Abstract
Nowadays, in order to meet the requirement from different customers and to confirm functions of their product, network corporation usually do many test before shipping. To gain more profit, they setup more and more test stations to increase production capacity but lead to unstable test result and low yield rate. We further do some analysis to the defective products and find out the problem that is caused not by quality of the products but signal interference in test stations. More than ninety percent of retest products is good product. This may lead to lots of product retest procedure, low production capacity and time cost. In the thesis, we propose a method to lower the retest rate of Wi-Fi product by using six sigma DMADV structure. Firstly, we use QFD (Quality Function Deployment) table to convert voice of customer to design the product specification. Secondly, we use GRR (Gauge Repeatability and Reproducibility) to analyze measurement variation of each test station and to check the key factor of test station. Thirdly, we use TRIZ (Theory of Inventive Problem Solving) theory to activate the concept for designing test station. Fourthly, we use Taguchi Methods to refine the robustness of our method. In this way, we can control the WiFi signal path consistency and reduce multi-path interference. Finally, to verify the method, we do a pilot run. After analyze the report, this method can lower about 10% retest rate and to save test cost by at least 10 million dollars per year.