EIPC Technical Snapshot: Supporting Autonomous Driving

Estimated reading time: 5 minutes

EIPC’s 17th Technical Snapshot webinar on May 4 focused on developments in automotive electronics, particularly on advances in the technologies required to support the evolution of autonomous driving. The team brought together two expert speakers to present their detailed views on topics encompassed within “CASE,” the acronym that appears to be taking over the automotive industry.

The session was moderated by EIPC president Alun Morgan who welcomed participants and introduced the first speaker, Jan-Henryk Serzisko, strategic account director at ESI Automotive, with an overview of how ADAS requirements influence PCB electronics.

His presentation gave us an introduction to CASE, an understanding of changes in hardware and vehicle architecture as the technology progresses toward full autonomy, and a summary of changes to the supply chain and its key players.

He explained that the CASE acronym was more than just industry jargon, and had become the guiding principle for the future of the automotive industry, describing trends in connectivity (C), ADAS (A), shared (S) and electrification (E). All aspects of CASE are pushing growth in automotive electronics, and as a consequence software and electronics have become the focus of most automotive companies and their executives. 

ADAS (advanced driver-assistance system) is a fundamental element of CASE. The market for automotive sensors associated with ADAS is growing strongly and is forecast to outpace automotive sales. Serzisko indicated the extent of proliferation of sensors as vehicle autonomy level has progressed from basic driver assistance toward full automation. In terms of vehicle architecture, Level 2 autonomy requires 100 to 150 electronic control units (ECUs), whereas at Level 3 and higher the trend will be toward domain control units. Advanced processing will be required, and automotive safety and autonomy will rely on advanced packaging, with thermal management being a significant consideration.

He discussed the critical importance of reliability and the potential consequences of thermal and mechanical stress, commenting that the two most common failure modes—fracture and overheating—will each eliminate the ability to transfer data. In either case the car will not know what is happening or what to do next. It would effectively be blind and deaf. 

So how to improve reliability? Serzisko looked initially at improving solder joint reliability on BGA packages, and demonstrated the improvement in thermal cycling results using high-reliability Innolot alloys compared with standard SAC305. Advanced packages performing high-speed and power processing create more heat than conventional packages and therefore require greater thermal dissipation from the die. He discussed the attributes of copper leadframes and die-attach materials, and the benefits of silver-sintering in improving thermal conductivity at die-attach-to-substrate and die-attach-to-die interfaces. BGA solder-joint reliability on thermal cycling has been improved by a factor of four by the use of edge-bonding and corner-bonding techniques. Complete coverage of the package and the assembled printed circuit board with solvent-free conformal coatings have enhanced thermal performance while protecting from moisture and particulate contamination and further eliminating electrical issues, particularly in aggressive environments. 

Unsurprisingly, the automotive industry sought high-performance, low-cost packaging solutions. Flip-chip ball-grid-array (FCBGA) was gaining more adoption, particularly in safety applications, and the flip-chip chip-scale-package (FCCSP) was common for 5G capabilities. The collective requirement was for high-pin-count input-output devices, enhanced electrical properties and excellent thermal characteristics.

These attributes could be provided by embedded trace technology, which also enabled close-tolerance controlled impedance to minimise signal loss, critical in radar applications. Serzisko described process sequences for manufacturing embedded traces and two-in-one redistribution layers with fine lines and microvias. 

He concluded by repeating that thermal management remains the top reliability requirement, and that specific materials and processes both enhance reliability and allow space and weight savings.

The second presentation, given by Cédric Malaquin, team lead analyst for RF devices at Yole Développement, was an overview of the automotive radar market and technology trends. 

It was perhaps inevitable that his starting point would be CASE, “the acronym taking over the automotive industry.” His overall CASE market forecast, from a baseline of $73 billion in 2020, showed a compound annual growth rate of 12.4% to 147 billion USD in 2026, then compound annual growth of 9% to $318 billion in 2035. 

In the 2020 to 2026 period, the compound annual growth in electrification (E) was forecast at 27.67%, whereas ADAS (A) was 10.39% and connectivity (C) 8.97%. The highest growth forecast was in sharing (S) at 53.45%.

As the trend toward electrification and autonomous vehicles continues, new names are entering the market to compete with the traditional players. Malaquin’s impressive series of graphs showed the rise and subsequent tailing-off of light vehicle sales per level of autonomy over the period 2015 to 2050, from Level 0 through Levels 1-2, 2+, 2++, 3-4, to Level 5. Symbolic milestones on his time axis were “feet-off” in 2019, “hands-off” in 2029, “eyes-off” in 2039, and “mind-off” in 2049. 

It is not surprising to see top Tier-1 companies developing sensors to fulfil the autonomous goal, and the multiplication of sensors in vehicles will have direct impact on the complexity of the electrical/electronic architecture. This proliferation of sensors—cameras, ultrasound, long-range and short-range radar, and LiDAR—demand more ECUs, more computing power, more software, and more complex electrical/electronic architecture.

OEMs are currently still using a distributed electrical/electronic architecture with roughly one ECU per function, but there is an ongoing evolution from a distributed architecture to a centralized architecture. Eventually all domain controllers will be centralised into one super-computer, and there will be a paradigm change in radar sensing toward imaging radar. For fully automated vehicles, elevation capability will become essential, with tracking of stationary targets for a perception map. This will require a substantial increase in the number of tracked objects and a dramatic improvement in the dynamic range. Multiple suppliers were launching 4D imaging radar, and silicon was evolving to suit the requirements of the radar architecture, with performance-driven cascaded-CMOS becoming mainstream. 

Malaquin demonstrated the competitive landscape in automotive radar using a chart that showed established manufacturers of silicon devices for power, networking, RF, and processing, followed by Tier1.5 and Tier 1 electronic component suppliers, together with automotive OEMs. The accompanying lists of newcomers in each area are enormous, with some familiar and many unfamiliar names.

The cumulative start-up fundraising invested on radar development over the last seven years exceeds $1 billion, with most companies focusing on improved performance and imaging radar. The automotive radar platform market had been $5.8 billion USD in 2021 and was poised for a compound annual growth of 13%, with a 25-times growth in volume between 2021 and 2027. 

The momentum in 77 GHz radar is continuing and the phase-out of 24 GHz has started. 2022 is expected to see the introduction of commercial imaging radar, which is forecast to represent 30% of the market by 2027. 

The radar semiconductor market was $1.6 billion in 2021, with processor and RFICs representing 90% of the total. Growth in sensors, with increased semiconductor content per sensor, is expected to reach 14% compound annual by 2027. 

Malaquin concluded that the CASE automotive megatrend is continuing to push ADAS forward, and that Level 2.2 is considered to be the “sweet spot” in the short term.

Thanking the presenters and all who had participated, and especially Kirsten Smit-Westenberg and Tarja Rapala-Virtanen for organising a very successful event, Alun Morgan reminded everyone of the upcoming live EIPC conference in Örebro, Sweden, on June 14 and 15, including a visit to the Ericsson 5G Development Centre in Kumla.