Carnegie Mellon University
Multiple Defect Diagnosis
Experiments have demonstrated cases where failing circuits
are affected by multiple defects. With increasing device and
interconnect density, coupled with more pronounced variability,
it is likely that more and more failing circuits will be affected
by multiple defects. In order to increase yield and improve
design quality, it is important to understand why circuits
fail. Among current multiple defect diagnosis methodologies,
some make restricting assumptions about defect behaviors.
These methods cannot effectively diagnose circuits affected
by defects with different behaviors from what the methods
assume. Other multiple defect diagnosis methods make restricting
assumptions about failing pattern characteristics. These methods
fail when no or few failing patterns with the assumed characteristics
In this talk, a generalized approach for diagnosing multiple
defects based on a new error propagation analysis is presented.
In this approach, no assumptions are made about defect behaviors
or failing pattern characteristics. Results from simulation
and silicon experiments are used to demonstrate the effectiveness
of this new multiple-defect diagnosis methodology.
Xiaochun Yu received her B.S. degree in Microelectronics from Fudan University in 2005. She is currently a Ph.D. candidate in the department of Electrical and Computer Engineering at Carnegie Mellon University, where she is advised by Professor Shawn Blanton. Her research interests include various aspects of VLSI testing and diagnosis.