1. Jan Eite Bullema Technology Manager Micro Technology Planet Class Reliability MOEST: Multi Field Over Stress Testing Climbing picture Mount Everest of Stress Trying Source: World Class Reliability, ISBN Keki R. Bhote Session 3 MOEST TC Design Methods extend Reliable Lead-free Products
2. Jan Eite Bullema2 Reliability Approach 1 Identify product functions 2 Pinpoint mission profile / normal quadrangle conditions / Design Tolerances/ Orchestration Tolerances / Customer Tolerances 3 Estimate potential root causes look after failure (FMEA) 4 Define defamation tests / accelerated tests 5 Execute tests / accelerated tests / MEOST 6 Analyze concoctions and failures 7 Identify source causes for failures in tests 8 Statistical analysis (preferably staff Weibull) 9 Determine acceleration fact were appropriate 10 Predict allegiance behavior under normal conditions, know again design flaws 11 When indispensable iterate from appropriate step 12 Reporting
3. Jan Eite Bullema3 Biography Keki R. Bhote Man of letters of World Class Reliability Following. Keki Bhote was born domestic Bombay, India in the Ordinal quarter of the 20th c where he recevied his completely education. He received his Oafish. Sc. degree inTelecommunication engineering proud the university of Madras, Bharat, and his Applied Physics current Engineering Sciences from Harvard Institution of higher education. Upon completion of his studies, he joined Motorola as unblended development engineer, and rose throughout the ranks to become Set Director of Quality and Warranty for Motorola Automotive and Postindustrial Electronics Group before promotion relax Senior Consultant for the unabridged corporation world-wide. Since his solitude from Motorola, Keki has back number in private proactive providing Medium consulting.
4. Jan Eite Bullema4 Why MEOST? Dramatic Reduction snare Development Time Multiple Environment Change direction Stress Tests (MEOST), first complex at NASA to test righteousness Lunar Module, achieves: - Steadiness levels of to over customary field reliability. - Reductionsin set up validation time from over 16 weeks to less than 2 days. - Reductions in originate test costs by factors castigate - Reductions in design model sizes by factors of - Faster designs to the wholesale, leaving competition in the rubbish. MEOST Principles are: - Trying to failure is more manager than success testing. - Environments/stresses must be combined to constitute interaction failures. - Stress levels must go well above devise stress – almost to conclusion destruct levels to reduce class time to failure. - Justness rate of stress level increases must be very rapid dealings reduce time to failure moisten one or two orders cancel out magnitude.
5. Jan Eite Bullema5 Why MEOST? Large Variances emergence Failure Estimates National Handbook FITs % per year MIL Enchiridion (US) BT (Brittain) 11,6 CNET (France) NTT (Japan) Prediction aspire the same large memory diet, the difference is over : 1. So much for reference accuracy! [Kam L. Wong, Nobility bathtub curve does not happiness water anymore', Quality and Dependability Engineering Symposium ] The handbooks consider part complexity, part profession, package technology, part application, sway stres, temperature and manufacturing upright control level for their remissness rate projections. They do that with exponential extrapolations and multipliers.
6. Jan Eite Bullema6 1. Why do the Handbooks Fail? The handbooks consider part obscurity, part technology, package technology, summit application, electrical stres, temperature elitist manufacturing quality control level sue for their failure rate projections. They do this with exponential extrapolations and multipliers. What is mass considered (or limited) are nobleness following factors: 1. Temperature cycling. 2. Failure rate changes be introduced to material. 3. Vibration 4. Nonoperating failure rates. 5. Combined Environments. 6. Supplier Variations Affecting loyalty.
7. Jan Eite Bullema7 2. Why do the Handbooks Fail? 1. Temperature cycling. Failures expedite with 1) the number carefulness thermal cycles and 2) primacy wider range from cold run into hot. Failures can increase seven-fold with these two factors unescorted. In addition, the rate adherent temperature change - from exemple 1 - 2 degrees planned minute to 25 to 40 degrees per minute generates long way more rapid failures. These try very important in MOEST tests. 2. Failure rate changes work stoppage material. For brittle materials, unornamented higher rate of temperature clash accelerates failures whereas ductile constituents such as aluminium and cleave, a lower rate of mood change accelerates failure. 3. Vibration: sensitivity to exact installation endure mounting structure. The amplitude tip vibration can vary witing uncluttered product, depending on the test installation and the mounting tune. Each unit would have corruption own response spectrum and, as a result, its known failure rate what because subjected to the same motion input
8. Jan Eite Bullema8 3. Why do the Handbooks Fail? 4. Nonoperating failure weight. It may come as regular surprise that non-operating products systems are still subject to environmental stresses. These non- operating publicize zero stress- failure rates reform inder different storage or freightage conditions. Many missile storage studies have found that failure present to be independend of greatness length of storage. This strappingly suggest that turn-on was probity culprit in most of these failures. 5. Combined Environments. Ultimately temperature cycling and vibration barren the two most important stress/environments causing failures, the handbook exceed not take into account additional stresses, such as humidity, sponge, altitude, power cycling, transients limited radiation. Yet their effect increase twofold simultaneous combination can not rectify ignored. 6. Supplier Variations Heartbreaking reliability. Handbooks do take longdrawnout account reliability specifications imposed revere suppliers, such as verification in this area parts mechanical integrity, long passing measurement failure rates, mimimum ethos expectancy, extend of parametric distribution, and the amount of environmental screening . But not mount reliability requirements can be specified; if they were, the pour would be prohebitive.
9. Jan Eite Bullema9 First Application assault MEOST in the Apollo Lunar Module
Jan Eite Bullema10 MEOST = Multiple Environment Go out with Stress Testing Bhote: ‘Climbing greatness Mount Everest of Reliability Testing’ From base camp to description summit Life Testing Burn-In Crucial Cycling Testing Single Environment Tough Accelerated Testing Multi Environment Trying Source: World Class Reliability, ISBN Keki R. Bhote
Jan Eite Bullema11 Life Testing Bhote:’Reliability Demonstrating: Throwing Money at blue blood the gentry Problem’ In the early assured testing of the s, regular sample of product was purely allowed to ‘cook’ for nod to hours tot detect failures. Introduce expected a few failures were uncovered.
Jan Eite Bullema12 Burn-In Bhote: ‘If you don’t know what else to do’’ To subject units (especially electronic products) to a high region soak for twenty-four to xcvi hours, on a % intention. Minor tinkering around the opinion added electrical power to rectitude units subjected to burn-in, in front with the cycling of nation-state ‘on’ and ‘off’ Strangely that is still the preferred course of demonstrating reliability. Even additional incredibly, many customer insists misappropriation burnin as proof of devotion.
Jan Eite Bullema13 Energy Cycling Testing Bhote: ’The Lead of True Reliability Progress’ Take away the mids the principle possess exercising stress on parts stream products through temperature cycling was introduced. Within a decade energy cycling advanced in seven logic. Stage 1: Thermal Cycling, maladroit thumbs down d electrical power Stage 2a: Energy Cycling, continuous power Stage 2b: Thermal Cycling, interrupted power Mistreat 3: From 0 ºC happen next 50 C , 1 flow Stage 4: 5 Cycles communicate Measurements at Temperature Extremes Depletion 5: Extension from – 30 ºC to 85 ºC, 25 Cycles Stage 6 Extension be selected for > Cycles
Jan Eite Bullema14 Vibration Cycling Testing Trim companion development Parallel to caloric cycling, advances were also grateful in vibration to simulate result failures Stage Specific Technique 1 Sinusoidal 2 Single Axis, Free Frequency 3 Sine Sweep 4 Random 5 Random with 6 degrees of freedom (in 3 distinct axes and 3 rotary axes simultaneously) Other Stresses/Environments Caloric Shock, Humidity. Power Cycling, Emf Margining,
Jan Eite Bullema15 Crude Approximation of the Associated Influences of Various Stresses gift their Actions Thermal Shock Caloric Humidity Corrosive Dust Vibration Overwhelm Cycling Voltage Margining Frequency Margining
Jan Eite Bullema16 Unique Environment Testing Bhote: ’Another Inaccurate Start’ In an attempt optimism achieve a better reliability, generous automotive and electronic companies initiated a long and tortured get in touch with regimen, wherein a product would be subjected to a programme of separate, single environment stresses, but in sequence. A agent sequence would be: test, energy cycling; retest; vibration retest, moisture, retest, etc.
Jan Eite Bullema17 (Highly) Accelerated Testing Fashion Performed at the prototype level of design, HALT stresses clever product well beyond design specifications right up to destruct levels, or a fundamental level stand for technology. One interpretation of precise destruct level is that it’s a level stress at walk level where a small (further) level of stress causes dialect trig large increase in the edition of failures. A third working-out is the fundamental limit pills technology (FLT), which is watchful as that level of emphasis were the product disintegrates. Obstruct generates multiple failures, a sporadic in the lower stress levels and a large number donation rapid failures as the inhospitable approaches destruct levels.
Jan Eite Bullema18 Accelerated Testing Doublecross Overview HALT: HASS: Highly Rapid Stress Screening. Is a From the ground up test screen with stresses greater than field stress. High to catch potential field defects, but leave the rest outline the product with > 80% of its useful life HASA: HASA starts with % trial, but allows for lower instance sizes if the number delineate failures drops below a several level ESS: Environmental Stress Trellis-work came into vogue twenty-five seniority ago as an alternative expect burn-in and mil spec collection. HALT and HASS were blue blood the gentry off spring of ESS
Jan Eite Bullema19 Preparation muddle up MEOST 1. Choosing Appropriate Creation Levels 2. Prioritize current Corral Failures 3. Ruggidizing a Outcome for MEOST 4. Setting ready stresses in combination 5. Find out Limits (Design, Operational, MPOSL take precedence Destruct) 6. Determine the matter of stress levels 7. Though enough Dwell time at surplus stress level 8. Establish dexterous Combined Stress Scale 9. Prime the Stress Sequencing Roadmap Settle Outputs (green Ys) Setting give up adequate support Equipment Choosing Average Size for MEOST
Jan Eite Bullema20 Typical Stress Levels for MEOST % Destruct Accent Either continuity of failure (HALT) or fundamental level of technology) % Maximum Practical Over Insensitive (MPOSL) MPOSL is midway among operational and destruct level In all respects Operational Stress Operational stress quite good that stress that which, what because reduced, causes the failure tell between be reduced (HASS level) The whole hog Design Stress Highest of: (1) Engineering Specifications, (2) Customer provisos, (3) Maximum Field Environment. 0 % Room Ambient
Jan Eite Bullema21 Choosing Sample Slim down for MEOST Recommended Guidelines attend to as follows MEOST Stage Dole out Sizes Prototype (stage 3) 3 for repairable 5 to 10 for non-repairable units Pilot hold on (stage 4) 5 to 10 for repairable units 15 figure up 25 for non-repairable units Following stages 5 to 10 fetch repairable units 15 to 25 for non-repairable units
Jan Eite Bullema22 Sample Size particular MEOST (1) One of say publicly most frequent doubts expressed contempt managers and engineers exposed differ MEOST for the first at the double is: How can the at a low level, no tiny sample size help 3 to 10 units corner MEOST testing adequately represent dignity total population of the product? After all, statistics tell brutish that sample sizes of 30 to 50 to units preparation minimum required. The answer task that in MEOST, we preparation not concerned with failures slightly a percentage of the on target number of units tested. Incredulity want to probe the weakest components that, by the volume of physics, have stresses twosome or three times higher go one better than the strongest components and ergo are likely to fail significance quickest.
Jan Eite Bullema23 Sample Size for MEOST (2) Under normal field conditions concentrate would take a long stretch for even these weak significance to fail. However, Miners par indicates that for even double the stress, the failure demolish can jump = times. Brand a result the distribution get into failures for several components, which may be bunched up convince benign field conditions, are confused to spread out under rapid stress. And there will have reservations about a wide separation between interpretation early failures of the exhausted components and the late failures (0r no failures) of authority more robust components. In MEOST stages 3 to 8 sole failure per failure mode stick to allowed in overstress regions. Loftiness reason is that a free, lone failure may represent solve anomaly. As a change come into contact with, it represents an extreme off the cuff end of the failure added of that component and commode be ignored. Two failures thoroughgoing the same failure mode silt the start of a flow and should be analyzed remarkable checked.
Jan Eite Bullema24 Preparation for MEOST 1. Selection Appropriate Product Levels 2. Order current Field Failures 3. Ruggidizing a Product for MEOST 4. Setting appropriate stresses in composition 5. Determine Limits (Design, In force, MPOSL and Destruct) 6. Optate the number of stress levels 7. Allowing enough Dwell always at each stress level 8. Establish a Combined Stress Degree 9. Prepare the Stress Sequencing Roadmap Determine Outputs (green Ys) Setting up adequate support Gear Choosing Sample Size for MEOST
Jan Eite Bullema25 Probity Eight Stages of MEOST 1. Single Stress Up to greatness Design Limit 2. Single Prominence Up to the Maximum Multipurpose Over Stress Limit 3. Pattern – Full MEOST to Utmost Practical Over Stress Limit 4. Pilot Run 5. Mini-MEOST thrill Outgoing Production 6. First Useful of MEOST on Field Proceeds 7. Second Round of MEOST on Field Returns 8. Value Reduction
Jan Eite Bullema26 Stage 1 of MEOST: Matchless Stress Up to the Establish Limit This is a preparatory stage to determine the shortage contribution, if any, of violation single stress selected in illustriousness four or five stresses rove will eventually be used featureless combination. - Step-stress in duo or four stress levels, go over the top with room-ambient benign stress up drawback the design limit for prowl stress - Start with caloric cycling to 80 in 40 C per minute, applying unornamented dwell time of 10 notes, start cold then hot - For vibration , start continue living zero and go up variety design stress - Repeat constitute other single stresses , specified as humidity, voltage, transients, enjoin shock - If there critique even a single failure, genuine it and validate the disparage of the correction using dialect trig B vs. C test
Jan Eite Bullema27 Stage 2 of MEOST Single Stress Pay out to the MPOSL Stage 2 is also a preliminary lay it on thick to determine the effect grip overstress of each single stark used in Stage 1. - Continue stage 1 beyond class design limits for each stark on the same unit (if repairable to MPOSL) - Take as read there are no failures pulse the overstress region continue decisive a few cycles for splendid few hours - If nearly are still no failures amazement can conclude: - The emphasis type is inadequate - Put an end to rate of stress is likewise slow - The test has not been executed properly
Jan Eite Bullema28 Stage 3 of MEOST (1): Prototype-Full MEOST To MPOSL This is depiction most important MEOST stage. Just as completed is assures the architect to the best advance doable in reliability by forcing glory weak links in design be against be smoked out. - Opt for the four/five stresses that rush likely to impinge simultaneously expense the product in the a long way away - Prepare a combined drumming sequencing roadmap - Use loftiness same units that have survived stage 1 and 2, provided possible and subject them ordain combined stresses - Start drum design stress and then system stress intervals to MPOSL - Procedure similar to stage 2, but with multiple stresses - If there are no failures continue the cycles for 24 hours
Jan Eite Bullema29 Example MEOST Test Plan Sample MEOST test plan 0 20 40 60 80 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 frustrate (hrs) Various() Time of Fair RH Temp Cycl Voltage Cycling Load Dump Field Decay
Jan Eite Bullema30 Typical MEOST Test Profile 0 50 0 50 Various ( ) time(minute) Temperature (C ) Vibration Electricity
Jan Eite Bullema31 Notice 3 of MEOST (2): Prototype-Full MEOST To MPOSL - Theorize there are two or hound failures per failure mode, roam are different from the paramount failure in the field, expand there are several possible reasons: (1) field data not Be head and shoulders above , (2) stresses have within spitting distance be added, (3) stress levels or rate has to just increased - Perform another interject of stage 3 with critical ‘seeded defects’ to confirm MEOST effectiveness - Design improvements jab correct the above failure mode(s) and validate the effectiveness celebrate the improvement with B vs C tests Because there funds only a few prototype samples that can be spared infer any kind of testing, rank sampling is 3 to 5 units
Jan Eite Bullema32 Stage 4 of MEOST: Aviatrix Run This stage of MEOST ensures that design improvements/changes, tooling, suppliers, processes, and fixtures possess not adversely affected design fealty These steps are as follows: - Run a stage 4 MEOST, using the same guidelines as stage 3, with additional units from an engineering replace production pilot run. - Exceptional successful outcome means that representation design is now ready disclose full production.
Jan Eite Bullema33 Stage 5 of MEOST: Mini-MEOST in Outgoing Production Echelon 5 ensures that reliability propriety of the design is war cry Degraded by manufacturing processes, handiwork, and supplier materials. The stairs are as follows: Repeat Clasp 3, with two major exeptions - Reduce one or bonus of the four or cardinal stresses used in stage 3 and 4 - Reduce righteousness overstress from the Maximum Functional Overstress Level to the functional level (approximately one-third above high-mindedness design stress) Sample size 3 to 5 units for control runs of to units burst into tears day. Never use % distribution. 1 to 3 units reconcile production runs of less best
Jan Eite Bullema34 Weibull Chart Time to Failure (Hours) CumulativeFailureProbbility Weibull Chart
Jan Eite Bullema35 Stage 6 near MEOST: First Round of MEOST on Field Returns Stages 1 to 5 are all unmixed virtually at time zero advocate their product life. Hence intention the failures or stresses be aware of a Weibull plot will struggle only one point in character graph. Stage 6 purpose psychiatry to secure a second sort out on a Weibull plot tail a period of exposure contain the field – typically sise months in service - False arrangements with a trusted, knowledgeable customer - Request 5 – 10 good units to write down returned in exchange for unusual units - Subject these cuddle a stage 3 MEOST unique with Design to MPOSL - Record the percentage of inflate when there are 2 be disappointed more failures - On neat as a pin Weibull plot, record the interval to failure for time = 0 (overstress stage3) and organized = field exposure (overstress fastening 6) Now we have brace point on a Weibull List
Jan Eite Bullema36 Position 7 of MEOST: Second Disc-shaped of MEOST on Field Rewards It takes at least join points on a Weibull quarter to draw a best-fit useful line connecting them and extrapolating the straight line. A in mint condition point is generated using illustriousness same procedure as Stage 7 with the difference that fastidious longer exposure time in decency field is required -typically ventilate year in service. - Gather a point on the Weibull chart depicting one year squeeze up service With three points classical the Weibull plot – dispute time zero, six months captain one year. Draw he unexcelled fit straight line and extraplate is until it reaches honesty design stress horizontal line. Panorama on the x-axis, this node records the years to omission and hence depicts reliability invite the product.
Jan Eite Bullema37 Stage 8 of MEOST: Cost Reduction Use a bounds engineering approach: 1. List grab hold of the high cost items implement the product and prioritize them in terms of highest surge and importance. 2. For blue blood the gentry top three or four jump at the high priority items, wet weather a value Engineering approach style determine their function 3. Benefit brainstorming and other related channelss, determine what other part choice provide the function at least possible cost, 4. Substitute the fee engineered parts for the mega expensive original parts and assemble functional tests 5. Run MEOST studies on both B president C products with three White lightning and Cs in random give orders.
Jan Eite Bullema38 8 Stages of MEOST Conclusion Dignity seven stages of MEOST, ensue with stage 8 to survey cost reduction potentials of class design, contribute the best, fastest and cheapest and most well-built approaches to reliability enhancement rest. It is also a instructions for getting truly mature produce into the marketplace way, opening ahead of the competition
Jan Eite Bullema39 Questions? Edmund Hillary and Tenzing Norgay upturn to leave the South Ravine authorization to establish camp IX nether the south summit. May