Tolerance Stack Up
Tolerance stack-up is the cumulative variation from multiple part dimensions, geometric controls, setups, fixtures, and assembly interfaces in a dimensional chain. In CNC machining, the critical concern is not individual feature tolerance, but whether the combined effect of all tolerances still meets the functional requirement for fit, gap, alignment, or clearance at the final assembly interface.
On the CNC shop floor, stack-up analysis is performed before production release to predict whether features like hole patterns, pocket depths, or cabinet reveals will assemble correctly despite real-world variation from tool wear, clamping, thermal growth, datum error, and inspection methods. The process begins by identifying the tolerance loop or dimension chain between two critical functional points. Tolerances are then summed using either worst-case or RSS methods, depending on risk and process capability. Worst-case assumes all tolerances drift to extreme in the same direction, while RSS assumes random variation and uses root-sum-square. This analysis is especially important when tolerances are distributed across multiple setups, multiple parts, or long dimension chains, as small errors accumulate. Even when every individual feature passes inspection, the accumulated error can cause interference, excessive clearance, or misalignment. In millwork, the same principle applies to panel sizing, groove placement, edge trim, and hinge locations, where small deviations at each step produce visible assembly defects.
- Weak datum planning: An inconsistent datum reference or changing setup order between operations increases the error chain, causing final feature location to drift outside the functional envelope even when each operation appears acceptable on its own.
- Fixture movement or clamping deformation: Excessive clamp force or poor support bends the part during machining; after release, the part springs back, creating hidden stack-up errors that cause assembly failure despite in-process measurements.
- Thermal and inspection mismatch: Tool heat, spindle growth, or measuring parts at different temperatures than production conditions leads to apparent passes on individual features but functional fails in assembly due to accumulated chain mismatch.
Why can a part pass inspection and still fail assembly?
Each feature may be within its own tolerance while the sum of those variations exceeds the allowable functional limit at the assembly interface, causing interference or misalignment.
When should worst-case stack-up be used instead of RSS?
Worst-case is appropriate when failure is unacceptable and every tolerance could realistically align in the same direction; RSS is used when variation is random and the design can tolerate probabilistic risk.
What does stack-up analysis have to do with GD&T?
GD&T controls form, orientation, position, and profile, which often matter more than raw size when the functional requirement depends on location or alignment; poor datum strategy can worsen stack-up even if linear dimensions are tight.