Based upon a uniform distribution inside an orthocell in the toleranced parameter space, it is shown how production yield and yield sensitivities can oe evaluated for arbitrary statistical distributions. Formulae for yield and yield sensitivities in the case of a uniform distribution of outcomes between the tolerance extremes are given. A general formula for the yield, which is applicable to any arbitrary statistical distribution, is presented. An illustrative example for verifying the formulae is given. Karafin's bandpass filter has been used for applying the yield formula for a number of different statistical distributions. Uniformly distributed parameters between tolerance extremes, uniformly distributed parameters with accurate components removed and normally distributed parameters were considered. Comparisons with Monte Carlo analysis were made to contrast efficiency.
[1]
John W. Bandler,et al.
Automated network design with optimal tolerances
,
1974
.
[2]
Ka-Ho Leung,et al.
Idealized statistical models for low-cost linear circuit yield analysis
,
1977
.
[3]
John W. Bandler.
Optimization of design tolerances using nonlinear programming
,
1974
.
[4]
W. Oehmisch,et al.
Coolidge, J. L., An introduction to mathematical probability. Dover Publications, Inc., New York 1962. XII + 214 S., $ 1,35
,
1965
.
[5]
T. Scott,et al.
Regionalization: A method for generating joint density estimates
,
1976
.