CALCE Research: Making Electronics Reliable
and Cost-Effective
written by
Thomas Ventsias
In 1985, when the
U.S. Army asked the University
of Maryland to update a
handbook for predicting the failure rate of electronic components, it seemed at
the time a fairly straightforward request.
The Military
Handbook for Reliability Prediction of Electronic Equipment (known as
MIL-Hdbk-217) and its commercial equivalents were widely viewed at the time as
the industry standard all over the world. The goal behind MIL-Hdbk-217 was to
ensure that the electronic components, such as transistors, diodes, resistors,
capacitors, and switches, used in electronic systems could operate for long periods
of time under demanding conditions.
After reviewing
the document, however, Maryland
researchers concluded that MIL-Hdbk-217 and the testing procedures behind it
were basically flawed. "The methods in place [in 1985] were statistically
oriented" says Peter Sandborn, associate professor of mechanical
engineering in the A. James Clark School of Engineering. "[The Army] would
ship 100,000 circuit boards, and after a period of time, would see how many
came back with problems." From that information, he says, the U.S. Military
built a model to determine how many circuit boards would fail. The major
shortcoming of MIL-Hdbk-217 was its inability to address the fundamental
cause of why and how electronic components and assemblies would fail over
time. Consequently, these models were inaccurate, inapplicable as a predictive
tool for new technologies, and could not be used to improve product design.
Researchers at Maryland offered an
alternative. With research and input from across the academic disciplines, faculty
in the Clark School started to push for a new process
called "physics of failure (PoF) analysis". Based on the fundamental
"physics" of each failure mechanism in each component (how it is
built, how it operates, and under what conditions it fails) new models could be
generated to predict aging, degradation, and failure over time. Using this same
PoF knowledge, Maryland
researchers soon developed new software that could "virtually"
qualify electronic components via computer simulation.
These new ideas
would form the basis for one of the largest research centers within the Clark
School of Engineering.
The Center for
Advanced Life Cycle Engineering (CALCE) Electronics Products and Systems Center
(EPSC), established in 1986, is now an internationally recognized leader in
reliability assessment of electronics based on PoF analysis.
CALCE has grown into a consortium that has received almost $45 million in
combined research support in the past 15 years. The center employs more than
100 faculty, research staff, and graduate students from almost every engineering
discipline.
Much of CALCE's
research is driven by the 50 - plus industrial partners who make up the CALCE
consortium. A virtual who's who of leading electronics, aviation, automotive,
semiconductor, computer and telecommunication companies like Lucent, Microsoft,
and DaimlerChrysler, the consortium promotes research in areas that have an
across-the-board impact on industry.
Current research
at CALCE focuses on applying PoF knowledge for complex tasks
such as designing electronics for reliability; accelerated testing; life
consumption monitoring; supply chain management and parts obsolescence
modeling; reliability of microelectromechanical systems (MEMS) technology;
the role of electromagnetic interference in the design of electronics; and
developing new environmentally friendly materials such as
lead-free solders for "green" electronics.
An example of
CALCE research is its fundamental work on computational failure models such as
micromechanical simulations of cyclic loading in viscoplastic polycrystalline
alloys like solders. These models capture complex creep fatigue damage
processes under vibration and thermal cycling of electronic interconnects.
Motivated by the understanding obtained from such detailed models, simpler
models have been developed to facilitate design. These simulation approaches
have been implemented into virtual qualification software for timely
reliability assessment.
Getting It
Right the First Time
In today’s fast-
paced electronics industry, where time-to-market means everything, the use of
physics of failure reliability modeling and virtual qualification can
significantly shorten the time interval between design iterations.
"A month late
to market on a mobile phone can be a financial disaster," says Peter
Sandborn. But that's exactly what can happen if a manufacturer builds a phone
and things start to fail in the initial production phase. Or worse yet,
Sandborn says, if a company ships "bad" phones with inherent design
flaws, not only do expenses increase dramatically by having to cover warranty
costs, but the potential loss of customers can be extremely damaging.
"There are big ramifications if these things go wrong," he says,
"and a big upside if they can get it right the first time."
Helping design
engineers "get it right" in the early stages of product development
is a large part of CALCE's mission. This is accomplished by providing designers
with the know-how to design in reliability rather than using a trial-and-error
iteration to achieve reliability after the product is designed. When the CALCE
approach was used on a particular General Motors product, they reduced
development time by over 10% and increased first pass success by over 60%. PoF
software tools developed at CALCE for analyzing circuit cards (calcePWA) and
components (CADMP II) greatly facilitated that effort. CALCE's
Physics-of-Failure software and methods have saved over $80 million in U.S. military
programs, several million dollars on Westinghouse radar systems, over a million
dollars on an AlliedSignal engine control module, and similar amounts on other
military and commercial programs. Numerous commercial suppliers are now working
with the CALCE to qualify new technologies, demonstrate reliability of new
hardware, and trouble shoot existing warranty returns and field failures.
"With the
scope, volume and time to market factor in the electronics industry today,
there are a lot of people who are very interested in the types of problems that
we are working on here at CALCE," says Sandborn.
A Smart
Approach to Testing
CALCE has earned
much of its international recognition for its breadth of expertise in
developing accelerated life cycle testing methodologies for electronic parts
and assemblies. Most integrated circuit boards or electronic assemblies are
tested in the prototype stage to see if they will meet their expected life
cycle. For electronic components used by the military or the aerospace industry,
almost all need strict certification by agencies such as the Federal Aviation
Administration.
Using
state-of-the-art laboratories, researchers at CALCE have developed accelerated
test methods that can effectively put many year's worth of wear and damage on
an electronic component controllably in a few days or weeks. "It is a
deceptively complex thing to do," says Abhijit Dasgupta, professor of
mechanical engineering. "First, you need to understand what are the
'relevant' aging mechanisms, he says. "There's no point in
designing an accelerated test that will stimulate failures that the product
won't ever experience in the field but it's still done [in industry] all the
time. Then you have to quantify the extent to which the test environment has
accelerated the relevant aging mechanism, so you can extrapolate the test
results to the use environment."
After determining
the relevant testing factors, researchers at CALCE can create conditions of
high temperature, high humidity, vibration, shock and impact, electrical
stresses, and even high pressure or radiation if the electronics will be used
in avionics or spacecraft.
In CALCE's high-
altitude chamber, for example, researchers have represented changes in air
pressure from sea level to 60,000 feet in the same time frame that it takes for an
F16 jet fighter to do the same. "Companies who install a lot of electronic
equipment on an aircraft, British Aerospace, and Honeywell, for example, have come
to us to understand the effects of this type of environmental change on electronic
failures," says Chris Wilkinson, a researcher at CALCE.
Putting the
Parts in Place
Supply-chain
management is a hot topic nowadays, and CALCE is leading the way in research
involving electronic parts selection and obsolescence. Parts obsolescence
is a major concern for low-volume, long-life electronic systems like those used
in modern aircraft. One of the largest cost factors in designing a new airplane
is maintaining an uninterrupted supply of parts.
When Honeywell
wanted to build a new engine controller for an aircraft with a 10 year
production life, they came to CALCE to optimize their parts selection for
reliability and cost. "Due to certification requirements you can't change
microprocessors on an airplane every 12 months like the PC industry," says
Sandborn. With modern computer-generated models to predict obsolescence of
electronic parts, CALCE can tell design engineers which part to put into their
system, and can also predict when that specific part is likely to need replacement
due to failure or obsolescence.
"CALCE was
established to meet the needs of a growing electronics design and manufacturing
industry," says Michael Pecht, professor of mechanical engineering and the
director of CALCE. "We have grown significantly in 16 years, and can
readily offer the latest resources and tools to help engineers assess, mitigate,
and manage risks in electronic products. Our goal is simple: to offer the
highest quality research environment to our sponsors, and to provide the
world's best knowledge base for building reliable, competitive electronic
systems."