Research On Improvement Of Production Rate Of Wireless Base Stations In ENC Company Based On Six Sigma

The business markets of today are continuously becoming more globalized making the quality of delivered products and services more important for companies to achieve customer satisfaction and long-term business growth. One essential part of quality work is to reduce the costs of poor quality which is partially represented by yield loss. The purpose of this thesis was to investigate why yield loss occurs in manufacturing of radio base stations at Nanjing Ericsson Panda Communications Ltd. and how it can be reduced. The work of the thesis was limited to one product group and carried out as an improvement project in accordance to the five phases of Six Sigma’s DMAIC (Define, Measure, Analyze, Improve, and Control).In the define phase the average yield loss for the last weeks in 2015 was calculated as 84 percent and compared to the target level of 90 percent in 2016. A high-level understanding of the assembly and testing process was also created through a SIPOC map. The measure phase established a more detailed understanding of assembly and testing by creation of process flow charts for the whole process. A more detailed understanding of the test procedure was also developed by mapping the different connection interfaces between the radio base station and the test station. This description also made it clear that the test can be divided into test groups consisting of test points which in turn consist of individual test cases constituting measurement values and limits. The flow of information around the test process was also mapped to provide knowledge of where necessary data could be found and collected.The analyze phase started with a breakdown of yield loss where measurement values categorized as outliers were found to represent a large portion of yield loss. Looking into the outliers-category, radio base stations with no fault code given (products that have failed the first test and passed a retest without any action) was identified as interesting. Through brainstorming and an affinity diagram workshop, about causes to outliers with no fault code given, connection issues and different measurement values were highlighted for analysis. Further construction and analysis of Pareto charts showed that some interface connections and measurement values were more frequently involved in outliers with no fault code given. By looking into test points for relevant interface connections and combining these with measurement values the test point External Alarm Check was found relevant for root-cause analysis.An initial time series analysis of the occurrence of External Alarm Check did not lead to any clear root-causes. Through further investigation of the problem together with a test engineer in the production it was identified that the test point failure was caused by improper connection of two cables used during the test. Further analysis of what could cause this improper connection indicated problems with missing connection checks, routing of cables, fixing of test equipment, connector design, and maintenance routines. The analysis of this test point also highlighted problems related to unclear working routines and insufficient documentation of actions for products that pass a retest.Based on the findings the improve phase focused on developing improvement suggestions to minimize the impact from identified issues. ENC were recommended to stabilize and align the test station environment, use connection checks for all test points, develop robust connectors, evaluate maintenance routines, standardize retest procedures, and ensure documentation of actions. At last the control phase emphasized the importance of continuous monitor and execution of the improvement suggestions.