The automotive industry is currently experiencing one of the greatest breakthroughs in its history – not only where electromobility is concerned, but also in terms of electric/electronic architectures (E/E). When looking at the growing trend towards SDVs (“Software-Defined Vehicles” – vehicles with characteristics and functions that are essentially controlled by software, i.e. a software-centric electronic device on wheels), you cannot help but notice the huge difference compared to today’s vehicles that are equipped with up to 100 control units, making an “over the air” software update virtually impossible. A central computer structure with interfaces to all sensors and actuators does not appear feasible under these conditions.
Self-driving cars require a huge number of cables with considerable lengths to be laid throughout the entire vehicle. Therefore, it will take several intermediate steps to transition to the SDV, where domain controllers in connection with zone controllers will find their way into the vehicle. This will be supported by the increasing introduction of Ethernet technology, which is being further developed in the multi-gigabit direction as specified in the IEEE802.3ch, IEEE802.3cy and IEEE802.3cz committees. Thanks to this technology, it will be possible, in future architectures, to reduce the number of plug-in connectors for self-driving vehicles to a strict minimum. Despite these steps, connecting the sensors and actuators to the powerful zone controllers will be a huge challenge. This is because over 100 quick plug-in connections to control devices are required in the vehicle. In addition, connecting the right slot on the control unit to the right cable is an enormously complex process for employees on the OEM’s assembly line. Another challenge is finding the necessary space on the control units for such a large number of slots, as the installation space for connecting the cables is limited. Therefore, new solutions are required that will reduce the number of plug-in connections while at the same time enabling maximum flexibility so that all of this can be achieved within the OEMs’ final vehicle assembly cycle times, and thus in an economically viable way. Therefore, flexible, integrated plug-in connectors need to be developed which make optimum use of the installation space available and significantly reduce the number of plugging processes.
Complex onboard networks: too many control units, too many connections
A closer look at today’s vehicles shows that between 10 and 25 high-speed plug-in connectors are integrated into the control devices. These are distributed around the vehicle and connected to sensors and actuators via assembled cables. In addition, there are countless interfaces with sensors and actuators that have low data rates and supply the control units with the necessary signals. Through the increasing number of functions which require high-speed connections, such as infotainment, ADAS and self-driving functions, the number of control units in vehicles is increasing dramatically. This is leading to a significantly higher need for data cables and is making it considerably harder to produce compatible software interfaces. In addition, in recent years, the component market has developed in such a way that different plug-in systems are available for data connections with high data rates, which are incompatible with each other. This is leading to a strong diversification in technical solutions.
Development of onboard network architectures in the vehicle
In order to understand why there is such a high number of applications, we first need to take a look at the architectures that are in use today or will be used in the vehicles of the future to connect and control the many sensors and actuators.
1) Distributed architecture
Distributed architecture has evolved over time and generally involves adding one separate small control unit for every sensor and actuator. With the number of sensors and actuators increasing to over 100, the number of plug-in connections and cables on control units and the head unit is also increasing. The limited space for accommodating the increasing number of plug-in connectors and cables on control units has led to a change in thinking towards more flexible and scalable architectures.
2) Domain architecture
With domain architecture, it is possible to combine all sensors and actuators in one application group. The functions of what were previously many individual control units are mapped by software in the domain controller. The different application groups are each mapped with one domain controller. The control units have slots for the respective applications (sensors and actuators) and a connection to the main control unit in each case. In this system, applications such as ADAS can be integrated as an independent functional unit. This has resulted in the miniaturization of connectors in order to create sufficient space for the numerous applications. The data rates required between control units are also increasing.
3) Zone architecture
With this architecture, 1 to 2 HPCs (High Performance PCs) are installed into the vehicle which control the increasing number of autonomous functions like a central brain. To achieve this, zone controllers are used to process the data (aggregation) and send it to the HPC which takes over control of the actuators. With increasing automation or autonomy of the driving functions, this HPC is equipped with the necessary redundancy to ensure the required decision-making reliability A modular and flexible approach for the connectors is required in order to fulfil the different architectures of the OEMs.
Automotive Ethernet – reducing complexity
Investigations into different chip and control unit manufacturers show that the number of sensors is going to grow over the next few years. Different analyses and studies show that between 25 and 80 high-speed plug-in connections to the control units will be required per vehicle. This varies widely depending on the architecture and the number of sensors in the vehicle. Considering sensors alone, up to 30 cameras, 20 radar and 10 Lidar sensors are integrated into self-driving vehicles. There are also various infotainment interfaces, as well as Car-to-X, which will play an important role in the future in networking with other vehicles and infrastructures. The challenges include assessing the collected data in real time and sending it to the very high-performance HPCs where decisions about the driving situation are made. These then transmit further orders to the actuators that control the vehicle in order to ensure safe, autonomous driving. This requires a wide range of high-performance plug-in connectors and cable technologies to be made available and integrated into the control units. As different interfaces have been established by the different OEMs, the greatest degree of flexibility is required in order to meet the varied customer requirements. On the other hand, the aim of car manufacturers is to automate the installation of cable sets into the vehicle and to reduce the number of cables. This poses major challenges for cable assemblers as they have to integrate the high-speed cables into cable harnesses in an automated manner. This integration is not possible in one step, which is why attempts are being made to standardize the basic architecture with Automotive Ethernet via a sheathed two-wire system. Ethernet makes it possible to use zone controllers which are equipped with specific “intelligence” and enable the cable lengths of the cable harness to be reduced and simplified. At the same time, modularization of the plug-in connectors at the controllers is necessary so that the controllers can be flexibly adapted to the customer’s requirements.
Multi-connectors for flexible adaptation to the onboard network architecture
To solve this dilemma, the integration of high-speed plug-in connectors into multi-pin headers is being investigated. This involves pressing multi-connectors into individual contacts by means of a special “stitching process”. The intended solution aims to insert high-speed plug-in connectors into these headers (circuit board connectors) and attach them using a similar process. Two approaches will be pursued depending on the customer’s requirements.
The first option is to facilitate the integration of existing plug-in connector interfaces (e.g. H-MTD, MATE-AX, HFM etc.) into a control unit header so that they can be positioned on the control unit and applied in one step. The advantage of this is a simplified manufacturing process for Tier1 suppliers.
The second option is a completely new approach. It consists of a construction kit which can be equipped with different interfaces according to the customer’s requirements. The basic pre-requisite is a standardized modular dimension for modules which can be equipped with power and signal contacts as well as with high-speed plug-in connectors. This system operates independently from the original interface with modules. For Tier1 suppliers and OEMs, this option results in high flexibility in adapting their control units to the architecture’s requirements.