Multiple and solid data background scheme for testing static single cell faults on SRAM memories

Memory testing is a method that requires an algorithm capable of detecting faulty memory as comprehensively as possible to facilitate the efficient manufacture of fault free memory products. Therefore, the purpose of this thesis is to introduce a Data Background (DB) scheme to generate an optimal March Test Algorithm (MTA) for detecting faults of memory that are undetectable using existing algorithms. The present research focuses on two types of Static Single Cell Faults (SSCFs): Write Disturb Faults (WDFs) and Deceptive Read Destructive Faults (DRDFs). These faults are undetectable by existing algorithms with insufficient operation. To date, the main effort in this field of research is to improve fault detection by modifying or adding an operation sequence in the MTA. A relatively small number of test approaches have worked on the DB scheme instead of the MTA to improve fault coverage. However, these approaches were designed to improve the fault coverage for detectable faults only. Thus, the present research develops a new DB scheme to be applied to existing MTA to detect two WDFs and two DRDFs. Two methods are proposed in this project. In Method 1, a multiple DBs generator with a bit-adjacent DB management scheme is applied for the selected MTA. This method is evaluated in terms of function and performance differences between the proposed MTA and existing MTA using the User Defined Algorithm (UDA) available in the MBISTArchitect tool. Findings show that both MTAs have the same testing time. However, the existing MTA of the Memory Built-In-Self Test (MBIST) required a bigger area overhead and consumed more power. Hence, Method 1 is not suitable to be used with the MBIST for System on Chip (SoC). For Method 2, suitable solid DBs are used to provide higher fault coverage instead of using the existing MTA. The new MTA is defined by designing an automation program called DB generator. The DB generator computes all the possible DBs and filters the list of preferable DBs using efficient combination logic. The proposed MTA is obtained after the eliminating procedure of the preferable DB list using the SQ generation rule. Finally, the fault coverage will be calculated manually by doing fault evaluation analysis using Fault Primitives (FP) rules. Results show that WDFs and DRDFs are successfully detected with each proposed MTA. The proposed MTAs are also able to detect other SSCFs, such as Transition Fault, Stuck-At Fault, Incorrect Read Fault, Read Destructive Fault, and State Fault. Finally, based on the SQ generation rule, and the development of the DB generator, MTAs are generated. The present research demonstrates that the DB generator and proposed MTAs, such as March CL-1, March Cl-2, March SR-1, and March SR-2, are successfully applied and designed, with up to 100% fault coverage.

Text FK 2013 142 - ir.pdf Download (1MB)

Actions (login required)

View Item