2007 Introduction MEOST

Jan Eite Bullema
Technology Manager Micro Technology
World Class Reliability
MOEST:
Multi Environment Over Stress Testing
Climbing the Mount Everest of Stress Testing
Source: World Class Reliability, ISBN 0-8144-0792-7 Keki R. Bhote
Session 3 MOEST
TC Design Methods for Reliable Lead-free Products

Jan Eite Bullema2
Reliability Approach
1 Identify product functions
2 Identify mission profile / normal usage conditions /
Design Tolerances/ Engineering Tolerances / Customer Tolerances
3 Estimate potential root causes for failure (FMEA)
4 Define appropriate tests / accelerated tests
5 Execute tests / accelerated tests / MEOST
6 Analyze products and failures
7 Identify root causes for failures in tests
8 Statistical analysis (preferably through Weibull)
9 Determine acceleration factors were appropriate
10 Predict reliability behavior under normal conditions,
identify design flaws
11 When necessary iterate from appropriate step
12 Reporting

Jan Eite Bullema3
Biography Keki R. Bhote
Author of World Class Reliability
Mr. Keki Bhote was born in Bombay, India in the 2nd quarter of the
20th century, where he recevied his early education. He received
his B. Sc. degree inTelecommunication engineering from the
university of Madras, India, and his M.Sc.in Applied Physics and
Engineering Sciences from Harvard University. Upon
completion of his studies, he joined Motorola as a development
engineer, and rose through the ranks to become Group Director of
Quality and Assurance for Motorola Automotive and Industrial
Electronics Group before promotion to
Senior Consultant for the entire corporation
world-wide. Since his retirement from Motorola,
Keki has been in private proactive providing
Telecommunication consulting.

Jan Eite Bullema4
Why MEOST?
Dramatic Reduction in Development Time
Multiple Environment Over Stress Tests (MEOST), first developed at
NASA to test the Lunar Module, achieves:
- Reliability levels of 10:1 to 100:1 over traditional field reliability.
- Reductionsin design validation time from over 16 weeks to less than 2 days.
- Reductions in design test costs by factors of 5:1.
- Reductions in design sample sizes by factors of 10:1.
- Faster designs to the market, leaving competition in the dust.
MEOST Principles are:
- Testing to failure is more important than success testing.
- Environments/stresses must be combined to produce interaction failures.
- Stress levels must go well above design stress – almost to total destruct levels
to reduce the time to failure.
- The rate of stress level increases must be very rapid to reduce time to failure
by one or two orders of magnitude.

Jan Eite Bullema5
Why MEOST?
Large Variances in Failure Estimates
National Handbook FITs % per year
MIL Handbook 217 (US) 4240460 317.3
BT (Brittain) 700 11,6
CNET (France) 37870 33.0
NTT (Japan) 37940 33.1
Prediction for the same large memory board, the difference is over 600 : 1.
So much for cookbook accuracy! [Kam L. Wong, The bathtub curve does not hold water
anymore', Quality and Reliability Engineering Symposium 1988]
The handbooks consider part complexity, part technology, package technology,
part application, electrical stres, temperature and manufacturing quality control
level for their failure rate projections. They do this with exponential
extrapolations and multipliers.

Jan Eite Bullema6
1. Why do the Handbooks Fail?
The handbooks consider part complexity, part technology, package
technology, part application, electrical stres, temperature and
manufacturing quality control level for their failure rate projections.
They do this with exponential extrapolations and multipliers. What
is not considered (or limited) are the following factors:
1. Temperature cycling.
2. Failure rate changes with material.
3. Vibration
4. Nonoperating failure rates.
5. Combined Environments.
6. Supplier Variations Affecting reliability.

Jan Eite Bullema7
1. Temperature cycling. Failures accelerate with 1) the number of thermal
cycles and 2) the wider range from cold to hot. Failures can increase
sevenfold with these two factors alone. In addition, the rate of temperature
change - from exemple 1 - 2 degrees per minute to 25 to 40 degrees per
minute generates far more rapid failures.
These are very important in MOEST tests.
2. Failure rate changes with material. For brittle materials, a higher rate of
temperature change accelerates failures whereas ductile material such as
aluminium and solder, a lower rate of temperature change accelerates failure.
3. Vibration: sensitivity to exact installation and mounting structure. The amplitude
of vibration can vary witing a product, depending on the exact installation and the
mounting structure. Each unit would have its own response spectrum and, hence,
its known failure rate when subjected to the same vibration input

Jan Eite Bullema8
4. Nonoperating failure rates. It may come as a surprise that non-operating
products systems are still subject to environmental stresses. These non-
operating or zero stress- failure rates vary inder different storage or shipping
conditions. Many missile storage studies have found that failure appear to be
independend of the length of storage. This strongly suggest that turn-on was
the culprit in most of these failures.
5. Combined Environments. While temperature cycling and vibration are the
two most important stress/environments causing failures, the handbook do not
take into account other stresses, such as humidity, dust, altitude, power
cycling, transients or radiation. Yet their effect in simultaneous combination
can not be ignored.
6. Supplier Variations Affecting reliability. Handbooks do take into account
reliability specifications imposed on suppliers, such as verification of parts
mechanical integrity, long term measurement failure rates, mimimum life
expectancy, extend of parametric measurements, and the amount of
environmental screening . But not all reliability requirements can be specified;
if they were, the costs would be prohebitive.

Jan Eite Bullema9
First Application of MEOST in the Apollo Lunar Module

Jan Eite Bullema10
MEOST = Multiple Environment Over Stress Testing
Bhote: ‘Climbing the Mount Everest of Reliability Testing’
From base camp to the summit
Life Testing
Burn-In Testing
Cycling Testing
Single Environment Testing
Accelerated Testing
Multi Environment Testing
Source: World Class Reliability, ISBN 0-8144-0792-7 Keki R. Bhote

Jan Eite Bullema11
Life Testing
Bhote:’Reliability Demonstrating: Throwing Money at the Problem’
In the early life testing of the 1960s, a sample of product was simply
allowed to ‘cook’ for 1000 to 3000 hours tot detect failures. As
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
temperature soak for twenty-four to ninety-six hours, on a 100 %
basis.
Minor tinkering around the edges added electrical power to the units
subjected to burn-in, along with the cycling of power ‘on’ and ‘off’
Strangely this is still the preferred method of demonstrating
reliability.
Even more incredibly, many customer insists on burnin as proof of
reliability.

Jan Eite Bullema13
Thermal Cycling Testing
Bhote: ’The Dawn of True Reliability Progress’
In the mid-1960s the principle of exercising stress on parts and
products through temperature cycling was introduced. Within a
decade thermal cycling advanced in seven stages.
Stage 1: Thermal Cycling, no electrical power
Stage 2a: Thermal Cycling, continuous power
Stage 2b: Thermal Cycling, interrupted power
Stage 3: From 0 ºC to 50 C , 1 cycle
Stage 4: 5 Cycles with Measurements at Temperature
Extremes
Stage 5: Extension from – 30 ºC to 85 ºC, 25 Cycles
Stage 6 Extension to > 100 Cycles

Jan Eite Bullema14
Vibration Cycling Testing
A companion development
Parallel to thermal cycling, advances were also made in vibration
to simulate product failures
Stage Specific Technique
1 Sinusoidal
2 Single Axis, Single Frequency
3 Sine Sweep
4 Random
5 Random with 6 degrees of freedom
(in 3 distinct axes and 3 rotational axes simultaneously)
Other Stresses/Environments
Thermal Shock, Humidity. Power Cycling, Voltage Margining,

Jan Eite Bullema15
Crude Approximation of the Relative Influences of
Various Stresses and their Actions
Thermal Shock
Thermal
Humidity
Corrosive
Dust
Vibration
Power Cycling
Voltage Margining
Frequency Margining

Jan Eite Bullema16
Single Environment Testing
Bhote: ’Another False Start’
In an attempt to achieve a better reliability, some automotive and
electronic companies initiated a long and tortured test regimen,
wherein a product would be subjected to a series of separate,
single environment stresses, but in sequence.
A typical sequence would be: test, thermal cycling; retest; vibration
retest, humidity, retest, etc.

Jan Eite Bullema17
(Highly) Accelerated Testing
Methodology
Performed at the prototype stage of design, HALT stresses a product well
beyond design specifications right up to destruct levels, or a fundamental
level of technology.
One interpretation of a destruct level is that it’s a level stress at that level
where a small (further) level of stress causes a large increase in the
number of failures.
A third interpretation is the fundamental limit of technology (FLT), which is
defined as that level of stress were the product disintegrates.
HALT generates multiple failures, a few in the lower stress levels
and a large number of rapid failures as the stress approaches destruct
levels.

Jan Eite Bullema18
Accelerated Testing
An Overview
HALT:
HASS: Highly Accelerated Stress Screening. Is a 100% test
screen with stresses higher than field stress. High enough to
catch potential field defects, but leave the rest of the product with
> 80% of its useful life
HASA: HASA starts with 100 % sampling, but allows for lower
sample sizes if the number of failures drops below a specified level
ESS: Environmental Stress Screening came into vogue twenty-five
years ago as an alternative to burn-in and mil spec series. HALT
and HASS were the off spring of ESS

Jan Eite Bullema19
Preparation for MEOST
1. Choosing Appropriate Product Levels
2. Prioritize current Field Failures
3. Ruggidizing a Product for MEOST
4. Setting appropriate stresses in combination
5. Determine Limits (Design, Operational, MPOSL and Destruct)
6. Determine the number of stress levels
7. Allowing enough Dwell time at each stress level
8. Establish a Combined Stress Scale
9. Prepare the Stress Sequencing Roadmap
10. Determine Outputs (green Ys)
11. Setting up adequate support Equipment
12. Choosing Sample Size for MEOST

Jan Eite Bullema20
Typical Stress Levels for MEOST
200 % Destruct Stress
Either continuity of failure (HALT) or fundamental level of
technology)
170 % Maximum Practical Over Stress (MPOSL)
MPOSL is midway between operational and destruct level
130 % Operational Stress
Operational stress is that stress that which, when reduced,
causes the failure to be reduced (HASS level)
100 % Design Stress
Highest of: (1) Engineering Specifications, (2) Customer
requirements, (3) Maximum Field Environment.
0 % Room Ambient

Jan Eite Bullema21
Choosing Sample Size for MEOST
Recommended Guidelines are as follows
MEOST Stage Sample Sizes
Prototype (stage 3) 3 for repairable
5 to 10 for non-repairable units
Pilot run (stage 4) 5 to 10 for repairable units
Subsequent stages 5 to 10 for repairable units

Jan Eite Bullema22
Sample Size for MEOST (1)
One of the most frequent doubts expressed by managers and
engineers exposed to MEOST for the first time is: How can the small,
no tiny sample size of 3 to 10 units in MEOST testing adequately
represent the total population of the product? After all, statistics tell us that
sample sizes of 30 to 50 to 100 units are minimum required.
The answer is that in MEOST, we are not concerned with failures as a
percentage of the total number of units tested. We want to probe the
weakest components that, by the laws of physics, have stresses two or
three times higher than the strongest components and therefore are likely
to fail the quickest.

Jan Eite Bullema23
Sample Size for MEOST (2)
Under normal field conditions it would take a long time for even these
weak components to fail. However, Miners equation indicates that for even
doubling the stress, the failure rate can jump 210
= 1024 times. As a result
the distribution of failures for several components, which may be bunched
up under benign field conditions, are going to spread out under
accelerated stress. And there will be a wide separation between the early
failures of the weak components and the late failures (0r no failures) of the
more robust components.
In MEOST stages 3 to 8 one failure per failure mode is allowed in
overstress regions. The reason is that a single, lone failure may
represent an anomaly. As a change occurrence, it represents an extreme
low end of the failure distribution of that component and can be ignored.
Two failures of the same failure mode is the start of a trend and should be
analyzed and checked.

Jan Eite Bullema24
Preparation for MEOST
1. Choosing Appropriate Product Levels
2. Prioritize current Field Failures
3. Ruggidizing a Product for MEOST
4. Setting appropriate stresses in combination
5. Determine Limits (Design, Operational, MPOSL and Destruct)
6. Determine the number of stress levels
7. Allowing enough Dwell time at each stress level
8. Establish a Combined Stress Scale
9. Prepare the Stress Sequencing Roadmap
10. Determine Outputs (green Ys)
11. Setting up adequate support Equipment
12. Choosing Sample Size for MEOST

Jan Eite Bullema25
The Eight Stages of MEOST
1. Single Stress Up to the Design Limit
2. Single Stress Up to the Maximum Practical Over Stress Limit
3. Prototype – Full MEOST to Maximum Practical Over Stress Limit
4. Pilot Run
5. Mini-MEOST in Outgoing Production
6. First Round of MEOST on Field Returns
7. Second Round of MEOST on Field Returns
8. Cost Reduction

Jan Eite Bullema26
Stage 1 of MEOST:
Single Stress Up to the Design Limit
This is a preliminary stage to determine the failure contribution, if
any, of each single stress selected in the four or five stresses that
will eventually be used in combination.
- Step-stress in three or four stress levels, from room-ambient
benign stress up to the design limit for that stress
- Start with thermal cycling -20 to 80 in 40 C per minute, applying
a dwell time of 10 minutes, start cold then hot
- For vibration , start with zero and go up to design stress
- Repeat with other single stresses , such as humidity, voltage,
transients, and shock
- If there is even a single failure, correct it and validate the
effectiveness of the correction using a B vs. C test

Jan Eite Bullema27
Stage 2 of MEOST
Single Stress Up to the MPOSL
Stage 2 is also a preliminary stage to determine the effect of
overstress of each single stress used in Stage 1.
- Continue stage 1 beyond the design limits for each stress on the
same unit (if repairable to MPOSL)
- If there are no failures in the overstress region continue testing a
few cycles for a few hours
- If there are still no failures we can conclude:
- The stress type is inadequate
- Te rate of stress is too slow
- The test has not been executed properly

Jan Eite Bullema28
Stage 3 of MEOST (1):
Prototype-Full MEOST To MPOSL
This is the most important MEOST stage. When completed is
assures the designer to the best advance possible in reliability by
forcing the weak links in design to be smoked out.
- Select the four/five stresses that are likely to impinge
simultaneously on the product in the field
- Prepare a combined stress sequencing roadmap
- Use the same units that have survived stage 1 and 2, if possible
and subject them to combined stresses
- Start at design stress and then step 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
Example MEOST test plan
-60
-40
-20
0
20
40
60
80
100
120
1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91
time (hrs)
Various()
Time of Day
RH
Temp Cycl
Voltage Cycling
Load Dump
Field Decay

Jan Eite Bullema30
Typical MEOST Test Profile
-50
0
50
100
150
200
0 50 100 150 200
Various ( )
time(minute)
Temperature (C )
Vibration
Voltage

Jan Eite Bullema31
Stage 3 of MEOST (2):
Prototype-Full MEOST To MPOSL
- If there are two or more failures per failure mode, that are
different from the predominant failure in the field, then there are
several possible reasons: (1) field data not OK , (2) stresses
have to be added, (3) stress levels or rate has to be increased
- Perform another round of stage 3 with deliberate ‘seeded
defects’ to confirm MEOST effectiveness
- Design improvements to correct the above failure mode(s) and
validate the effectiveness of the improvement with B vs C tests
Because there are only a few prototype samples that can be
spared for any kind of testing, the sampling is 3 to 5 units

Jan Eite Bullema32
Stage 4 of MEOST:
Pilot Run
This stage of MEOST ensures that design improvements/changes,
tooling, suppliers, processes, and fixtures have not adversely
affected design reliability
These steps are as follows:
- Run a stage 4 MEOST, using the same guidelines as stage 3,
with new units from an engineering or production pilot run.
- A successful outcome means that the design is now ready for full
production.

Jan Eite Bullema33
Stage 5 of MEOST:
Mini-MEOST in Outgoing Production
Stage 5 ensures that reliability integrity of the design is not
Degraded by manufacturing processes, workmanship, and supplier
materials. The steps are as follows:
Repeat Stage 3, with two major exeptions
- Reduce one or more of the four or five stresses used in stage 3
and 4
- Reduce the overstress from the Maximum Practical Overstress
Level to the operational level (approximately one-third above the
design stress)
Sample size 3 to 5 units for production runs of 100 to 1000 units
per day. Never use 100 % sampling. 1 to 3 units for production
runs of less than 100

Jan Eite Bullema34
Weibull Chart
Time to Failure (Hours)
CumulativeFailureProbbility
Weibull Chart

Jan Eite Bullema35
Stage 6 of MEOST:
First Round of MEOST on Field Returns
Stages 1 to 5 are all performed virtually at time zero in their product
life. Hence plotting the failures or stresses on a Weibull plot will
yield only one point in the graph. Stage 6 purpose is to secure a
second point on a Weibull plot after a period of exposure in the field
– typically six months in service
- Make arrangements with a trusted, competent customer
- Request 5 – 10 good units to be returned in exchange for new units
- Subject these to a stage 3 MEOST starting with Design to MPOSL
- Record the percentage of overstress when there are 2 or more
failures
- On a Weibull plot, record the time to failure for time = 0 (overstress
stage3) and t = field exposure (overstress stage 6)
Now we have two point on a Weibull Chart

Jan Eite Bullema36
Stage 7 of MEOST:
Second Round of MEOST on Field Returns
It takes at least three points on a Weibull plot to draw a best-fit
straight line connecting them and extrapolating the straight line.
A new point is generated using the same procedure as Stage 7
with the difference that a longer exposure time in the field is required
-typically one year in service.
- Add a point on the Weibull chart depicting one year in service
With three points on the Weibull plot – at time zero, six months and
one year. Draw he best fit straight line and extraplate is until it
reaches the design stress horizontal line.
Projected on the x-axis, this intersection records the years to failure
and hence depicts reliability of the product.

Jan Eite Bullema37
Stage 8 of MEOST:
Cost Reduction
Use a value engineering approach:
1. List all the high cost items in the product and prioritize them in terms of
highest costs and importance.
2. For the top three or four of the high priority items, use a value
Engineering approach to determine their function
3. Using brainstorming and other related methods, determine what other
part will provide the function at least cost,
4. Substitute the value engineered parts for the more expensive original
parts and run functional tests
5. Run MEOST studies on both B and C products with three Bs and Cs
in random order.

Jan Eite Bullema38
8 Stages of MEOST
Conclusion
The seven stages of MEOST, along with stage 8 to explore cost
reduction potentials of the design, contribute the best, quickest and
cheapest and most powerful approaches to reliability enhancement
known.
It is also a recipe for getting truly mature product into the
marketplace way, way ahead of the competition

Jan Eite Bullema39
Questions?
Edmund Hillary and Tenzing Norgay about to leave the South Col
to establish camp IX below the south summit. May 1953

2007 Introduction MEOST

More Related Content

What's hot (20)

Similar to 2007 Introduction MEOST (20)

More from Jan Eite Bullema (17)

Recently uploaded (20)

2007 Introduction MEOST

Editor's Notes