Application Of The Improved March Algorithm In Memory Exceptional Test

Present memory test is facing challenge with the constant growth of the memory density and the development of the deep submicron VLSI technologies. As the size of the semiconductor craft is getting smaller, the possibility of capture failure is getting higher. Therefore the memory test research becomes more and more important in PC manufactory industry. It is apparent that to choose an effective algorithm and accurate fault model is the key to get to the balance of limited test time and affordable expense versus quality products.This thesis introduces today's popular memory test application and methods. It analyzes the traditional fault model and the test algorithm; it also talks about the March algorithm and its advantage, evolution as well as the features of the improved algorithm. Although all the PC parts have been thoroughly tested before they come to assembly, there are high failure reports from end users. The memory test time increases rapidly with its size. This highly impacts the shipping volume and operation expense, that's why it becomes the bottleneck of the modern computer manufactory industry.The benefits of an improved memory test algorithm——March-TBL are discussed in the thesis in detail, i.e. reduced test time based on current test coverage. However, the primary focus of this thesis is to unveil a new method to optimize the March algorithm by removing the failure classification from the test which is not necessary for manufactory diagnoses. It designes and implements a memory test prototype system based on this improved algorithm, introduces this structure, presents the design specification and execution results in order to test the effectiveness of the algorithm.As an expectation, this thesis suggests that the MBIST test is a trend in the PC memory test, which will significantly reduce test time and control the cost.