TECH FOCUS: TESTiNg worst case, then the probability of all five being in the last 25% of their respective tolerance intervals is 6 per billion (0.0235). The figure on the prceding page yields the probability of multiple characteristics aligning at tolerance edges. This figure shows that worst-case testing can yield very low risk if several variables contribute equally to the variation of the undesired event. It can be seen from the figure that 30% or more of the tolerance range defined as worst case provided risk level below one in a million. The key assumption for this figure is that all characteristics contribute equally to the variation of the output. If a single UPD SmartBrief Ad 0622.qxp_1/2 Page Horizontal 4/14/22 5:08 PM Page 1 characteristic dominates the variation of the output, then the tolerance end point is required, but if two characteristics contribute equally, then using the last 10% of the tolerance window yields very low risk. The sensitivity of an output parameter to input characteristics should always be assessed early in the design phase to reduce risk and design verification costs. A substantial test savings can be had, and risk can be managed effectively by understanding the probability of characteristics aligning at worst case. The page-14 figure is easily reproduced for any process capability or multiple processes with different capabilities. n . TO UPDATE June 2022 15https://www.sae.to/update_sblanding https://bit.ly/SAE_SAEUpdate