Issue link: https://iconnect007.uberflip.com/i/1275152
AUGUST 2020 I SMT007 MAGAZINE 83 Figure 8: Example of bimodal anomalies. Figure 9: Mapping of component anomalies to potential root causes. There are also anomalies that show bi- modality (Figure 8). This may indicate a com- ponent source issue, perhaps from dual vendor sources. An action to verify if the correct com- ponent has been mounted or an incoming sam- ple measurement before component consump- tion could point the customer to the root cause. To summarize the cause and effect associ- ated with the component anomalies predicted using this method, refer to Figure 9. As these anomalies are in real-time and alerts are sent to the production teams, quick action can be taken to resolve performance and qual- ity issues that couldn't be predicted or detected until a much later time. Figure 10 shows the interface that maps the degradation anomaly to the actual test probe location for immedi- ate action. Calculating the ROI To provide tangible ROI and savings to the factory, we decided on a 12-month period for the calculation and metric tracking. The first important metric would be the first-pass yield (FPY), which is the quantity of boards that pass test the first time it is tested over the total num- ber of boards tested the first time. This is impor- tant as it indicates the robustness of the manu-