The car industry in Taiwan has been focusing on the rear-end manufacturing and assembly, and not much was put in for system design at the front end. However, as part of the car industry is turning the attention to self-sustainable development, the intelligence of cars is now the priority of development. Taiwan’s companies which are good at system integration have had their eyes on this big piece of pie, and started putting in more resources than ever from the chipsets at the front end to the software at the rear for the development of vehicle technology that matches what is coming.

The idea of vehicle automation has been there for years. However, the so-called automation is not having the car to drive on its own. It is that all the internal components and environmental detection capability of a car are connected to the centralized control in order to allow real-time response when something happens and the system to step in automatically for intervention.

Development of Vehicle MCU for Application and Driving

On the other hand, the sensors in the travel control for engine, suspension, tire pressure and others are integrated and regulated by intelligent central control unit, allowing improved safety and fuel consumption and reducing harm to environment. The new environment-friendly power system consisting of hybrid power, lithium batteries and fuel cells requires smarter power and fuel consumption management. The heart to all of these is the MCU.

As the California Clean Air Act (CAA) was born in the 1970s, MCU was first used for the engine control of a car. However when MCU was first introduced, the industry started using the HC08 of Motorola, 8051-based C500 series of Siemens and other similar 8-bit MCUs. Due to the limits of data bus and performance of 8-bit MCUs, the algorithm code had only around 1,000 lines for engine control, and the total capacity was merely 4KB or even less. The exhaust of engine could not even comply with the 2nd-stage EU standard, and the control object was merely the single engines.

With the increased demands for energy saving and environmental protection, it is a basic requirement to comply with the 4th or even 5th stage EU exhaust standard. The poor performance of traditional 8-bit MCUs forces themselves out of the engine control system. Today, the 16-bit MCUs become the mainstream of engine control, and their 32-bit counterparts are going to put on a good show in hybrid power and battery engine control.

It is difficult to find the way into the car MCU market; those left behind have no choice but trying to squeeze their way in.

Despite the great opportunity in vehicle ICs, the winners of the market always took it all in the old days. It is simply a mission impossible for newcomers to get into the supply chain that connects the car manufacturers and their suppliers in a short period of time. Even a veteran IC designer like Holtek has to find a way to work with Japanese component suppliers in order to get the ticket into the market.

Therefore, in addition to build up their own capabilities and look for any possible way to get in the market, the IC designers may start from the after-market. The products in this market include M/E equipments, dash boards, airbag deployment sensor modules, front and rear lights, electric rear-view mirrors, electric chairs, on-board TVs, on-board A/V equipments, parking sensors, anti-theft protection, navigation systems and other peripheral equipments. They need to work hard in the new markets. However, it is necessary to avoid getting caught up in specific markets (such as China) in order to stay away from non-economic factors and protect themselves from damage to competitiveness.

The acceleration of introduction schedule is the next challenge for vehicle MCUs

For car systems, the MCUs have been there for years. First, they were used in engine injection systems. The ECU used is a type of MCU. As car are more and more electronic, MCUs are everywhere in a car from bumper to bumper. However, the environment on board of a car is a very harsh and punishing to electronic systems. There are inevitably excessive vibrations, temperature, humidity and collisions, and any possible challenging scenario has to be taken into account for a car.

For vehicle MCU elements, the tolerance has to reach the level of 0.1-1% as in an aircraft, and 25G is the design requirement for the strength and resistance against vibration similar to that of an aircraft as well, not to mention that the fluctuation of power supply is 5 times as much as that for an aircraft. Apart from satisfaction with these mind-blowing specifications, a unit like this must not be as expensive as those used in an aircraft. Therefore, from the launch to the introduction of an MCU product into a car, it requires years of time-consuming tests and verifications. The MCUs that are allowed to install in a car are mostly stable products that have survived extended testing and verification rather than the latest products just launched. This is the reason why the development of vehicle MCU technology is not as fast as that of general consumer electronics.

The traditional car industry is so keeping to themselves and conservative that it makes car manufacturers very difficult to accept products equipped with highly advanced architecture. Nevertheless, as the cars are more electronic, the built-in applications are more complicated, and the on-board HMI is getting closer to human, the progress in car systems demands more and more calculation capability from MCUs. And because of so, the functions of car electronics are not going to catch up with the demands of the time if everything still follows the old introduction schedule. Therefore, the cooperation between car manufacturers and IC designers must start from the very beginning of concept design in order to build a product that satisfies the demands for the environment in which cars are used and the performance asked for a car, and only by doing so, it is possible to accelerate the introduction of new technology to build a car that fits the market demands.

MCU performance at different levels for the architecture of smart travel control system

The high-performance 32-bit products are getting popular in car electronics, but it does not mean that the 32-bit gadgets will replace the existing 8 or 16-bit ones. On the contrary, in high-end luxury cars, there are plenty of products that have different performance. For example, there are many subsystems in a car, such as power, suspension, pressure, vision, and sensors. The designers want them as simple as possible. A subsystem that generates simple signals may have a cheap 8-bit MCU in it. As the subsystems become complicated, that’s where 16-bit MCUs come to play. For image transmission which requires a lot of bandwidth and performance, 32-bit MCUs are capable of dealing with the image signals.

The subsystems are there only to feed the signals to the central control, and therefore, the simpler they are the more reliable they can be. In order to process the signals from all the subsystems, high-performance 32-bit MCUs can be used in the central control. There are so many to choose from single core to multi-core units depending on how complicated the system is. Some MCUs even have DSP processor attached or dedicated hardware wiring for large quantity of logic algorithms and the processing of all information. Finally, the processed signals are transmitted to the central control of the car, and the car will identify what to do with the signals based on the programming or the driver will make the decision.

32-bit MCUs will become the core of car system

The introduction of MCUs at different level is one way to make up the intelligent control system of a car. For the control of costs and maximization of performance, different level of MCUs in charge of different work loads is the most reasonable design. Thanks to the concept of intelligent central control, all information is collected and processed here for decision making. The core has to be powerful, or the system will be overloaded when all the data come in, let along making any decision.

The 32-bit MCUs are not just found in the central control among all car electronics. Let’s take the traditional fuel system as an example. The accuracy of fuel injection is now more and more demanding for injection engines. The sampling and processing speed of an 8-bit MCU are no longer satisfactory to the modern environment regulations. Also, the next generation hybrid engines or even battery-powered engines have high demands for voltage control and recharging control. When the driver steps on the gas pedal, as the electronic system is used instead of the traditional hydraulic control, the power feedback, linear power output and other functions require extremely precise communications between systems in order to provide plenty of power and ensure the life of batteries.

On top of that, the batteries of hybrid power car or battery-powered cars have to power not only the engine but also the surrounding sensors and other subsystems. Functions like the interactions and feedbacks between subsystems, system startup and shutdown control and many other applications require the support of high-performance architecture to make things happen.

There is no longer gray area for MCU architecture

Looking at the history of MCU development, there is nothing else that is as successful as the 8051. Today, the 8051 is THE standard that everybody follows. However, this architecture is qualified for a museum. Despite the efforts of IC designers to put in their own stuff for improvement of performance, function and many others, it is not going to change the fact that the legendary 8051 is an 8-bit processor.

In fact, 16-bit architecture has never been really in fashion, although the industry says the 16-bit architecture has its position. However, I have to say, its position is nothing but ambiguity. If a developer has doubt for 8-bit MCUs and second thought about their 32-bit counterpart, we should expect that IC designers come out and shout: “the 16-bit MCUs are the solution!” But in fact, this position is hardly noticeable as the performance of 8-bit MCUs keeps growing and the price of 32-bit MCUs is dropping.

For the future of MCU market, we can definitely say it’s either 8-bit or 32-bit! There will be no gray area in between!

ARM joined the MCU battle and shocked the industry

The suppliers that jumped into the32-bit MCU market at the beginning, such as Renesas, Freescale, and STMicroelectronics, started mostly with the architecture that consists of Power Architecture. Renesas adopted the RISC architecture of SH. These suppliers have control of their own markets with these architectures. For example, Renesas rules the Japanese market, while Freescale focuses on European and American markets. However, as ARM’s Cortex architecture enters the MCU application field, it is now a whole new ball game.

Some of the 32-bit MCU suppliers have claimed simultaneously that their products’ position is specific and they have their own market shares, but these suppliers are extremely sensitive about the ARM architecture, possibly afraid of falling behind the competitors.

ARM architecture is known for mobile applications. This architecture is widely known in the industry for its ease to develop. It seems that the full package of supporting and transitional solutions in terms of hardware and software are provided as the continuation from its persistence in SoC area after entering MCU territory. Is this why IC designers are diving into the ARM architecture? I think not!

In fact, the industry has been looking to establish a hardware standard for everybody to follow, hoping to achieve maximum product reusability and compatibility on the same platform. After all, most of the MCU architectures went their own way, and the programs were not compatible at all. As soon as one architecture is chosen, everything that comes next, including product development and compatibility, has no choice but following its footsteps. The point is none of these architectures has the biggest piece of market. For now, it is still profitable despite that everybody has its own piece of territory, but today where globalization and standardization are inevitable, supporting these architectures popular to only a handful of people is nothing but shutting oneself from the rest of the world.

I believe that the gray area between these architectures will be gone eventually. The market follows the mainstream, and the mainstream creates standards. With an established standard, more will follow. For now apart from ARM, you can count the suppliers that supply built-in processors that are qualified to be the market standards with only the fingers of one hand, such as SH, 68K and MIPS, but none of them is able to create a ripple as big as ARM can.

Because of so, the fact the ARM enters the market is pushing the other MCU architectures out of the market. If clients find that it is easier to develop products with ARM, control the launching schedule and provide a reasonable price (well, ARM is a bit cocky when it comes to licensing, but with a relatively large quantity, the price per chip will not be too ugly), I can’t think of any reason why MCU with ARM architecture will not succeed.

In the foreseeable future, these IC design will still maintain multiple parallel product lines. Just how many products in these lines will survive after the market selection? Let’s wait and see.