Premium
Mathematical modelling of test limits and guardbands
Author(s) -
Healy Sandra,
Wallace Michael,
Murphy Eamonn
Publication year - 2009
Publication title -
quality and reliability engineering international
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.913
H-Index - 62
eISSN - 1099-1638
pISSN - 0748-8017
DOI - 10.1002/qre.1001
Subject(s) - datasheet , limit (mathematics) , reliability engineering , product (mathematics) , quality (philosophy) , quality assurance , production (economics) , six sigma , computer science , test (biology) , engineering , electrical engineering , mathematics , manufacturing engineering , operations management , mathematical analysis , paleontology , philosophy , geometry , external quality assessment , epistemology , lean manufacturing , biology , economics , macroeconomics
To gain market share on product performance, suppliers will routinely specify product datasheet (DS) performance to tighter than six sigma specifications. To maintain quality levels in this instance, the product must be 100% tested in production to screen out parts that are outside specification. Because of test measurement error and temperature drift, the test limits (TLs) used must be guardbanded tighter than the DS limit to ensure that the shipped product meets the parts per million requirement. These guardbands are protection zones that protect the customer from receiving parts that are outside specification. A review of current guardbanding techniques finds that they are conservative in nature. New guardbanding models are presented, which calculate the limit sets required for production test, quality assurance test and DS limit specifications. These models account for both test measurement error and temperature drift. Model results show that the guardband required varies with Cpk of the TL, test measurement error and temperature correlation coefficient. Copyright © 2009 John Wiley & Sons, Ltd.