Rapid Failure Mode Precipitation Testing (RFMPTTM

1 Achatz and Kleyner!RFMPT, October 11, 2013 Andre Kleyner, Global Reliability Engineering Leader Delphi Electronics & Safety Rapid Failure Mode Precipitation Testing ( rfmpttm ) ASTR 2013, Oct. 9-11, San Diego, CA Tom Achatz, PE Global Technical Integration Engineer General Motors Company Achatz and Kleyner!RFMPT, October 11, 2013 RFMPT in Design for Reliability (DfR) QFD, Requirements definitions, Benchmarking, Product usage analysis Understanding of customer requirements and specifications DFMEA, Cost trade-off analysis, Lessons Learned Probabilistic design, Tolerance Analysis FEA, Warranty Data Analysis, DRBFM, Reliability prediction Lessons Learned, Reliability Block Diagrams HASS, Control Charts, Re-validation, Audits, Look Across, Lessons Learned, ORT HALT/RFMPT, Evaluation Testing , DRBTR, Reliability Growth modeling, Change Point Analysis Design and Process Validation Accelerated Test, RFMPT Reliability Demonstration.

2 1. Identify 2. Design 3. Analyze 4. Verify 5. Validate 6. Monitor and Control Achatz and Kleyner!RFMPT, October 11, 2013 Difficulties to correlate HALT failures with field failures, therefore HALT is not a life test. Inherent lack of control over vibration profiles -- the effect of which is to impart loads bearing little or no resemblance to field conditions. Use of HALT is further limited when ancillary environmental inputs such as humidity, spray, shock, high voltage, electrical loads, etc. must be included. For some power electronics products the existing HALT chambers are simply too small. ASTR 2013 Oct 9-11, San Diego, CA RFMPT Addresses Some of HALT s Limitations Achatz and Kleyner!

3 RFMPT, October 11, 2013 Phase Title Environment Condition Time/Cycles Comments 1 Functional Verification -40/25/105 C Lower, nominal, and upper design temperatures with nominal set points for all non-temperature inputs (voltage/current, etc.). N/A Verify initial functional status and establish baseline for every DUT. 2 High Temperature Pre-aging (Optional) 105oC Temperature set to upper design temperature. Units on test are operational with nominal electrical duty cycle; full I/O fault monitoring. 300 hours Accelerate microstructural changes such as tin whiskers, delamination, outgassing, etc. that otherwise occur very gradually in service.

4 3 Combined Environment Vibration; thermal cycling; electrical load cycling; humidity; corrosive spray Simultaneous 3-axis shaker system (see Figure 2) with field-correlated vibration profiles (See Table 3); 85% relative humidity; 5% NaCl spray Step stress vibration (see Figure 2 and Table 3); humidity controlled above 50oC; corrosive spray applied every 15 minutes Vibration profiles derived from field measurements on similar product and correlated to reflect one life over 15 hours of Testing . After first life, increase vibration levels by dB per step every 6 hours. Continue to test all parts to Failure , repairing where possible.

5 4 Functional Verification -40/25/105oC Nominal set points for all non-temperature inputs (voltage/current, etc.) N/A Verify final functional status; quantify performance degradation; identify failures Suggested RFMPT Phases Use functional requirements for continuous test monitoring Quantify test to field correlation Maintain careful records of time/cycles to Failure Make reliability predictions, identify/fix failures, track reliability growth Achatz and Kleyner!RFMPT, October 11, 2013 Example RFMPT Test Flow Identify potential design weaknesses quickly Multiple concurrent environmental stresses Correlate to field (customer to test relationship) Root cause failures Track times to Failure Continuous monitoring essential Track reliability growth (A/B comparisons) Replace non-repairable parts Use results to shorten PV Testing and improve timing between design turns RFMPT is reliability development not validation Achatz and Kleyner!

6 RFMPT, October 11, 2013 Example RFMPT Road Load Data Acquisition customer survey data linked through damage models to correlate RLDA and RFMPT Tri-axial Accelerometers at Component Mounting Points (Inputs) Internal Tri-axial Accelerometers to Understand Responses Measure other variables such as temperature, input/output voltage and current, output torque, coolant flow, road grade, etc. Achatz and Kleyner!RFMPT, October 11, 2013 Failure MODES: delamination, insulation breakdown, loss of isolation, capacitor wear out, circuit interruptions (opens/shorts), increased leakage current Failure MECHANISMS: creep, diffusion, out-gassing, electro-migration, tin whiskers, dendritic growth MODEL: Arrhenius INPUT: Activation energy, time at temperature histogram, solar load, NAAWS, test temperature OUTPUT: Test hours at prescribed temperature histogram of severe customer usage over target design life RFMPT Pre-aging Correlation Process Achatz and Kleyner!

7 RFMPT, October 11, 2013 January 9, 2014 Example RFMPT Pre-aging Results Achatz and Kleyner!RFMPT, October 11, 2013 RLDA CUSTOMER TARGET TEST PARAMETERS DERIVE PROFILE Run Test RFMPT Vibration Correlation Process Failure MODES: circuit interruptions (opens/shorts), broken wire bonds, material separation, loss of isolation, component separation MODEL: Lalanne/Dirlick fatigue damage spectrum using Halfpenny mission profiling technique INPUT: RLDA time histories, S-N slope/intercept, target life OUTPUT: Equivalent damage triaxial PSD profiles for prescribed accelerated test time Achatz and Kleyner!RFMPT, October 11, 2013 Example RFMPT Vibration Profiles Each axis controlled at field correlated levels.

8 Achatz and Kleyner!RFMPT, October 11, 2013 Failure MODES: circuit interruptions (opens/shorts), broken wire bonds, material separation, loss of isolation, capacitor wear out Failure MECHANISMS: mismatched CTE s, fatigue MODEL: Coffin Manson (inverse power law) INPUT: Coffin-Manson parameter, target T histogram (RLDA/PUMA), test T OUTPUT: Required thermal cycles for prescribed T RFMPT Temperature Cycling Correlation Process Service Life mServiceTestmaxminTNTT = Achatz and Kleyner!RFMPT, October 11, 2013 Number of total equivalent accelerated cycles Time (Life) occurrences per X miles =TLifedAc celerate T Time Time T T Cold Climate RFMPT Temperature Cycling Correlation Process Achatz and Kleyner!

9 RFMPT, October 11, 2013 Example RFMPT Temperature Profile Achatz and Kleyner!RFMPT, October 11, 2013 Time (minutes)Motor 1 Motor 2 Motor 3 Coolant Temperature ( C)Coolant Flow Rate (l/min)1501505-20 to 75c10250505-20 to 75c10350505-20 to 75c1041501507-20 to 75c1051501507-20 to 75c1061501507-20 to 75c1071503007-20 to 75c1081503007-20 to 75c1091503007-20 to 75c10104005020-20 to 75c1011504005-20 to 75c101250505-20 to 75c1013150507-20 to 75c1014150507-20 to 75c10151501507-20 to 75c101630015010-20 to 75c101730015010-20 to 75c101830015010-20 to 75c1019504005-20 to 75c102040015020-20 to 75c10 Peak Current (A)Example RFMPT Electrical Profile Achatz and Kleyner!

10 RFMPT, October 11, 2013 Example RFMPT Setup Automotive hybrid power inverter Simultaneous 3-axis vibration with hydraulic actuator Temperature cycling (chamber not shown) Duty cycle applied with active loads and power supply Achatz and Kleyner!RFMPT, October 11, 2013 Comparison of HALT and RFMPT Feature HALT RFMPT Special equipment Y N Repetitive shock machine Y N Rapid thermal transitions Y Depends Suitable for large and heavy devices N Y Representative fixturing N Y Time to run Up to a week Up to a month Fully correlatable to field environment N Y Well defined environmental stressors N Y Stressors other than vibration/temperature special standard Ability to tailor vibration profile N Y Can be used to identify operational limits Y Y Can be used to identify destruct limits Y N Suited for larger sample sizes N Y Useful for correlating analytical models N Y Suitable for