DPMO and OMI are good tools to calculate PCB assembly line yield and to compare and benchmark electronic PCB assembly in the supply chain. Issues that arise with the implementation of the DPMO and OMI indices might be as follows:
•Confusion over the utility of both functions. DPMO is easier to calculate than OMI and therefore will become the more commonly used function.
•DPMO/OMI deployment will require extensive training of assembly labor as well as management and support staff such as process and quality engineers to interpret the rules for calculating defects.
•Guidelines will have to be defined for certain defect conditions to assure the independence of component, placement, and termination defects
•Some components might have different defect rates than others. For example, mechanical, through-hole (TH), and surface mount technology (SMT) components can all be part of the assembly line process. Each will have a different defect rate, and they should not be lumped together in one defect number.
•DPMO concepts require knowledge of the actual number of termination opportunities, which are readily available in manufacturing but do not get finalized until late in the design and development process for electronic products (after PCB layout). Intermediate metrics such as the ratio of components versus termination opportunities might be more useful in the design stage, especially for de' sign for manufacturing (DFM) input，before the design in “hardened” after PCB layout. This intermediate metric was shown in Example 4.3.2.
•DPMO is an example of the attribute quality problem in six sigma. The notion of striving for “six sigma in everything that we do” is not directly shown with the use of one or two indices such as DPMO and OMI. Individual process quality as well as total assembly line quality should be examined. In DPMO, the emphasis is on a modified defect rate. In the next section, an alternate method for calculating and comparing quality of assembly lines using back calculated or “implied” Cpk is discussed with examples.