Issue link: https://iconnect007.uberflip.com/i/474043
78 The PCB Magazine • March 2015 variation to the mean and spec limits is the pro- cess capability, or Cpk. The less variation in a process, and the closer the variation is to the mean, the higher the Cpk number. With all the statistical tools available, the formula is not im- portant for this purpose, but what is important is recognizing what this number means. It is generally accepted that a Cpk of less than 1.33 would indicate a process that is not capable of consistently meeting customer requirements, and a Cpk of 2.0 would represent a six sigma level. The sigma level represents how many standard deviations, or sigmas, it takes to reach the spec limits on either side of the mean. In other words, in a three sigma process it takes three sigmas to reach the LSL and three sigmas to reach the USL. Sigma Levels The sigma level is a difficult concept to un- derstand during the early stages of process im- provement, so I will try to simplify this as much as possible. When a process is referred to in sig- ma terms, it is stating how many sigmas (stan- dard deviations) it takes to reach the specifica- tion limits from the mean. Statistical rules state that the amount of variation that falls within each group, or sigma level, is repeatable and can be quantified. It is important to remember that these rules are constant regardless of what sigma level a process is operating under. Most organizations have not achieved a 99% yield, much less a three-sigma level. As Figure 1 shows, in a normal distribution, 99.7% of the variation will fall within +/– three standard de- viations, or sigma levels. While that may appear to be a very good yield on the surface, this trans- lates into 2700 DPPM that will fall outside of the specification limits. The areas outside of the spec limits are called the process tails, and again referring to Figure 1, these tails fall outside the spec limits and represent defective product. As we saw earlier in this chapter, a three sigma pro- cess results in an awful lot of defective product. Now let's look at a six sigma process, where it takes six standard deviations, or sigmas, to reach each spec limit. Again, statistical rules state that 99.9997% of the variation will fall figure 1: a normal distribution, where 99.7% of variation falls within +/– three standard deviations, or sigma levels. BEST PRACTICES 101, PART 5: PROCESS CAPABILITy continues Point oF View