Review the Two Different Methods Devised by the Reliability Analysis Center (RAC)

Although the dormant or non-operating failure rate of a system is often assumed to be 0, this is generally not the case. To take dormant states into account in reliability predictions, the Reliability Analysis Center (RAC) has devised two different techniques:

• Dormant failure rates can be estimated through the use of dormant environments and conversion factors as described in "Topic A16: Dormant Analysis" of the Rome Laboratory Reliability Engineer's Toolkit (Commercial Practices Edition).
  
• Dormant failure rates can be calculated by PRISM RACRates models, which recognize and account for the fundamentally different failure mechanisms involved in dormant storage as described in the RAC PRISM® User's Manual.

This document summarizes how the RAC conversion factor method and PRISM RACRates models take dormant states into account when calculating failure rates. For detailed information about these methods, please refer to the above-referenced RAC publications.

Conversion Factor Method

Estimations of dormant failure rates are determined based on operating failure rates and taking into account the variation between the operating and dormant environments. To determine the dormant failure rate, the active failure rate of each part is first computed. Then, the active failure rate is multiplied by the appropriate value from a table of active-to-passive conversion factors. This conversion also takes into account the part type.

For example, when an integrated circuit is in a ground environment when active and also when dormant, the failure rate of this part in the dormant environment is estimated to be 0.08 times its active failure rate. If the same part is placed in an active environment of space and a dormant environment of ground, it has an estimated dormant failure rate of 0.30 times its active failure rate.

Once both the active and dormant failure rates for a part are computed, the duty cycle is used to compute the total failure rate of that part:

        λ_part = λ_active * Duty Cycle + λ_passive * (100 − Duty Cycle)

Assembly failure rates are then computed by adding up all of the individual part failure rates. Finally, the overall system failure rate is computed by rolling up all of the individual assembly failure rates.

RACRates Model Method

PRISM RACRates models recognize and account for the fact that the failure mechanisms involved in the dormant state are fundamentally different from those in normal operation. This method is vastly different from the conversion factor method, which assumes that the failure mechanisms involved in the dormant state are the same as those in normal operation but occur at a lesser rate.

To provide accurate failure rate predictions that take into account both operating and non-operating conditions, RACRates models consider five separate contributions to the total part failure rate:

• Operating base failure rate (λ_OBπ_Gπ_DCOπ_TO)
• Non-operating or "environmental" base failure rate (λ_EBπ_Gπ_DCNπ_RHT)
• Temperature cycling base failure rate (λ_TCBπ_Gπ_CRπ_DT)
• Failure rate due to electrical overstress (λ_EOSπ_G)
• Failure rate due to solder joints (λ_SJ)

RACRates models for computing failure rates are available for the following part types: capacitors, diodes, integrated circuits, resistors, transistors, and thyristors. While there is also a RACRates model for predicting the failure rate of software, it is structured very differently from those for predicting part failure rates and is therefore not addressed in this article.

The RACRates model for predicting the failure rate of an integrated circuit follows:

λ_P = (λ_OBπ_Gπ_DCOπ_TO) + (λ_EBπ_Gπ_DCNπ_RHT) + (λ_TCBπ_Gπ_CRπ_DT) + (λ_EOSπ_G) + (λ_SJ)

Where:

		λP	=	Predicted failure rate (in failures per million calendar hours)
		λOB	=	Operating base failure rate
		πG	=	Reliability growth factor
		πDCO	=	Operating duty cycle factor
		πTO	=	Operating temperature acceleration factor
		λEB	=	Non-operating or "environmental" base failure rate
		πDCN	=	Non-operating duty cycle factor (proportional to time in the non-operating state)
		πRHT	=	Non-operating temperature and relative humidity acceleration factor
		λTCB	=	Temperature cycling base failure rate
		πCR	=	Temperature cycling rate acceleration factor
		πDT	=	Temperature cycling ΔT acceleration factor
		λSJ	=	Failure rate due to solder joints
		λEOS	=	Failure rate due to electrical overstress

In the RACRates model, the operating base failure rate contribution (λ_OBπ_Gπ_DCOπ_TO), temperature cycling base failure rate contribution (λ_TCBπ_Gπ_CRπ_DT), failure rate due to electrical overstress (λ_EOSπ_G) contribution, and failure rate due to solder joints (λ_SJ) contribution all affect the calculation of the operating failure rate for the part. The non-operating base failure rate (λ_EBπ_Gπ_DCNπ_RHT) contribution affects the calculation of the non-operating failure rate for the part.

Once both the operating and non-operating portions of the RACRates model are computed, they are added together to compute the total failure rate for that part:

        Total Failure Rate for Integrated Circuit = [Operating Portions] + [Non-Operating Portions]

        λ_P = [(λ_OBπ_Gπ_DCOπ_TO) + (λ_TCBπ_Gπ_CRπ_DT) + (λ_EOSπ_G) + (λ_SJ)]+ [(λ_EBπ_Gπ_DCNπ_RHT)]

As with other modeling methods, the assembly failure rates are then computed by adding up all of the individual part failure rates, and the overall system failure rate is computed by rolling up all of the individual assembly failure rates.

The Relex Reliability Prediction Module supports both of these RAC methods for taking dormant states into account in reliability predictions. If you would like additional information about how these methods are incorporated into Relex, please email info@relex.com.