The role of quality engineering is to improve speed of R&D work by using SNR and OA as explailled in Chapter one. However, R&D workers are very busy if they conduct their work for each product plan. Quality engineering deals only functionality. So we can develop robustness of basic technology for a family of products. This chapter explains the following two strategies: (1) Develop machine function to produce various kinds of parts with simple specimen instead of gears and parts used for any product. See first half of this chapter. (2)Use generic function instead of objective function. Taking engine hardware/software function, we should use chemical reaction speed but not transducer function, fuel energy to mechanical energy.

Other quality engineering strategies to develop robustness using only computer software will be introduced next chapter.

Recently, as for DC motor used for a power tool, small and high output are progressing by magnetically high efficiency. On the other hand, magnetic torque change depended out of balance becomes large, and problems, such as vibration and noise, often occur. When this torque unevenness is coped with, it is in the tendency for a motor performance to fall generally. Although reduction of the torque unevenness by experiment is reported also before, the examination by the simulation is not reported. Then, this research estimated torque unevenness reduction by the simulation in the standard S/N ratio. Consequently, the optimal combination, which reduces torque unevenness to 1/6, was found out,with the motor performance maintained. And the effect of each design parameter has been grasped and it can be reflected in the design.

The purpose of tolerance design in the Taguchi methods is a judicious balance of quality and cost. A major effect obtained from tolerance design is an anticipated cost reduction and improved productivity through the appropriate relaxing of tolerances of factors having small effects. Few case studies in which tolerance design was used have been reported, however. The reason is the difficulty of performing experiments, because of the difficulty of creating experimental samples with the width of the tolerance controlled at both ends. Recently, there have been many research reports in which Taguchi methods are combined with numerical simulation. In a numerical simulation, it is easy to create an experimental sample with the width of the tolerance controlled at both ends. A larger number of such research reports would bolster the significance of tolerance design. In the present study, tolerance design for the parameters of an optical filter was carried out by numerical simulation, and factors enabling cost reduction were found.

Polygon motors procured from parts manufacturers are installed in large numbers in copiers, printers, etc. A technique for quickly evaluating their structural merits was developed by using the functional evaluation methodology of quality engineering. The polygon motor structure was the control factor. Noise affecting motor performance was used as a noise factor, and stability under the influence of intentionally applied noise was analyzed and evaluated. Besides being able to indicate the structural merits of polygon motors by their stability with respect to noise affecting motor performance under product usage conditions,this technique was capable of predicting relative reliability in an evaluation period of only a few days by including motors with well established field performance records as benchmarks in the group of motors evaluated.

The evaluation of the results of research and development is a crucial issue in R&D management. Results achieved by development departments are often evaluated as to whether the performance of the finished product or technology meets quality targets. Because quality engineering lays weight on the basic research and development process, however, it is not readily amenable to evaluation by conventional quality target methods. To promote corporate research and development activities by aggressively using quality engineering, managers must be able to evaluate results in the research and development process appropriately. This report proposes a method in which results are evaluated by the gain of the S/N ratio and the evaluation is linked to financial rewards, By trying this method out in their own department, the authors were able to demonstrate the possibility of evaluating R&D activities quantitatively and objectively by using the gain of the S/N ratio as an index value. Engineers and management-level personnel inside and outside the company were surveyed for their opinions on this proposal. Greater approval of result evaluation by gain was obtained from personnel at companies extensively using quality engineering. This survey also brought to light problems to be overcome before this result evaluation method can be formally systemized, however. The concludinig part of this report summarizes these problems and bonsiders possible solutions,

On-site conditions do not always permit quality engineering experiments to be perfbrmed. It is therefore important for experiments to be possible without making extensive use of company equipment, In the present study, experiments in automatic image recognition were carried out to explore methods of research that are feasible in any situation. The MT system was used to distinguish characters and pictures in an image, and the results were evaluated. Circumstances demanded that the experiments be performable at home. This problem was solved by the use of a scanner connected to a personal computer.