| Summary: | Memory testing is significantly important nowadays especially in SOC’s design, due to their rapid growth in the memory density and design complexity in smaller chip
area and low power design. Thus, test time in memory testing is a key challenge to accelerate time to market, high yield and low test cost in high volume manufacturing. Test time reduction in memory testing is important in industry, as test cost is directly related to validation time of each product on the tester. There are lots of memory algorithms used for memory testing, including the galloping column algorithm (GalCol). The GalCol algorithm test is important to detect unique coupling and transition faults. However, the existing GalCol algorithm takes huge test time due to its test complexity. To overcome the test time issue in industry, reduced GalCol algorithms with solid data background are
proposed. The reduced GalCol algoritms have similar test behavior as original GalCol algorithm with major difference in the number of galloping of the target cells. The galloping of target cells are reduced to first and last 8, 16 and 32 of cells of every base cell. This project is progressed in two stages, which are the software development using INTEL software and Synopsys tool and test implementation on INTEL production flow.
These algorithm are verified on 15 units of 64KB L2 SRAM memory. In this project, test time reduction and consistent pass fail test results are achieved in the reduced GalCol algorithm tests. The GalCol X8 algorithm obtains the highest test time reduction of about
79.5% at 600MHz and 75.7% at 1.6GHz with consistent pass or fail test results comparable to original GalCol algorithm in the HVM test flow.
|
|---|