Issue link: https://iconnect007.uberflip.com/i/597087
30 The PCB Magazine • November 2015 2. The yellow zone on either side of the green zone makes up the remaining 50% of the specification tolerance (25% for each yellow zone) 3. The red zone is everything outside of the specification limits; in other words, the zone where you are making bad parts. Control Charts The minimum number of samples needed to be able to perform any meaningful statisti- cal analysis is 30, which per the above sample size methodology would be 30 days of data col- lection and pre-control. That being said, it is important to note that the more data available to analyze, the more accurate a picture of the process you will get. However, this needs to be balanced with real world expectations to avoid Paralysis by Analysis. The length of data collec- tion will depend on the specific process and out of the box stability, but a 30-90-day duration is reasonable. After the data collection period, if the pre- control results appear to be able to stay within the green and yellow zones, the next step is to move the process into the control chart phase. Control charts are a graphical representation of the current state of a process, and should be implemented at the operator level to maximize effectiveness. A control chart's true function is to provide real-time feedback to control and improve a process, which means that the data displayed on the charts must help front-line op- erators make better process decisions. All control charts have three basic compo- nents: 1) a process center, or mean; 2) an up- per and lower control limit; and 3) an upper and lower specification limit. Data between the control limits and the specification limits signal that a process adjustment is needed. Data ex- ceeding the specification limits would be con- sidered out of acceptable limits and defective. The most common form of control charts is the X-Bar and R chart. X-Bar refers to the average of the data in each sample and plotted in the top half of the chart. "R" stands for the range of the data in each sample and is plotted in the bottom half of the chart. The range of the data sample is simply the difference between the highest and lowest value in the sample of five. normal Distribution In a stable process, data will be randomly centered on the process mean and contained within the control limits. This is called a nor- mal distribution, or bell-shaped curve, and is probably the most recognized and most wide- ly-used statistical distribution. The reason for this is that many physical, biological, and social parameters obey the normal distribution, and thus are said to behave normally. When data takes on a normal distribution, it simply means that most of the data will fall around the cen- ter, or mean, and that the remaining data will fall progressively farther from the mean with decreasing frequency. Cpk Analysis Every product has an optimum value, and because every process has variation, it also has a tolerance. This is defined as specification lim- its, with both an upper and lower spec limit (USL, LSL) surrounding the optimum value. To review, when a product or process is outside of either of these spec limits, bad product is be- ing produced. How well the process variation is centered on the mean and contained within these spec limits is called process capability. The relationship of this variation to the mean and spec limits is the process capability, or Cpk. The less variation in a process, and the closer the variation is to the mean, the higher the Cpk number. Again, with all the statistical tools available, the formula is not important for this purpose, but what is important is recognizing what this DATA AnALyTICS THroUGH STATISTICAL TeCHnIQUeS FeATure " The length of data collection will depend on the specific process and out of the box stability, but a 30-90-day duration is reasonable. "