous studies have proposed an approach involving a two-phase method. Although the two-phase method quickly finds a feasible solution, there is no guarantee that the solution is good. Besides, in recent years, PCB technology provides dense assembly capability and increased chip packaging; this implies that the scale of the two-probe routing problem becomes larger and the problem is difficult to be solved. We therefore propose a new approach, where we formulate a two-probe routing problem as a combinatorial optimization problem and then, we solve the combinatorial optimization problem by using genetic algorithm.We expect that our approach is useful for solving the large-scale problems. In computational experiments, we compare the results of our approach with those obtained using the two-phase method, and we show that our approach outperforms the two-phase method.