Takaki Yoshida

MD-SCAN method for low power scan testing

As semiconductor manufacturing technology advances, power dissipation and noise in scan testing have become critical problems. Our studies on practical LSI manufacturing show that power supply voltage drop causes testing problems during shift operations in scan testing. In this paper, we present a new testing method named MD-SCAN (multi duty-scan) which solves power supply voltage drop problems, as well as its experimental results applied to practical LSI chips.

Practical application of automated fault diagnosis for stuck-at, bridging, and measurement condition dependent faults in fully scanned sequential circuits

A practical fault diagnosis system based on combination automatic test pattern generation (ATPG) and fault simulation is described. Our fault diagnosis system deals with conventional stuck-at and bridging faults, as well as measurement condition dependent (MCD) faults in order to diagnose those faults causing different behavior with measurement condition such as supply voltage and temperature, using single stuck-at based diagnosis techniques. Experimental results with a practical very deep submicron (VDSM) LSI circuit shows that a defective chip can be efficiently diagnosed using our diagnostic algorithm and newly proposed MCD fault model.

An effective fault simulation method for core based LSI

We examined effective usage of fault simulation to reduce enormous handling time for fault simulation, and applied it in our LSI development. Random sampling method, DFS (Distributed Fault Simulation), a selection of most suitable FPP(faults per pass) and elimination of hyper faults are applied here to realize necessary handling speed of dozens of times faster than the present usage. It is effective in a fault simulation to simulate the best vector first that increases the fault coverage most. Furthermore, we would like to give a new suggestion that the density of mask patterns is taken into consideration as a factor of fault coverage and also its physical correctness is estimated.