PRISM Forum - Message Replies

Posted by: George Pyryt (gpyryt@draper.com )
Organization:Draper Lab
Date posted: Fri Jun 9 15:58:04 US/Eastern 2000
Subject: Non-op electrical overstress factor
Message:
During the calculation of failure rates, Prism includes a serperate term for electrical overstress failure rate for all models. If I am trying to calculate the failure rate during storage over a long period of time, I do not think the electrical overstress should be that large of a factor. Is there a suggested factor or approach that I could use to reduce the part failure rate by when evaluating storage reliability. The failure rate Prism gives for a given resistor for example can be 2 orders of magnitude larger than if using the Mil-Std-217 data reduced by a factor of .01 for storage. The Prism failure rate I used included your previous suggestion of using 0 degree temperature rise for each component and 0% duty time.

David Dylis Subject: Non-op electrical overstress/ESD factor
Reply Posted by: (ddylis@alionscience.com )
Organization: Reliability Analysis Center (RAC)
Date Posted: Wed Jun 14 7:29:53 US/Eastern 2000
Message:

To calculate the failure rate during storage of a resistor, the analysis of each
component must incorporate a 0 degree temperature rise and 0% duty cycle.  Dormant failure
 rates were calculated for varying types of resistors using PRISM, RADC-TR-85-91 (Impact of
 Nonoperating Periods on Equipment Reliability) and MIL-HDBK-217F Notice 2 (Parts Count
 method) modified by an operating to dormant adjustment factor from the Reliability
 Toolkit: Commercial Practices Edition.  Differences in predicted failure rates between
 each method is highlighted below:
 
 All Failures are in Failures/10^6 hours
 
 Fixed, Carbon Composition (RC/RCR)
 PRISM Failure Rate 0.002073
 RADC-TR-85-91:  0.0001512
 MIL-HDBK-217F Notice 2 (Adjusted):  0.00132
 MIL-HDBK-217F Notice 2 (Not Adjusted):  0.0066
 
 Fixed, Film (RN)
 PRISM Failure Rate 0.001097
 RADC-TR-85-91:  0.00024
 MIL-HDBK-217F Notice 2 (Adjusted):  0.00222
 MIL-HDBK-217F Notice 2 (Not Adjusted):  0.0111
 
 Fixed, Network, Film (RZ)
 PRISM Failure Rate 0.002494
 RADC-TR-85-91:  0.001032
 MIL-HDBK-217F Notice 2 (Adjusted):  0.00096
 MIL-HDBK-217F Notice 2 (Not Adjusted):  0.0048
 
 Fixed, Wirewound. Power (RW)
 PRISM Failure Rate 0.001627
 RADC-TR-85-91:  0.001368
 MIL-HDBK-217F Notice 2 (Adjusted):  0.0039
 MIL-HDBK-217F Notice 2 (Not Adjusted):  0.0195
 
 Fixed, Thermistor (RTH)
 PRISM Failure Rate 0.006747
 RADC-TR-85-91:  0.00648
 MIL-HDBK-217F Notice 2 (Adjusted):  0.00084
 MIL-HDBK-217F Notice 2 (Not Adjusted):  0.0042
 
 Variable, Wirewound (RT)
 PRISM Failure Rate 0.012097
 RADC-TR-85-91:  0.002376
 MIL-HDBK-217F Notice 2 (Adjusted):  0.00144
 MIL-HDBK-217F Notice 2 (Not Adjusted):  0.0072
 
 Note:  The PRISM results in this analysis do not take into account the PRISM System Level
 Multipliers which could reduce (or increase) the PRISM calculated failure rate when
 component are integrated into an assembly/system based on system processes used.
 
 The following assumptions were made during these calculations:
 Environment:  Ground Benign (PRISM Defaults for Ground Stationary, Indoors)
 Duty Cycle: 0% in PRISM
 Temperature Rise:  0C in PRISM
 MIL-HDBK-217 Adjustment factor:  0.2 from active to passive ground benign  (Source: Reliability
 Toolkit: Commercial Practices Edition, Reliability Analysis Center (RAC), pg. 178).
 Note that factors vary between 0.03 and 1.0 based on environment of interest for resistors.
 
 As seen by this analysis predicted dormant failure rates vary amongst part types and
 method of prediction.  PRISM is based on the RAC’s most recent data and modeling
 development efforts while models developed in RADC-TR-85-91 and MIL-HDBK-217F Notice 2
 were developed in 1985 and 1995 respectively.  In fact, MIL-HDBK-217 was not developed
 to predict dormant failure rates and an adjustment factor was developed and published in
 the Reliability Toolkit as a means to develop a dormant failure rate estimate due to
 lack of tools to properly estimate dormant failure rates.
 
 PRISM has been developed to address the shortcomings of current reliability prediction
 techniques with one of these shortcomings being the ability to predict dormant
 failure rates.  Models in PRISM have been developed to allow the user to predict
 failure rates of a system based on its actual usage including dormancy.   The
 electrical overstress factor in the models addresses the probability of EOS/ESD
 causing component failure.  During the development of PRISM component models, it
 was assumed that the EOS failure rate is independent of the duty cycle. The reason
 for this assumption was that EOS is an externally applied stress and not a function of
 power being applied. Parts in storage may still be affected by this factor due
 to handling, etc