Making Cars Better

Electronics can improve everything from gas mileage to the user experience, but they have to be reliable enough under extreme conditions and power efficient.

popularity

The automotive industry, with its double-digit growth, is a very attractive market for equipment manufacturers. This growth is explained not only by the increasing number of cars produced for the Asia market, but also by the shift of basic customer expectations for things such as more hybrid and electrical vehicles, more sophisticated infotainment requirements, and more high-end features.

On top of volume (more cars), there is a need for new technologies. We are getting further and further away from our grandfather’s car. For instance, premium vehicles operate with 100 million+ lines of code. This code ensures safety and security by controlling thousands of embedded electronic devices and sensors. Actually, electronics systems contribute to most of automotive innovations and new features. The so-called “intelligent” cars, with semi-autonomous driving features (self-parking, advanced cruise control, collision-avoidance, etc.) rely on 32-bit MCUs. The complexity and number of these automotive MCU increases year after year. A decade ago, the associated cost for electronics and software content used to be below 20% of the total cost of a vehicle. Today the cost can reach up to 35%. Consequently, cost is a serious constraint for the electronic suppliers. Somehow, this cost constraint is driving the race for electronic innovation.

One way to control cost is the “all-in-one” approach. Instead of using multiple microcontroller units with each controlling a single function, chip suppliers are integrating multiple functions into a single microcontroller. By reducing the number of chips, reliability improves and integration gets easier and cheaper for the automotive manufacturers. This is the case of NXP with their chip for infotainment applications. The engineers have managed to combine up to six different integrated circuits (ICs) in one single IC. With this single product, they manage to address the worldwide range of digital radio frequencies (U.S., Europe and Asia radio bandwidths). Such a strategy is very important to be successful in the automotive electronic market, and not just for infotainment. A single MCU needs to take into account the worldwide constraints, such as the different standards from one country to another. This is the way to ease the productivity challenges (i.e., cost) for car manufacturers.

Standards are another consideration. The automotive electronics industry typically involves many standards for security and safety purposes. There are standards for electrostatic discharge protections (ESD – IEC_61000-2-2), for electromagnetic compatibility between devices (EMI/EMC – IEC_6000-4-4), and for functional system safety (ISO 26262). This last one actually becomes a “must,” especially with automatic driving control features.

These standards and constraints are mandatory for OEMs. After all, it could be a matter of human life. As a result, they will directly impact and drive the physical integrity constraints. For example, some electronic controller units should have a lifetime of 15+ years with a power consumption of 10 watts—both of these in very stressful environments (vibration, dust) with severe temperature conditions (up to 150C°). And they should survive to 15KV ESD discharges on a 28nm technology. This is not trivial!

Reducing the cost, increasing the reliability, and maximizing the performances at the same time is a major challenge for the automotive IC designers. Power consumption is one of the key criteria for MCU performance. A few years ago, the main power constraint was mainly the “off” state leakage, when the car is turned off. Today, the automotive chip power specifications are more sophisticated and involve real management. The power has to be optimized in different operational modes: Off, Standby, On, Full Speed, and so on. But power is not free, and it impacts the performance of the car itself. This makes perfect sense not only for electric vehicles’ autonomy, but also for combustion-engine automobiles. Based on some estimates, 100 watts corresponds to a consumption of 0.1L/100Km (≃0.04 gal/100mil). Considering the number of chips within the car and the fuel economy regulations, there is no place for wasted power. Therefore, there is no other choice but for designers to simulate this power usage as early as possible (RTL level), in every possible use case, to better control and optimize the final chip performances.

Then there’s chip reliability, which relies on the quality of the physical implementation. The task is not trivial for designers, and depends upon the final requested MCU architecture (see figure1). Integration between sensitive analog IPs, noisy digital IPs, memories (ROM, RAM, FLASH), I/O interfaces, voltage regulators, power switches, protection circuits (e.g., ESD) and different set of sensors (e.g., thermal) is a major challenge for designers. Simulation is the only choice designers have to ensure integrity and reliability of the final chip to perform as desired in the automobile (especially from the prototyping to the tape-out phase).

ansysphoto
Figure 1: Illustration of MCU Architecture

An effective solution calls for the ability to mix different domains with different and lots of data. The environment must solve a number of questions. How to take advantage of RTL power signature for a gate-level simulation? How to model transistors’ behavior inside an analog IP during a top-level digital simulation? How to include package and board models for a chip level power-thermal interaction simulation?

The ideal environment goes beyond chip-level simulation, allowing interoperability and data exchange between different teams: front-end, analog, top-level digital, package, board and system level engineers. This brings up the need for accurate modeling and use of those models for system-aware chip and chip-aware system simulations. This environment needs to enable, both at chip or system level, power integrity, reliability (ESD and electromigration), thermal, and especially in case of automotive applications, EMI/EMC analyses. For safety reasons, to validate electromagnetic compatibility between electronic devices, the MCU’s power signature and the body of car itself have to be modeled and simulated.

Simulation used to be popular to increase the productivity in the automotive industry— cars’ shapes are simulated to optimize aerodynamics, fuel engines are simulated to maximize performance, and many mechanical parts are simulated to minimize content with optimum reliability. Now simulation is also becoming key for automotive electronics and associated software. It is a productive answer to the automotive industry dilemma: tradeoffs between cost, safety and performances. Simulation is enabling innovation.