Tolerance Stack-up Analysis By James D. Meadows [updated] ✦ [ INSTANT ]

The stamping process produced flatness variation that was not normally distributed—it was bimodal (two peaks) due to die wear. The RSS method gave a false 99.7% confidence, but actual failure rate was 15%. Additionally, three angular tolerances (folded brackets) created non-linear stack-up that linear analysis ignored.

An automotive sensor bracket assembly had a 15% failure rate during final alignment. The gap between the sensor face and the target wheel was supposed to be 0.5 +/- 0.2 mm. The team had used an RSS analysis, assuming all stamped metal parts were normally distributed. tolerance stack-up analysis by james d. meadows

Meadows is best known for challenging the status quo of the traditional and Root Sum Square (RSS) statistical methods. While these methods are taught in most engineering schools, Meadows argued that they are often misapplied, leading to either over-engineered products or unexpected assembly failures. The stamping process produced flatness variation that was