Statistical Boundary
Statistical quality control boundaries separate common cause variability from assignable cause variation on manufacturing control charts. Surface mount assembly lines rely on shewhart limits to monitor deposit height, solder paste area, and component placement coordinates. Calculated at three standard deviations above and below the process average, these boundaries define the expected range of stable machine performance.
The three-sigma convention balances the economic cost of investigating false alarms against the risk of overlooking genuine machine degradation.
Process Surveillance
Automated inspection tools generate real-time control charts to maintain line performance within statistical boundaries. When paste volume readings transgress established shewhart limits, line operators halt printing to clean stencils or inspect squeegee mechanism pressure. Random natural variations fluctuate harmlessly between the upper and lower control bands, signaling that the printing or placement operation remains in statistical control.
Systematic shifts, tool wear, and paste dry-out manifest as out-of-bounds observations or non-random run patterns. Surface mount capability depends on distinguishing these operational shifts from baseline process noise. Unlike engineering tolerance specifications, these boundaries reflect actual machine capability rather than contractual design limits.
Operational Scope
Non-normal distributions reduce the mathematical efficiency of standard three-sigma calculations, leading to unexpected false alarm frequencies. Skewed paste volume data sets require data transformation or rational subgrouping before applying shewhart limits to automated inspection results. Small sample sizes impair the accuracy of calculated standard deviations, destabilizing control lines during initial production setup runs.
Autocorrelated process data produced by high-speed continuous processes can generate misleading out-of-control signals on automated assembly lines. Control boundaries must be recalculated following tooling changes, maintenance overhauls, or raw material formulation revisions. Operating procedures mandate immediate root-cause investigations whenever consecutive subgroup points violate statistical rules.