TY - JOUR
T1 - ProTaR
T2 - An infrastructure IP for repairing RAMs in system-on-chips
AU - Huang, Chao Da
AU - Li, Jin Fu
AU - Tseng, Tsu Wei
N1 - Funding Information:
Manuscript received September 2, 2006; revised March 10, 2007. This work was supported in part by the National Science Council, Taiwan, R.O.C., under Contract NSC 95-2221-E-008-152 and by MOEA Taiwan, R.O.C., under Contract 96-EC-17-A-01-S1-002.
PY - 2007/10
Y1 - 2007/10
N2 - Complex system-on-a-chip (SOC) designs usually consist of many memory cores. Efficient yield-enhancement techniques thus are required for the memory cores in SOCs. This paper presents an infrastructure intelligent property (IIP) for testing, diagnosing, and repairing multiple memory cores in SOCs. The proposed IIP can perform parallel testing for multiple memories, and serial diagnosis or repair for one memory each time. In the repair mode, the proposed IIP can execute various redundancy analysis algorithms. Therefore, the user can select a better redundancy analysis algorithm for each memory core being tested according to its redundancy structure. Simulation results show that the proposed IIP needs less test time and redundancy analysis time than the processor-based built-in self-repair scheme. We also have realized the proposed IIP for four types of memories - two 8 K × 64 bit SRAMs, one 4 K × 16 bit SRAM, and one 2 K × 32 bit SRAM - based on TSMC 0.18-μm standard cell technology. Simulation results show that the area overhead of the IIP is only about 4.6%.
AB - Complex system-on-a-chip (SOC) designs usually consist of many memory cores. Efficient yield-enhancement techniques thus are required for the memory cores in SOCs. This paper presents an infrastructure intelligent property (IIP) for testing, diagnosing, and repairing multiple memory cores in SOCs. The proposed IIP can perform parallel testing for multiple memories, and serial diagnosis or repair for one memory each time. In the repair mode, the proposed IIP can execute various redundancy analysis algorithms. Therefore, the user can select a better redundancy analysis algorithm for each memory core being tested according to its redundancy structure. Simulation results show that the proposed IIP needs less test time and redundancy analysis time than the processor-based built-in self-repair scheme. We also have realized the proposed IIP for four types of memories - two 8 K × 64 bit SRAMs, one 4 K × 16 bit SRAM, and one 2 K × 32 bit SRAM - based on TSMC 0.18-μm standard cell technology. Simulation results show that the area overhead of the IIP is only about 4.6%.
KW - Built-in self-repair (BISR)
KW - Diagnosis
KW - Infrastructure intelligent property (IIP)
KW - Random access memories (RAMs)
KW - System-on-chip (SOC)
KW - Test
UR - http://www.scopus.com/inward/record.url?scp=34648860557&partnerID=8YFLogxK
U2 - 10.1109/TVLSI.2007.903940
DO - 10.1109/TVLSI.2007.903940
M3 - 期刊論文
AN - SCOPUS:34648860557
SN - 1063-8210
VL - 15
SP - 1135
EP - 1143
JO - IEEE Transactions on Very Large Scale Integration (VLSI) Systems
JF - IEEE Transactions on Very Large Scale Integration (VLSI) Systems
IS - 10
ER -