TOOL SUPPLIERS HAVE RISEN TO THE CHALLENGE POSED BY VEHICLE makers and auto electronics suppliers to help reduce system problems, reduce system cost, and reduce time to market. In general, this means greater involvement with tools earlier in the design process. The answers are similar in some cases and quite different in others but in either case, provide options for users. Standardization efforts with strong support from OEMs, tier one and semiconductor suppliers as well as design tool suppliers have made significant progress. This report focuses on the design tool options available to help electronic systems designers cope with the increasing complexity in today's vehicles.
The increasing system complexity in a modern vehicle has been quantified in several areas. According to Gartner/Dataquest report, Electronic Production and Semiconductor Consumption, Nov-ember 2005, worldwide semiconductor consumption for automotive electronics is projected to grow 6.9% from 2005 to 2010. At the same time, the percent of the average vehicle's cost due to automotive electronics could increase from about 23% in 2004 to 40% in 2010.
From a software complexity standpoint, there is an estimated 4 million lines of code in a 2006 vehicle. “The complexity is in-creasing,” says Chris Washington, product manager, LabVIEW, National Instruments. “The amount of time these engineers have to get these products out the door that are increasing in complexity is shortening. The quality standards are increasing and more and more the differentiators, specifically in electronics, differentiators in vehicle and other products, oftentime becomes the firmware — the embedded technology.”
Figure 1 shows the distribution of warranty expenses in vehicles. According to the 2003 HAWK study, more than 50% of all breakdowns directly result from faulty software and electronics. At the same time, as shown in Figure 2, the cost to correct these problems increases as the program progresses from the design to the coding and finally into the test phase.
Knowing these facts provides justification for the increased use of design tools but not necessarily the solution. So the OEMs have challenged the tools suppliers to help solve the problem. “The high-level conceptual answer is more simulation, and less physical prototyping and less manual testing,” said NI's Washington. Even in the most innovative design approach, with early simulation, OEMs can find problems earlier in the design or development cycle. There is a side benefit as well. Increased use of simulation reduces the number of prototypes that have to be built, which involves considerable cost and time at the complete vehicle level.
To address the problems sooner, automotive OEMs are pursuing more robust designs. “The issue with robust design is how do you design systems that are repeatable, predictable, reliable and proven,” said David Smith, Synopsys Scientist. Smith is also the chairman of SAE's Electronic Design Automation (EDA) Committee. Since, the committee's efforts are driven by OEMs, its activities provide interesting insight into what the car companies expect.
The SAE EDA committee was established to create and promote standards for the development and use of electronic design automation practices and tools within the Ground Vehicle industry. There are four areas that the committee is dealing with, driven by automotive OEMs:
- Model-based embedded systems engineering (MBESE).
- Supply chain model exchange.
- Data dictionary (standard component definitions).
- General area of design and verification methodology, which has been assigned SAE document J-2748.
In the first area, MBESE has widespread acceptance of model-based engineering or model-based design for software development and several companies also model physical processes. The committee wants to link verification of both models together. As Smith pointed out, they have techniques for doing these things separately — the trick is doing them together. That is one of the areas the committee is looking at. Half of the effort is a design methodology problem and half is a tool problem.
The second area, supply chain model exchange, deals with improving the modeling process throughout the supply chain. Tool providers can help by developing modeling tools that people can use to simplify the process of creating models and characterizing them.
The third area, standard component definitions, is required to avoid different interpretations of terminology.
In the fourth area, design and verification methodology, the proposed J2748 VHDL-AMS statistical analysis packages standard is already in the approval cycle. The committee has approved the proposed draft and it is in the next step toward becoming a recommended practice. J2748 will add statistical modeling support into the VHDL-AMS language.
The SAE is certainly not the only organization developing standards to support improved system design that impact design tools.
The AUTOSAR (AUTomotive Open System ARchitecture) initiative (http://www.autosar.org/) was introduced at the VDI conference in Baden-Baden, Germany three years ago and has made significant progress. Its goal is to allow embedded software to be viewed as an interchangeable component within the automotive supply chain. Today, it has had a 2.0 release. Mentor Graphics is among the design tools suppliers who are members of the initiative. They also worked on the 2.0 release package. “From a vehicle standpoint, you are going to see adoption of AUTOSAR target packages in the 2007-2008 start of production time,” said Larry Anderson, transportations business development director, Mentor Graphics. One of the key benefits of AUTOSAR to OEMs is the ability to reuse software modules. Figure 3 shows the AUTOSAR architecture.
One of the historically complex and problematic areas of the vehicle is the wiring harness. Today, an average vehicle wiring harness consists of 800 wires, 2.5 km in length, and 1800 connectors. “In a complex luxury sedan you can have four to five kilometers of wiring,” said Mentor Graphics' Anderson. “Next to the engine, it is usually the second most expensive component.” Mentor Graphics tools have been used to help OEMs determine the optimum silicon-copper trade-offs. This can avoid surprises that can occur when consolidating functions into a single ECU. The adverse wiring impact could outweigh the savings from the silicon-level integration.
The wiring harness impacts the overall vehicle electrical and electronic systems. “As electronics have increased dramatically in cars, so have the power distribution and management issues,” said Paul Latiolais, senior marketing manager, Synopsys. “What happens is if you work with traditional methodologies you are going to miss this.” The increased control systems and components require a different design methodology for the engineers who routinely have dealt with wiring harness issues.
One of the problems that Synopsys has addressed is signal integrity analysis of wiring harness networks. With the vehicle's complex, distributed computing systems, the wires connecting these systems include twisted pairs, shielded wires, wire bundles, harnesses and more. One design tool aspect requires performing signal integrity on the data network to determine how much ringing is occurring.
Another critical harness area involves power distribution. With added complexity and increased power in the vehicle's generation and storage systems, there has been an increase in the amount of connector/terminal melting as well as failed splices. To address this, Synopsys has a line of harness verification tools. However, power is a specific problem that is being addressed by several design tools suppliers. Problems frequently occur from the use of extra power adapters. “They need to be able to do the analysis now, to be able to make sure that they can meet the power requirements, all the permutations of the power requirements, plus any additional optional things people provide and still meet their cost criteria,” said Synopsys' Smith. Traditional overdesign is not acceptable because OEMs cannot afford it. Figure 4 shows an approach that Volkswagen used in working with Synopsys to simulate and analyze the vehicle's power network. Synopsys' Saber Simulator and MATLAB/Simulink were used in the process.
Wiring harness verification includes the power distribution in the wire harness, in-vehicle networking, and signal integrity analysis for the data networking, as well as power network analysis for generation, storage and load analysis. The addition of hybrid technology adds high-voltage inverters instead of simple dc to the vehicle's wiring harness complexity. This requires electromagnetic compatibility (EMC) analysis in the modeling as well.
Wiring is certainly a key aspect for design tools but to have a more complete look, other areas must be examined. Two additional topics include model-based design and hardware in the loop. Additionally, in this final section, different approaches to tools and methodology are demonstrated.
Model-based design (MBD) can be applied to many different areas to address control problems and for code generation problems. MBD use is essentially universal in powertrain development. In this process, both OEM and powertrain control suppliers use the MathWorks products. Some users develop models to test out and design algorithms and some have the goal of automating code generation. “Where a lot of them are moving now is in rigorous processes — making sure that they can version control both the models and the code,” said John Friedman, marketing manager for MathWorks. “They are also looking to link those codes back to their initial requirements and then to link the tests that they have created to the requirements.” Figure 5 shows the concept of model-based design. Verification of these models must occur at the component, systems, and total vehicle level.
With the established and recognized position of design tools in several areas, many companies work together to provide optimum synergy of each company's tools. These partnerships exist in several areas. For example, Synopsys' Saber, MathWorks' Simulink and dSpace's rapid prototyping tool are shown working together in Figure 6 to provide closed loop correlation in an electronic throttle model .
Hardware in the loop (HIL) testing fits in the design validation side of product development. Essentially, HIL takes the embedded controller with the control algorithm and connects it to a computer running a real-time operating system (RTOS). The controller reacts as if it was in an actual vehicle. This allows a much lower test cost, as well as quicker testing, more test iterations, and greater flexibility.
“You can do it more quickly, you can do more tests, and you can get greater boundary tests,” said NI's Washington. An example of a test that could be conducted using this approach is simulating an oil leak until the point that the engine no longer operates. With real hardware, this would be a single occurrence test. With HIL, engineers can test areas that may not be practical to test — such as those at or outside of boundary conditions. In addition, Washington pointed out an example where one LabVIEW user, MicroNova electronic GmbH, used a LabVIEW FPGA module and National Instruments' PXI-7831R reconfigurable I/O module to develop a 12-cylinder fuel injection HIL simulator. The HIL solution allowed development before the actual hardware was available.
Specific activities at design tools companies reveal some of the variances in approaching electronics complexity. Five examples include a chip-based starting point, networking, microkernel/RTOS, power electronics and testing.
“What we see is primarily in three areas: the first one is mixed- signal, analog/mixed-signal (AMS), the second is around verification, verification certainly at the chip level and the third is system-level design challenges,” said Kelly Perey, vice president, marketing, Cadence Verticals Cadence Design Systems Inc.
Cadence-focused areas correspond to the three automotive levels: semiconductor, tier ones and OEMs. “The verification and system-level design issues are particularly interesting because they cross all three segments,” said Perey. “And what those three segments need varies slightly.” Cadence has a methodology called Plan to Closure that provides system-level verification all the way down to verification at the chip level. Customers in automotive and tier ones such as Bosch use the Plan to Closure system.
From its roots in semiconductor technology, Cadence links to the system level to other tools such as MathWorks, which is heavily used across automotive to capture system-level information. These links include language-level links into Cadence analog and mixed signal simulators that allow Cadence to validate that the chip works in the context of the system specification in MathWorks. This produces a verified model of the IC and verifies that the model works inside the system specification.
In powertrain and chassis control, the increased usage of the CAN and other protocols in feedback control systems that have to communicate with each other on a relatively frequent basis is an area of interest to Mentor Graphics. To simulate the actual network traffic, the user enters all the producers of signals and all the consumers of signals (i.e., engine speed, temperature, air flow, etc.). Mentor Graphics tool schedules all the signals to make sure the maximum latency of the signal is not exceeded and automatically packs all the frames in the desired protocol for data transport. A signal that does not satisfy the requirements is flagged for a design change.
For automotive electronic control units with minimal or no off-chip memory, such as those in in-car head units, handsets, dashboards and body electronics, Green Hills Software Inc.'s recently introduced micro-velOSity provides a royalty-free real-time microkernel and new CAN driver software for next-generation office-in-the-car systems. Addressing the low end in the platform architecture with a ROM footprint as small as 1600 bytes and RAM footprint as small as 1000 bytes, the micro-velOSity is API compatible with the company's velOSity, a mid-range real-time kernel, and INTEGRITY, a real-time operating system (RTOS). Green Hills integrates its family of operating systems with its MULTI development tools package.
Ansoft's Simplorer addresses power electronics applications and the interactions between electrical, electromechanical, electrothermal, electrohydraulic, and mechanical loads in these systems. Together, with its Maxwell finite-element-based electromagnetic field simulation software, the two tools provide a high level of accuracy in automotive design. Maxwell calculates inductance, resistance, torque, force, and other parameters in sensors, actuators and motors and generates a behavioral model of the device that includes the calculated parameters for use in Simplorer. Simplorer can be used as a circuit and block diagram simulator as well as state machine and VHDL-AMS design and is ideally suited for development in advanced hybrid vehicles according to Mark Ravenstahl, product marketing manager, Ansoft.
Since measurements with simulation reduce the number of tests, National Instruments wants users to make the most of the tests that are performed. Instead of just determining if a test passes or fails and then archiving the data, measurement and simulation allows correlation with models — the tools and equations on the design side — to improve the quality. To perform a higher percent of the design phase in simulation, the models have to get increasingly better so testing and correlation with real world measurement is essential. In mechanical design, NI has been working with mechanical design companies to allow this process to be easy and trackable to take measurements of a physical system and correlate these in the mechanical model. For the mechanical device, this can ensure the models and the boundary conditions are accurate.
Walden C. Rhines, Mentor Graphics presentation, http://www.mentor.com/
Ove Josefsson (Volvo Car Corporation), “AUTOSAR Overview (AUTomotive Open System Architecture), Snart Conference, Skövde, 2005-08-17.
Thorsten Gerke (Synopsys GmbH) and Carsten Petsch (Volkswagen AG), “Analysis of Vehicle Power Supply Systems Using System Simulation,” SAE 06M-98.
Liang Shao (Hitachi America Ltd.) and Jian Lin (General Motors), “An Electronic Throttle Model with Automatic Parameter Tuning,” SAE 2005-01-1441.
Randy Frank is president of Randy Frank & Associates Ltd., a technical marketing consulting firm based in Scottsdale, AZ. He is an SAE and IEEE Fellow and has been involved in automotive electronics for more than 25 years. He can be reached at [email protected].