Issue link: https://iconnect007.uberflip.com/i/976095
46 SMT007 MAGAZINE I MAY 2018 reviewed to determine how well the equation fits the raw data. The R-Sq term shown in the fitted line plot represents how much of the variation the predictor (placements per panel) is accounted for by the CPH equation. The closer to 100%, the better. This model's R-Sq (adj) term is 69%. Since the model is intended to be used as a barometer of how well a product is running on a given SMT machine, this R-Sq term is acceptable. The difference between the data points and fitted line are called residuals. Residuals represent the error or amount of variation not explained by the regression equation. There are four items to check to confirm that there are no unusual data points in the model. These are: 1. Normal Plot: Residuals form a straight line. 2. Histogram: Residuals appear to form a normal curve. 3. Versus Fits: Residuals are contained in a straight band, with no obvious pattern in the graph. 4. Versus Order: Residuals appears to be in statistical control. This process of reviewing the "placements per panel" versus CPH, identifying the "Max CPH" and modeling the ramp to "Max CPH" with a regression equation (linear, quadratic or cubic) was completed for all other SMT machine models Rockwell Automation uses. Figure 3: Fitted line plot of SMT machine CPH predicted by placements per panel for less than 300 placements per panel. Figure 4: SMT machine CPH residual analysis.