“Connecting the analog world with the digital”
eeNews Europe: How important is the automotive industry as a customer market for Analog Devices (ADI)?
Stefan Steyerl: Automotive has a high share of ADI’s global sales: In the last fiscal year it was 16%. Europe has traditionally been a strong automotive region, with a much larger share of business here. In addition, there is the trend that the value share of semiconductors in cars is rising sharply. This was already the case in the past because of the various electronic systems in the vehicles; now it is being driven additionally by electrification, Advanced Driver Assistance Systems (ADAS) and the automation of driving. Today, an average value of about $250 per car is assumed for semiconductors; with fully electrified vehicles at level 4 or 5 of automated driving, this content will more than quadruple.
eeNews Europe: Are there any development programs in the automotive industry that involve ADI? I’m thinking, for example, of Audi, which has a joint development program with several semiconductor manufacturers?
Steyerl: Exactly, Audi has the PSCP program (Progressive Semiconductor Program). ADI has been one of the partners since this program was introduced. As a chip manufacturer, such a program enables us to identify at an early stage which requirements OEMs will have on future electronic systems and, along with that, on semiconductor technology as a whole. This allows us to integrate these requirements into our development flow at an early stage so that we can have these components, solutions and systems available in time for the start of series production of the respective vehicles.
eeNews Europe: Do other OEMs also have such or similar programs in place? Perhaps also on an international scale?
Steyerl: I don’t know whether other OEMs with whom we work have a similarly structured and open R&D program as Audi. But of course all OEMs have their semiconductor strategies and have a similar approach to working with Tier 2 semiconductor manufacturers early in the development. Conversely, it is also our aim to identify corresponding development trends at an early stage. We do this with all relevant automotive manufacturers.
eeNews Europe: In which application areas does ADI focus?
Steyerl: Our focus as a semiconductor manufacturer is to connect the analog world with the digital world. Starting with the sensing, the acquisition of physical quantities, up to the provision of the corresponding processing capabilities for the resulting data. This general strategy of ADI can also be found in the automotive sector. We were the first semiconductor manufacturer in 1993 to have a MEMS acceleration sensor for airbag control in series production at vehicle manufacturers. We have continued this tradition of development from driver assistance systems to sensors for automated driving. There is an increasing need for sensors to understand what is happening around the vehicle. Several different technologies are relevant. First of all, radar, on which we have a very strong focus, and then lidar technology, in which we also have considerable activities. We do not sell cameras and camera sensors for automated driving, but our network technologies enable us to connect such cameras. We are also represented in simple driver assistance systems with reversing cameras via our video interfaces and codec modules.
These are our focal points in the area of safety. In addition, we focus on powertrains, especially electrified powertrains. Our focus here is on battery management systems. Years ago, we entered the traditional battery management system business, i.e. the management of 12V lead acid systems. Here we now have third-generation products on the market; we integrated everything on a chip that is necessary to measure such a battery. Such miniaturized systems sit on the terminal of the battery and record its charge state, voltage, currents and temperature. These systems are necessary for modern start-stop systems in micro hybrids. Today, we are the market leader in this segment and cooperate with Tier 1 suppliers, who in turn are the market leaders in their field. We have a market share of over 80%. This is a very interesting market, because start-stop systems have started in Europe, but can now be found in vehicles worldwide.
The acquisition of Linear Technology has also given us a strong focus on higher-voltage battery management systems. This ranges from 48 volts to 800 V traction batteries. It enables us to map cell monitoring in lithium-ion batteries and also interfacing them with higher-level systems. This will make the electrification of vehicles, especially measurements on lithium-ion batteries, a further focal point. In addition, these high-voltage batteries must be decoupled from the other electronic systems. We have solutions for this with our Isopower family. We are also active in the field of insulation technologies for gate drivers based on ADI I-couplers. These have been found in such vehicles for many years.
eeNews Europe: Will the voltage in the traction batteries continue to increase? 800 volts is already a lot, but for higher engine power, thicker cables would be needed.
Steyerl: I don’t think the development will lead to even higher engine power. Power is actually more than sufficient today. The bigger challenge lies in accelerating the charging of the battery systems on the one hand and reducing the costs for the battery systems on the other. In today’s EVs, battery cost is a crucial factor. The aim is to reduce cost on the one hand and at the same time to increase range. These two developments go hand in hand, because the vehicle developer can install more battery cells – but then the costs continue to rise. In view of these conflicting requirements, we are considering what contribution we can make on the electronics side to measuring the cells. If I can measure more accurately, I can make better use of the remaining range. This will indirectly reduce battery costs. Another important aspect: If I measure cell voltages with greater accuracy during charging and discharging, I can also increase battery life – and this is an important factor in electromobility, especially in terms of costs.
eeNews Europe: Another product area where ADI is active, is the Automotive Audio Bus or A2B. What is the situation here – are there new customers, new applications?
Steyerl: A2B is an example of a technology that we have developed in close collaboration with car manufacturers. The problem was originally that there were multiple audio sources in the vehicle, for example several microphones. The manufacturers had to connect each one individually and wire it to the head unit. The result was star-shaped wiring that increased both weight and cost. And it led to the problem that there was not enough space at the head unit for the numerous connectors. Out of this problem we developed the A2B bus. With its daisy chaining, the A2B addresses precisely these issues, the cabling no longer is star-shaped but ring-shaped, and the head unit only needs one connector for multiple signal sources. This technology thus addresses the issues of costs, space and weight. A2B can be found today in production vehicles …
eeNews Europe: Can you give examples?
Steyerl: The first OEM to bring this technology into series production was Ford. Today we have the A2B technology in many vehicles in Europe as well as worldwide; unfortunately we are not allowed to name them. Many other manufacturers are already planning with A2B as a technology for future model series.
eeNews Europe: Is A2B replacing the MOST bus?
Steyerl: That always depends on the specific application. There is certainly also a trend towards replacing MOST with other systems such as Ethernet. The focus of A2B is the audio sector, i.e. primarily the connection of microphones and loudspeakers.
eeNews Europe: What will we see next in the area of automated driving?
Steyerl: There are many different opinions, which are also subject to change. A year ago, the time horizon for the introduction of automatic vehicles at Level 4 / 5 was still seen within five or six years. From today’s point of view and taking into account the statutory approval procedures, I believe that fully automated driving will hardly reach the mass market before 2030. One has to distinguish very much between the different levels of automated driving. The same applies to the different submarkets: we have to distinguish between the market for private users and the fleet operators, whose robot taxis represent very limited use cases. Such autonomous vehicles are likely to exist earlier in certain narrowly defined application segments. OEMs and suppliers can of course use the development to learn from it. Nevertheless, the volume will remain very limited in the near future. What we will see earlier are vehicles according to autonomy levels 2+ and 3. The rules here are relatively unclear. Currently the industry has gone back a bit and talks about “level 2 plus”. This level actually already has the technology for L3, but leaves the responsibility completely with the driver. That’s why this should be considered as assisted rather than automated driving.
eeNews Europe: Do you see any demand for automated driving functions in the truck sector?
Steyerl: Yes, absolutely. Automation here is partly driven by safety considerations, partly by the pursuit of higher productivity.
eeNews Europe: There are reports that the first robotic vehicles will be trucks rather than cars.
Steyerl: Yes, it’s fair to say that. But it always depends on the autonomy level. I don’t think there will be big trucks in level 5 very soon, but quite likely we will see delivery vehicles with limited use cases, which automate certain tasks within a narrow area and with a manageable fleet.
eeNews Europe: Are there any developments in the field of automated driving where ADI not only sets the tone, but perhaps even sets the direction?
Steyerl: We have our radar chipset, and radar is one of the most important technologies to map partially or fully automated driving. According to current knowledge, three technologies – radar, lidar and camera systems – are needed to detect the environment. In radar systems, we are developing radar sensors based on 28nm CMOS technology. The 28nm RF CMOS technology is interesting because it represents the best process node on the cost side and at the same time it allows a higher degree of integration compared to the SiGe technology frequently used in radar technology. In other words, RF-CMOS makes much more possible, from the transmit-receive side to circuitry and signal conversion to the digital world, even processing data in a single-chip radar.
But nothing goes without challenges: RFCMOS makes it more difficult to build high performance systems with sufficient output power to look far into the distance and thus push the phase noise to an acceptable level and finally to achieve the desired high resolution – all this is generally more difficult to implement in RFCMOS than in SiGe technology. When we decided to switch to RFCMOS, we decided to achieve at least the performance of SiGe from the beginning. With our current 28nm RFCMOS chipsets we have achieved this. This enables us to map the high performance systems described above onto an RFCMOS chipset. With this technology, it is easier to detect a child who suddenly steps onto the road between two vehicles – the higher accuracy helps. And the radar sensor can look further into the distance and detect obstacles earlier. For example, obstacles lying on the road – for automated driving it is also important to know how high such obstacles are. At heights of up to 7 cm, the vehicle simply drives over them; at higher obstacles, the control systems have to brake or drive around them. The sooner you recognize something like this, the more time the automatic has to react.
eeNews Europe: Do such systems also detect obstacles that appear a little higher above the road, such as the famous crosswise standing truck?
Steyerl: That depends on how the radar system is set up and whether I also measure the height via the number of channels and not just in the horizontal. We can also implement such systems with our technology.