The Challenge:
50 years ago, electronics in a car was more or less restricted
to lighting and ignition. The overall schematic fitted on
one sheet and a handful of engineers mastered the related
functionality. Nowadays, things have become much more complex.
In the application fields power train, safety, comfort and
infotainment, thousands of microelectronic chips are designed
into a modern car and the embedded software content is measured
by hundreds of megabytes.
On top, electric mobility poses further challenges. Revolutionary
changes in a car's architecture will have to be implemented.
And while today's level of robustness of classical cars
has been accomplished over 125 years, at least the same
level has to be accomplished within 10-15 years for hybrid
and pure electric vehicles.
In effect, the complexity of any electronics in a car directly
drives the robustness requirements. Doubling the electronics
content simply calls for cutting failure rates of electronic
components in half, to maintain the overall quality level.
For integrated circuits today, the demand simply is: "zero
defect". Even more, the impending change from internal
combustion engines to electric motors with the related plethora
of innovations seriously exacerbates the situation, as in
many cases, we cannot rely on the experience with proven
concepts and components any more.
The Project RESCAR 2.0:
To address this situation, six partners covering the major
stages of the automotive value chain have formed a consortium
and set up a research project called RESCAR 2.0. These partners
are: AUDI AG, BMW AG (associated project partner), ELMOS
Semiconductor AG, Forschungszentrum Informatik (FZI), Infineon
Technologies AG and Robert Bosch GmbH. They will be supported
by the Fraunhofer-Institute IZM, Berlin, the Fraunhofer-Institute
IIS, Dresden, as well as the Universities of Bremen, Dresden,
Hannover and Tübingen. The project is coordinated by
Infineon Technologies AG. The goal of RESCAR 2.0 is to substantially
enhance the robustness of new electronic components for
applications in the field of electric mobility. RESCAR 2.0
is funded by the German Authorities (BMBF) with roughly
€ 6.5 million. The partners will jointly invest about
€ 13.3 million in this research activity.
Primary Approach:
The fields identified to address the improvement of robustness
in RESCAR 2.0 are:
- Define a robustness metric
- Set up a development flow to implicitly design-in the specifically required robustness
- Validate electronic systems with regards to their required robustness by simulation and measurement
- Identify the critical parameters on component and process level to maintain the robustness within the overall electronic system
The RESCAR approach is built upon the "Robustness
Validation" methodology, which was developed on a broad
consensus within the automotive industry in 2005 - 2009
and led to the publication of several handbooks1.
The major aspect, which deeply affects all the above points,
is how the electronics is used. The approach here is to
combine the knowledge of the partners on their various levels
along the value chain to better establish robustness targets.
The car manufacturer has a clear idea on how electronics
is applied. For example, power switches used to fire an
airbag may be switched only a very small number of times.
Power switches in a bridge to automatically shift gear -
especially if used in pulse-width modulation - may be switched
millions of times. Many more pieces of information, e.g.
temperature, voltage and current profiles, may be of interest.
All this information will be collected in a systematic way,
thus creating a standardized mission profile. This mission
profile is refined and translated along the value chain
firstly from car manufacturer to the Tier-1 supplier, which
provides the electronic control unit. The next step is from
Tier-1 supplier to Tier-2 supplier, responsible for the
development and fabrication of microelectronic circuits.
On this level, robustness needs to be addressed on the basis
of the related microelectronic failure mechanisms. One major
challenge of RESCAR 2.0 will be to bridge the gap from how
electronics is applied to how this affects potential failure
mechanisms. The forward translation maps the application
details to the stress applied on the microelectronics, which
in turn leads to degradations or complete failures of devices
or connections. In contrast to this, the backward translation
maps the occurring degradations or failures back to the
application level on which their impact can reasonably be
assessed.
Robustness Metrics:
We all know that it is difficult to improve a certain level
of quality, if there is no undisputed way to measure the
level of quality. This also holds for the robustness of
electronics. Currently, there are a couple of approaches
on the market to assess robustness. One way is to evaluate
how far the electronics' ratings and operating ranges can
be exceeded with regard to what is promised in the specification
or how far the parametric properties of the electronics
stay wide inside the specification windows.
Other approaches elaborate on the verification space, which
is spanned by temperature, supply voltage, and the many
more parameters defining the electronics and the rest of
the application. Think for instance about an electric drive
train and the related power bridges. Here is a small choice
of the parameters (or factors) in question:
Environment:
- Temperature
- Supply voltage
Electronics:
- Switching slew-rate
- On-resistance of power switches
- Current measurement op-amp gain
- Free-wheeling behavior
- Regulation delays
Mechanical part of drive application
- Armature inductance
- Armature resistance
- Torque constant / backEMF
- Motor friction and inertia
- Drive train frictions and inertias
- Driving load conditions
Easily, we end-up with some 20 parameters or more and all
of these parameters have a specification window. This simply
means that verification is to be performed in this 20-dimensional
space and we need to fully cover this space for a meaningful
verification. Robustness could also reflect the quality
of coverage of this verification space.
One task of RESCAR will be to evaluate candidates for robustness
measures and to select one candidate or a combination thereof.
Design-for-Robustness:
Designing for robustness in the context of RESCAR 2.0 means
to translate the electronics' requirements into constraints,
which can be maintained and checked in the design software
for microelectronic products. This process works for analog
as well as digital constraints, though the focus in RESCAR
2.0 will be on analog and mixed-signal functionality. In this
way, the requirements compliance and the related robustness
is "built-in" rather than "tested-out"
of a design. Unfortunately, this does only work for a part
of the requirements.
Robustness Assessment by Simulation and Measurement:
When it comes to the assessment of robustness, this can
be accomplished on the basis of simulations and measurements.
Obviously, these simulations and measurements will go beyond
the classical design verification.
One aspect to shed more light on could be to quantify the
stress, which is applied on for instance the transistors
of a power stage. This can be done taking into account the
application and even statistical variations of it. Knowing
about the stress, the microelectronics experts devise the
related degradations of the devices, which lead to related
changes in the device models used for circuit simulation.
The resulting behavior could be then simulated.
Measurements can equally be planned to assess robustness.
First of all, they need to be carried out to check the simulations
results. Moreover, much more experiments can be accomplished
in measurements rather than in simulation, as the measurements
are typically carried out in real-time.
Those simulations and measurements will have to be done
under lots of variations of factors to check out their influence.
Tons of simulations or measurements will produce tons of
result data for which a data mining approach will be set-up
to extract the essence of robustness according to the defined
metric.
Conclusion:
The provision of robust electronics at affordable cost
will be a major challenge for the foreseeable future. This
holds in general for the automotive domain and to an even
much higher degree for the very innovative field of electric
mobility. RESCAR 2.0 pursues a new approach to address this
question. Its results will form a substantial contribution
to the big change towards electric mobility in the automotive
industry, which we are currently experiencing. This big
change will trigger new decisions on winners and losers.
Those missing robustness most likely will not be on the
winner's side.
Robustness requirements of automotive electronics will
be standardized in 'mission profiles' and translated into
constraints to be used throughout the design process, validation
and test in the whole supply chain
Authored by:
- Georg Pelz, Principal Design Methodology and Product
Modeling, Infineon Technologies AG
- Julia Lau, Manager Physical Design Methodology &
Flow Development, Infineon Technologies AG
