As processing systems become more complex, Renesas believes timing silicon holds the key to enhancing SDV bandwidth and latency. By Stewart Burnett
The success of software-defined vehicles (SDVs) arguably depends on fast response times and higher bandwidth. The control functions handled within a SDV continue to become more complicated due to consolidation around single processing units. A system’s responsiveness is crucial for a wide range of features, from infotainment functions like media streaming and navigation to control systems for driver-assist or braking systems. An unexpected lag spike in either of the latter could lead to disastrous consequences for both a vehicle’s occupants and other road users. Two key system elements required to ensure systems remain responsive is bandwidth and latency.
Mastering latency—the delay between the start of data processing and transmission to its destination—is essential for SDVs to function as customers expect. For software to be faster and more responsive, there are two key variables to consider: the clock speed for the SoC/CPU and the speed of interconnection between processing systems. Software development is made easier when latency is reduced and becomes predictable.
Yimu Guo, Timing Product Manager at Renesas, believes that SDVs will benefit from a silicon-based clock system that ensures peak system performance with high processing bandwidth and low latency. The clocking solution can determine several key factors including power, speed, and cost. So, how might timing silicon help achieve optimal latency speeds, and how serious could a failure to prioritise this aspect of SDV design be for automakers?
Shrinking the quartz footprint
Timing is not a new problem for automotive, but it is becoming more complicated as vehicle functionality and hardware complexity increases. The compute-intensive nature of SDVs will require interconnect speeds an order of magnitude faster than incumbent solutions can offer. “Today, you’re looking at single-digit gigahertz throughput. However, as SDVs evolve in the future, we could reach data-centre class interconnect speeds,” Guo remarks. Silicon-based solutions are better-equipped to handle high performance largely due to their programmability.
Historically, discrete quartz crystal oscillators have been used for automotive timing, but they are now approaching their limitations. Oscillators can only generate a single output clock frequency, so, as the complexity of SDVs increases, many oscillators are required per vehicle. This results in a larger footprint and can be challenging to integrate, creating unnecessary design hurdles for vehicle engineers to overcome. By contrast, a single silicon-based timing device can be added seamlessly into an SDV control module, allowing for a more compact, integrated system design. “Silicon timing enables the cost optimisation of raw materials, which is a major issue for our customers,” remarks Guo.
Other advantages of silicon for timing applications include the ability to add purpose-defined functions using standard, scalable semiconductor manufacturing processes. Silicon-timing generators can be programmed to provide a wide range of output frequencies, whereas distinct quartz devices are limited to a single fixed frequency. This limitation will become increasingly pronounced as SDV development advances. “Purpose-defined functions can include diagnostic and monitoring features to boost the overall reliability and safety in the control unit, improving the overall driver experience,” Guo explains.
However, one of the reasons quartz has endured as a material in automotive is the simple nature of the ECU function. Historically, a low performing MCU with a single clock would perform its task in isolation. With the onset of SDVs, a single processing system must communicate and oversee all these functions. This invariably increases system complexity, resulting in higher speed clocks. The increase in higher frequency clocks would traditionally require many expensive and discrete quartz clocks. However, this scenario is more ideally suited to a silicon timing solution. One single silicon timing device can provide all the disparate clock frequencies within the design while also adding some digital functions for diagnostics and monitoring.
Optimal frequency selection
While Renesas and its peers in the timing segment continue to develop silicon-based solutions, Guo emphasises that the trade-off is already worthwhile. “Quartz oscillators are a known quantity and arguably the path of least resistance, but at some point you will need a clock tree that enables flexible frequency selection to meet the range of processing and connectivity features.” This is not an approach familiar to vehicle control unit designs, although it has existed in data centres for decades. To this end, Renesas offers a range of solutions that can generate different clock frequency outputs and direct them towards the appropriate processing and connectivity devices.
Guo highlights the growing perception that cars are becoming “data centres on wheels” but notes a key difference: the consequences of hardware failure in a car are exponentially more severe. “The quality requirements are much higher, and meeting them can be difficult. You need to make sure that your part is fully reliable while running inside the car.” Should an electronic braking system experience a communication issue due to a timing failure, for example, the results could be catastrophic. Fast and reliable response times are also crucial in the context of autonomous and driver-assist systems, which process vast amounts of data continuously from a variety of sensors, including LiDAR, radar and cameras. Should latency problems affect the continuous and timely transmission of this data, it could lead to a delayed reaction or failure to perceive an object in the car’s environment entirely.
Future-proofing
Guo believes environmental data transmission demands better quality clocking with lower noise levels, improving bit error rate alongside latency and bandwidth. One of SDVs’ primary benefits is their allowance for continuous iteration and feature set expansion through over-the-air updates—for example, upgrading from SAE Level 2 to Level 3. Such software advancements using common existing hardware will only place further demand on them, paving the way for more configurable solutions to step in.
Silicon timing enables the cost optimisation of raw materials, which is a major issue for our customers
While operational reliability is crucial, Guo suggests future-proofing an SDV with timing solutions that can support upcoming standards. This will help ensure the vehicle hardware remains responsive even as its software becomes increasingly complex. He highlights a feature contained within Renesas’ Autoclock product family that allows for an increase in system reliability by leveraging redundant input clocks. “If the system reference clock stops working or fails to be within the specification, Autoclock will automatically switch to the back-up if necessary.”
Ultimately, Guo believes the relevance of silicon timing will only increase as the automotive industry moves closer to realising SDVs. However, if automakers prioritise clocking during the development process, then they must ensure their vehicle solutions are as responsive in ten years’ time as they are at launch. “Hardware requirements are going up, and so is system complexity, even at the subsystem level. You’re going to need more and better clocks as a result,” he concludes. “As time wears on, the importance and usefulness of silicon timing compared to discrete, quartz-based timing will be realised.”
