The implementation of a six sigma program in an organization necessitates several major activities: understanding the design quality of new products as measured by six sigma, knowing the capability of current manufacturing processes, as well as being ready to adopt more capable processes for new products. In this section, each issue will be explained in detail with examples and case studies. Special examples will be given in discipline-specific designs in the next section.
8.2.1 Design analysis for six sigma
When a six sigma program is agreed to in the development of new products, the design team has to consider developing quality measures for all new designs. These include the design quality level for each element to be designed, as well as the quality level for module units and systems. The quality level could be expressed by any of the measures that were introduced in previous chapters, including units of sigma designs, Cpk, DPU (PPM), or FTY. It is important to note that the design quality measure is due only to the design as expressed in terms of component specifications, and not to the physical implementation of the design in manufacturing such as PCBs. The design defects are due only to improper designs, not to any variation in production. These will be calculated separately and combined in an overall new product quality, including design and manufacturing, as shown in Figure 8.5.
The application of six sigma in design is different than in manufacturing, since it will be based on the design components5 specifications and the proper use of components in the design, In order to obtain a good estimate of the quality of the design, the component specifications mast be known, and the design has to be modeled to obtain a distribution of the performance based on the component tolerance distribution.
The six sigma design estimate can be made of typical components as the design nominal and the components worst-case conditions as the specification limits. Components could be modeled as linear or normal distributions of values between the specification limits for one- or two-sided tolerances. Modeling results of Monte Carlo simulation using random selection of uniform data could be used to show a distribution of results of the design versus its specifications. An example of this process is given for a simple bandpass filter (Figure 8.6) whose specifications are described in Table 8.2.
Using the method outlined above, the results of the six sigma quality study are shown in Table 8.3, expressed as Cpk. These results are based on simulation of the design and a Monte Carlo distribution for each component, as shown in Table 8.3. The simulation results are recorded in terms of average and standard deviation for each of the bandpass parameters. Based on the specification and results of the simulation, the Cpk can be obtained for each of the specification parameters, as well as the defects per unit (DPU) and the expected first time yield (FTY). The FTY is based on the design and component se- lection, and does not contain the defects due to the manufacturing process variability.
The total expected quality of the bandpass filter is determined by either adding the defects (DPUs) or multiplying the yields for all of the design parameters. A composite design Cpk for the bandpass filter is calculated to be 0.26. Obviously，this does not meet the six sigma requirements, and selection of the tolerance of the components has to be tightened considerably.