Accession Number : AD1046444


Title :   Reliability Assessment of a Single-Shot System by Use of Screen Test Results


Descriptive Note : Journal Article


Corporate Author : NAVAL UNDERSEA WARFARE CENTER KEYPORT DIV WA KEYPORT United States


Personal Author(s) : Coate ,H ; Skaggs,M ; Paulus,M ; Kamel,S


Full Text : https://apps.dtic.mil/dtic/tr/fulltext/u2/1046444.pdf


Report Date : 01 Feb 2018


Pagination or Media Count : 18


Abstract : Field reliability prediction methods based upon early screening results typically involve tracking a temporal metric such as on-time across a constant stress testing regime in order to model wear-out. These methods have very limited applicability to single-shot systems because reliability is not driven by wear-out, and testing is often performed at varying stress levels. A new methodology is introduced to track and project the reliability for a single-shot system as it goes through a multi-staged screening process that produces no meaningful temporal metric and involves significant differences in test strength. The approach described here assumes that the defect density during testing takes the form of an exponential decay, although other mathematical functions can be substituted for the exponential. In order to apply the decay rate function to a discrete pass/fail test scheme, the approach provides for normalization of the disparate tests to constant stress by back-calculating and adjusting for test strength based upon previous screening results. This approach is more useful when reliability does not involve wear-out of parts, which is typically true of single-shot systems. However, it also potentially has utility for all programs that need to glean information about early failures caused by fabrication problems, so long as a discrete end point for the reliability requirement such as warranty termination has been established. The equations provide a tool with which reliability practitioners can estimate field reliability of a new lot of single-shot or warrantied systems based upon early screen results, as long as a complete set of data from previous lot testing is available. Utility for reliability growth estimation is also described. A numerical example is given to demonstrate application of the model. This paper was adapted from a limited-distribution technical report by the same authors (Coate and Skaggs, 2016).


Descriptors :   RELIABILITY (ELECTRONICS) , TEST METHODS , failure analysis , FAILURE (ELECTRONICS) , RELIABILITY ENGINEERING , LINEAR REGRESSION ANALYSIS


Distribution Statement : APPROVED FOR PUBLIC RELEASE