A Comparison of Statistical Distribution Functions to Predict BGA Attach Reliability

This paper presents a comparison of using different statistical distribution functions to predict ball-grid-array (BGA) solder joint reliability. Four functions are evaluated: two-parameter Weibull, three-parameter Weibull, mixed two-parameter Weibull and Lognormal. Statistical analysis demonstrates that the most commonly used two-parameter Weibull function may not present the best fit to the failure data; three-parameter Weibull and mixed Weibull may be better options in some cases by examining failure data histogram, data fit goodness, and potential failure mechanism change.