TABLE OF CONTENTS Once the historical/collected information becomes available, the next step is to calculate the Sigma metrics to identify the quality level of the current product/process. This section will discuss how to calculate the Sigma metrics value.
The Defects Are Randomly Distributed
If a sample were taken, say that 2.0 defects per unit were found. Now, if another sample were collected in the near future without changing any process, input material, and equipment, exactly same defect rate should not be expected. If the defect rate is different, this result does not necessarily mean that the process has become worse or better.
Based on the given factors (input material, process, and equipment), think about the likelihood of producing a unit (product/service) with no (zero) defects. This would be true only when there is no rework or repair. Therefore, yield and defect counts are related measures, and defect metrics are calculated separately depending on the database. There are two types of process databases:
Discrete process databases
Continuous process databases
Discrete Process Database Metrics
Metrics calculations can be divided into two steps:
Calculate the defect rate (the defect rate could be defects per hundred/thousand/ten thousand/hundred thousand/million).
Find the Sigma value in Table 3.18.
| Long-Term Yield (%) | Process Sigma (ST) | Defects per 1,000,000 | Defects per100,000 | Defects per10,000 | Defects per 1,000 | Defects per100 |
|---|---|---|---|---|---|---|
| 99.99966 | 6.0 | 3.4 | 0.34 | 0.034 | 0.0034 | 0.00034 |
| 99.9995 | 5.9 | 5 | 1 | 0.05 | 0.005 | 0.0005 |
| 99.9992 | 5.8 | 8 | 1 | 0.08 | 0.008 | 0.0008 |
| 99.9985 | 5.7 | 15 |
