Paul H Chang

US Patent:

6658616, Dec 2, 2003

Filed:

Nov 22, 1999

Appl. No.:

09/444781

Inventors:

Paul Chang - Endwell NY
David Pruden - Endwell NY

Assignee:

Cadence Design Systems, Inc. - San Jose CA

International Classification:

G01R 313183

US Classification:

714728, 714732, 714739

Abstract:

A method is defined which reduces the number of applied test patterns while maintaining identical fault coverage for a given set of weighted random patterns. Reduction is accomplished by simulating the weight sets in reverse order against a full (untested) fault list but allowing fault mark-off for initially effective patterns only (patterns which detected faults). This results in some patterns which initially detected only a few faults (due to their exposure to a small untested fault list) to detect a greater number of faults (due to their exposure to a full fault list), while other patterns which were initially effective become ineffective (detected no faults). Since the same faults are still detected there is no loss of coverage. Only those patterns which remain effective after exercising the reverse simulation are ultimately applied on the tester.