00330 r8ad54f747f0f287dd69fb4ed228e35

333

Optimizing Defect Levels and Losses from Gage Errors

The central composite design was chosen. See Montgomery (199la) and Box and Draper (1987) for further discussion of this design. Figurę 11 displays the design. With two variables the spacing for the axial point positions is 1.414. The experimental levels for the gage guard bands can be found in Table 1. The ranges for the factors in the experiment were chosen by the test engineering group.

Each of the experimental runs was performed by solving Eąuation (9) for the ppm(Def) and Eąuation (14) for the Gage Losses for each of the combinations of GB1 and GB2. Eąuation (14) was altered to accommodate a two gage system by removing the third gage. The results are shown in Table 2.

GB2

-a

-1

0,0

+ a

+1

GB1

Figurę 11. The Central Composite Design Chosen for the Experiment for This Two Gage Example.