A Hybrid In-Field Self-Test Technique for SoCs

Carbonara, Sara; Bernardi, P.; Restifo, Marco

doi:10.1109/dtis.2019.8735075

Cited by 2 publications

(1 citation statement)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…LBIST became alternative approach for ATPG (Automatic Test Pattern Generation) due to increased complexity and test cost requirement for functional blocks [1]- [3]. Electronic component general safety-critical applications [4] are referenced in IEC 61805 [5] and Automotive Safety Integrity Level (ASIL) for automotive domain is referenced in ISO 26262 standard [6].…”

Section: Introductionmentioning

confidence: 99%

Design and Implementation of DFT Technique to Verify LBIST at RTL Level

Vannal¹,

Siddamal²

2020

Adv. sci. technol. eng. syst. j.

View full text Add to dashboard Cite

According to IEC 61805 and ISO 26262 standards requirement inclusion of LBIST (Logic Built in Self-Test) became mandatory to achieve safety critical application such as automotive field. In such systems, once device is switched ON LBIST (Logic Built in Self-Test) is activated and testing of digital logic is performed. After safety subsystem says that the LBIST passed, the SoC (System on Chip) moves into the functional mode otherwise, the SoC moves into a safe state. In this entire start-up sequence the LBIST interacts extensively with the safety sub-system of the SoC. Startup sequence remains un-verified at RTL (Register-Transfer Level) leading to painful ECOs (Engineering Change Orders) and post Silicon issues in some cases. LBIST verification can only run if scan chains are present in design which is not the case at RTL. The paper describes design of a Design-for-Testability (DFT) technique to enable LBIST based system verification with different test approaches at RTL which eliminates the possibility of ECOs by catching most of the issues at RTL level. Simulation results are demonstrating the feasibility of the approach with emphasizing the benefits obtained on significant computational modules.

show abstract

Section: Introductionmentioning

confidence: 99%