EVs are coming
For a few years now, we have been saying that “EVs are coming”, but now we can say that “EVs are here!”. Driven by government legislation, environmental concerns and latterly by the rising cost of fossil fuels, the market adoption of the electrified (pure battery EV or hybrid) vehicle is on a steep ramp. Analysts predict that by 2030, 50% of the vehicles sold globally will be electrified.
One factor that will decide whether that prediction holds true or not is the cost of the EVs – at the moment, they are still too expensive, and the adoption is assisted by government subsidies. Economies of scale bring cost reduction, but other approaches are also needed. The introduction of the new domain and zonal architectures, particularly for new pure battery EV designs, will be a significant contribution to this effort.
Cars needs zone architectures
Over the past few decades of vehicle development, as more and more mechanical or electro-mechanical modules were added, the interaction of these modules drove the need to create a vehicle network. While the modules enable great new vehicle features, the organic growth of the network creates an architecture that is not well-suited to the complex, multi-functional nature of a car. Software and hardware architectures are getting too complex to develop and maintain through the life of a vehicle platform which extends to ~15 years.
Vehicle manufacturers are adopting two approaches to help
- Domain focus – consolidating logical functions and connecting ECUs, which combine to provide a vehicle function (e.g. body, propulsion). The consolidation assists the software architecture, via a hierarchical structure of the function within the domain, and abstracting that function from the rest of the vehicle.
- Zonal focus – to reduce wiring costs, this initiative connects nodes that are physically close together to a local zone gateway, independent of what function they provide. These zone gateways are then connected to each other and to the central computing node, by fast Ethernet.
These two approaches are orthogonal and complementary, they can co-exist on the same vehicle to achieve the benefits of both.
Data vs power distribution
The zonal vs domain architecture discussion typically focusses on data – how the functionality is distributed across the many ECUs in the vehicle and how the data is transferred between them.
However, we cannot forget about the distribution and transfer of power, certainly we do not want to power up all zones all the time.
Imagine the scenario where your car is parked outside your home for two weeks when you are on holiday. Of course your alarm is running, but occasionally this will “wakeup” various functions to assess the state of the vehicle. However, it needs to do this without activating the full network. Scenarios like this lead to a complex arrangement of wakeup and conditional powering of every node.
There are 2 general themes in these schemes:
- Central distribution – from central fuse box (possibly with additional satellite distribution boxes), connecting all individual nodes that need power, given the current condition of the vehicle (driver approaching, driving, parked, etcetera).
- Power distribution is a function of the zonal gateway. The ECU decides what power to apply to each of the edge nodes below it, depending on the current context of the vehicle.
It should be noted that to make power distribution cost effective, the older fuses and relays are replaced by eSwitches and eFuses.
Propulsion domain control
The area of propulsion is probably going through the most significant change. When the combustion engine was the only source of torque in the vehicle, control was much easier – the main challenge was how to comply with the various global emission standards.
However, adding an alternate torque source (e.g., an electric motor) brings new control challenges and new failure modes to consider. This is where the propulsion domain control function becomes critical – this is a strategy function that sits above the control of each of the various individual sub-functions – like combustion engine, electric motor, battery, DC/DC converter, on-board charger (OBC), chassis control, braking, etcetera. The domain controller makes the strategic decisions, based on the current context of the vehicle – decisions like which torque source to use (considering state of battery, terrain, traffic, etcetera), and the contrast, which anti-torque to use (fraction brakes or re-generating electric motor). Strategic decision making can be made for reasons of safety, but also for reasons of energy optimization – to maximize the range from the finite energy source on the vehicle; the battery. Abstracting strategy from real time control eases the development and manufacturability of the vehicle – changes in vehicle configuration, motor type, battery size, etcetera, are handled by the domain control function.
This domain controller can either be instantiated as a dedicated ECU or can be a hybrid zone/domain controller taking on the functions of for example the front left zone as well as propulsion domain, or housed as part of the central compute node looking forward in time.
S32 Real Time Processors
NXP has just extended the S32 Automotive Platform with a new class of safe, real-time processors that offer the critical deterministic behavior of safe microcontrollers (MCUs), but with an unparalleled combination of gigahertz speed, multi-application isolation and memory expansion capabilities. The new 16nm S32Z and S32E real-time processor families are ideal for the safe integration of cross-domain vehicle functions, like the propulsion domain function. Together, these devices support our customers’ diverse, end-to-end domain and zonal vehicle architectures to enable the software-defined vehicles of the future.
Real-time applications, like propulsion, operate consistently and predictably, so they are “deterministic”, completing execution within a defined maximum time. To meet these timing constraints, a processor core must meet the performance requirement, offer fast interrupt responses, and provide deterministic execution, which maps well to the Arm® Cortex®-R52 processor cores that are used in the S32Z and S32E families of processors. The Cortex-R52 processor cores also support functional safety and hypervisors which are needed for these safe, real-time applications.
The initial S32Z2 and S32E2 series that are sampling today include eight Cortex-R52 processor cores with split-lock support that operate at up to 1 GHz.
In the future, NXP will sample the S32Z1 series, which includes four Cortex-R52 processor cores for applications that require more performance than traditional automotive microcontrollers for real-time applications integration, but not as much as the S32Z2 series. This gives customers scalability with two families of processors that are pin- and software-compatible. The real-time processors will also scale into the future with even more performance and integration, with 5nm products in the future to offer a strong roadmap with software compatibility.
The S32Z2 series is focused on domain and zone control, but the S32E2 series add some interesting real time actuation capability, primarily for EV functions like inverter control, DCDC, etc. This enables the up-integration of these functions, allowing them to be controlled directly by the propulsion domain controller, thereby reducing BOM cost, size and weight – critical factors in the push for EV adoption.
Conclusion
Vehicle electrical architectures are going through a transformation. As the software content and complexity increases, to create a software-defined vehicle, the vehicle network is evolving into domain and zonal architectures. This reduces the cost of the associated wiring harnesses, but there are many ways to implement it, with the associated distribution of processing, data and power throughout the vehicle. A zonal architecture brings its own challenges, such as how to partition safety critical real time control operations, like propulsion. NXP has introduced a new class of real time processor which can be used in zonal, domain and vehicle control applications, to help make this transformation realizable.