Issue link: https://iconnect007.uberflip.com/i/455818
76 SMT Magazine • February 2015 leads in the test. Given the exposed edge area of these leads, this corresponds to a whisker densi- ty of 0.0018 whiskers per square mm. Recogniz- ing that this density is extremely low, a higher whisker density of 100 whiskers/mm 2 (~50,000x higher than tested) was also used in the model. SEM analysis of the whiskers in this study found that the longest observed whisker was ~20 mi- crons in length. This was then used as a baseline for estimating lengths for a lognormal distribu- tions with the 80% long whisker length set to 1, 2, 5 and 10x this length (20, 40, 100 and 200 mi- crons) and the 7% short whisker length set to be 25% of the long value. These parameters are ad- mittedly somewhat arbitrarily selected, but with the paucity of tin whiskers observed on these components there was limited data with which to justify alternate values. Because no whiskers were observed or expected on soldered areas of the connectors, the 'solder' and 'pad' whisker densities for the model were set to zero and only the 'lead' whisker parameters were adjusted. The connector had 30 rows of 6 leads, which resulted in a total of 168 sets of adjacent edge surfaces that could potentially be shorted with a tin whisker. The model estimates the number of shorts that would occur within a given connec- tor. In this analysis, the voltage between leads was assumed to be 5V; a larger voltage differ- ence would be predicted to lead to more shorts resulting from whiskers making contact with an adjacent surface. Figure 14 shows results of the model for the two whisker densities (0.0018 and 100 per mm 2 ) for both a full (unsoldered) connector and the portion of the connector that is not poisoned by solder. These results are depicted in terms of the inverse of the shorts per connector—which indicates the size of the population of connectors needed to expect one to have a tin whisker induced electrical short. For the present test results (with extremely low whisker density and only short whiskers being observed), this plot shows that the risk of a tin whisker electrical short is exceedingly low (less than one per trillion connectors). If whiskers grow to be 10x larger than those observed in this study, but have the same low density, the probability of a short falls to ~ one in a million. If whiskers formed with a much higher density and did grow to be significantly longer than the ones in this study, the model ultimately predicts that virtually all connectors would experience a tin whisker electrical short. To effectively use the model described in Reference [11] for these connectors, the model should be explicitly adapted to allow the user TIN WHISKer rISK aSSeSSMeNT OF a TIN SurFaCe FINISHed CONNeCTOr continues Feature Figure 14: estimated inverse connector failure rates.