eeNews Europe: What does this mean in practice?
Schiefer: We already sell SiC in industrial applications and are in the process of qualifying SiC for automotive applications. Currently, about 15 customers are testing a first SiC module. In my opinion, SiC will not be used on a larger scale in the market until around 2020 at the earliest - first in premium vehicles; in the mass market probably not before 2025.
eeNews Europe: And GaN?
Schiefer: Gallium nitride clearly plays an important role at Infineon. Our CoolGaN product family is already being used in high-performance power supply units for data centers. CoolGaN makes them more cost-efficient and compact and enables them to consume significantly less power. In the field of automotive electronics, we do not yet see any suitable areas of application in which GaN would have an advantage. Among other things, costs are currently an exclusion criterion. It is therefore difficult to say when there will be automotive-qualified, robust GaN components. In any case, the deployment of GaN in vehicles is more a thing of the future than SiC; so the introduction should not be expected before 2020.
eeNews Europe: In the automotive industry, there is talk of getting away from the many ECUs distributed in the car and to provide for several domain computers or even central computers, which then depict these ECUs virtually as software. This development would have a significant impact on the semiconductor content of cars. How do you see this development at Infineon?
Schiefer: With the increase in comfort and functionality as well as in-vehicle communications and automation, the computing power of the vehicle must also increase. This task could either be done by several powerful domain controllers or additional ECUs, of which there are already around 100 in the premium vehicle, for example. Automated driving and IoT functions are expected to bring further ECUs into the vehicle. However, this additional complexity can hardly be handled. It is therefore indeed advantageous to define two to four domains in the car where the domain controllers take over communication and complex computing functions. Such a domain structure can also assume functionalities that are added to the car. Classic ECUs will, continue to be needed; they are responsible for individual real-time tasks such as steering and braking. It is not yet conceivable that, for example, an airbag control unit could be implemented as a virtual ECU in software. Most of the ECUs, of which an average of around 60 currently communicate with each other in a medium-sized car, will continue to be implemented as classic ECUs in the car.
eeNews Europe: So there is no hope of simplifying the electronics architecture in the car?
Schiefer: I would put it this way: the complexity continues to increase. The domain structure will help to manage this additional complexity. A complete simplification seems difficult to be achieved. With or without simplification of the electronics architecture, however, one thing is quite clear: the car of the future will drive CO2-free, autonomous and data-secure.