Study on the Validity of Electronic Parts Stress Models
- 1 August 1971
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Reliability
- Vol. R-20 (3) , 136-142
- https://doi.org/10.1109/tr.1971.5216115
Abstract
By using data on the failure rates of electronic components collected by a European Bank, we try to form an idea on the validity of the various models (Bazovsky model, MIL HDBK-217A standard model, exponential model, RADC TR-67 models) that represent the variations of the failure rate with respect to the thermal and electrical stresses. Two different approaches have been programmed on a computer: 1) conventional, using the linear regression analysis 2) more modern using the Fletcher-Powell method or methods derived from it (Fletcher-Reeves or Davidon). The results are given for four types of components. The models studied reflect, more or less accurately, what happens in real life. We define and give a value to a criterion, allowing the model to be chosen closest to reality. This paper, apart from its practical interest on the validity of currently accepted models and its theoretical interest on the use of methods for seeking extremes, shows how the computer can assist the reliability engineer with the filing of data, the statistical processing of these data, etc., right up to the automatic tracing of networks of curves.Keywords
This publication has 5 references indexed in Scilit:
- SF System: Transistors, Diodes and ComponentsBell System Technical Journal, 1970
- An efficient method for finding the minimum of a function of several variables without calculating derivativesThe Computer Journal, 1964
- Function minimization by conjugate gradientsThe Computer Journal, 1964
- A Rapidly Convergent Descent Method for MinimizationThe Computer Journal, 1963
- The Modified Gauss-Newton Method for the Fitting of Non-Linear Regression Functions by Least SquaresTechnometrics, 1961