E
We utilize Contract Manufacturers to produce a variety (10-15 different assemblies) of printed circuit board assemblies that go into a residential product. During product sampling in our reliability lab we have seen instances of temperature / humidity induced failures when powered assemblies are cycled in an environmental chamber from cold to hot.
Bounded more by capacity (staffing and chamber time) our current sampling process does not use a standard sample size, but we would like to migrate the requirement to our suppliers so that any failures are found as early as possible ... potentially as a production gate ... rather than waiting for Reliability testing.
Does anyone have any experience with environmental testing in a high volume (6,000 - 10,000 boards per day) environment and the use of a sampling plan technique? Obviously, the confidence interval and sample size must be derived from our expected failure distribution and production volume; however, our failure distribution is really multiple small populations of issues as opposed to a defined distribution of defects. Cycle Time and the need to ship product daily are a concern and any use of gating a delivery based upon a sample environmental test is going to be problematic since a minimum cycle might be 6-8 hours to cycle the boards enough to find any solderability or component issues.
I appreciate any insights on testing and creation of a good sampling plan to give good coverage without stopping production to wait for a Pass/Fail on the environmental screen.
Bounded more by capacity (staffing and chamber time) our current sampling process does not use a standard sample size, but we would like to migrate the requirement to our suppliers so that any failures are found as early as possible ... potentially as a production gate ... rather than waiting for Reliability testing.
Does anyone have any experience with environmental testing in a high volume (6,000 - 10,000 boards per day) environment and the use of a sampling plan technique? Obviously, the confidence interval and sample size must be derived from our expected failure distribution and production volume; however, our failure distribution is really multiple small populations of issues as opposed to a defined distribution of defects. Cycle Time and the need to ship product daily are a concern and any use of gating a delivery based upon a sample environmental test is going to be problematic since a minimum cycle might be 6-8 hours to cycle the boards enough to find any solderability or component issues.
I appreciate any insights on testing and creation of a good sampling plan to give good coverage without stopping production to wait for a Pass/Fail on the environmental screen.