|M.Sc Student||Dekel Daniel|
|Subject||Reliability Model of Optical Surface Subjected to Laser|
|Department||Department of Quality Assurance and Reliability||Supervisor||Dr. Yefim Haim Michlin|
paper is about improvement of the reliability estimation model of optical
surfaces subjected to laser radiation (pulses).
currently in use is based on the MIL-HDBK-217 handbook used for assessing the
reliability of military equipment. Its laser reliability model, based on work
conducted in the1970’s predicts short MTBF’s for the element in question.
The major factors that influence the probability of damage are: kind of coating, flux intensity and number of pulses. The literature survey failed to find an up-to-date model. The literature approach of measuring the probability of damage in a single pulse as a function of the flux intensity revealed correlation with a log-normal distribution function (cumulative), but the range of probabilities range extends up to 10-3 , uncommon in ordinary practice. An approximation function was used in order to simplify the model, on the premise that extrapolation (ordinary practice) is acceptable in this case. It was assumed that a subsequent pulse is independent of the occurrence of its predecessor (“no memory”), permitting use of a binomial distribution. The analysis introduced an additional parameter, namely pulses per operating hour, which made it possible to convert the probability of single-pulse damage to the familiar form of failure rate.
The field reliability data of 350 specific designators were analyzed. The analysis yielded an estimate of the total operating hours and pulses. Failures of optical elements were identified. The least reliable element yielded an estimate of the probability of single-pulse damage in the range of 10-8 . A test under laboratory conditions yielded the damage probability per pulse in the same range.
The paper suggests an updated reliability model, incorporating of the stress parameter (ratio of flux intensity to damage threshold) and the number of pulses per operating hour.