FPGAs Helping Cars Get Smarter

One of the many interesting things we all learned during the pandemic is that supply chains can cause problems for lots of people. Automotive industry supply chains in particular are, well, really complicated. Even general consumers started to learn how important semiconductors are to the production of today’s increasingly intelligent cars and how, if chips are in short supply, vehicles can’t get produced.

What they probably didn’t quite get to is what a critical role FPGAs play in today’s latest car designs. In fact, it’s not uncommon for a technologically advanced car to have as many as 10-12 FPGAs within them, often of different types and performing a range of important functions.

Enabling flexible adaptation from today's architectures towards emerging Zonal Architectures

The increasing numbers of cameras and other visual sensors on cars, for example, is a great application for FPGAs because they can control and route the signal from on-car cameras into the main computing element in the vehicle, typically some type of SoC (system on chip). The bridging functions performed by FPGAs allow multiple high-resolution cameras to have their signals combined or multiplexed together near where the cameras are located—such as around the front or rear bumpers—and then have those signals separated again, or demultiplexed, when they arrive at the SoC.

Many modern cars are starting to offer these zonal architectures where groups of sensor, computing, and interconnect elements are combined into different computing “zones” and then the results from different zones are sent to the main SoC. One of the challenges of all these connections is that they can vary quite a bit from vehicle-to-vehicle. While many cameras and other sensors such as Radar and LiDAR use MIPI standards, other intra-car connections often use high-speed SERDES (Serializer/Deserializer) to enable longer distance connections. On top of that, other systems use older automotive standards including LIN (Local Interconnect Network) Bus, CAN (Controller Area Network) Bus, and even automotive Ethernet. Thankfully, appropriately designed FPGAs can work over all these different connection types and even serve as a bridge between different standards.

Real-time networking and motor-control

In addition to interconnection capabilities, one of the other key benefits of FPGAs is their ability to perform signal processing and analytics with low latencies and at a deterministic rate every time. In a real-time based application like assisted driving or even simply operating a vehicle, this consistency is essential to the performance of the car and its functions.

In electric/hybrid vehicles, for example, FPGAs can be used for fine-tuned motor control, power conversion, and inverters, because of their support for high performance PWM (pulse width modulation), all of which help maximize power and efficiency to extend the range of the vehicle.

Lowest power and smallest package

On top of these purely functional aspects of FPGAs, there are several inherent characteristics about certain FPGAs that make them well suited for automotive applications. First, because Electric Vehicles (EVs) run on batteries, low power consumption is an important factor to consider, making low-power FPGAs like those from Lattice Semiconductor well suited to the task.

In addition to focusing on low power, Lattice also offers a broad portfolio of FPGAs with support for industry leading small packages and a wide range of processing capabilities, including Lattice ECP™ family, Lattice MachXO™ FPGAs, and Lattice Nexus™ platform based FPGAs. Lattice also offers some of the highest density I/O (input/output) enabling increased flexibility and adaptability, yet another important metric for the auto market.

Cyber resilient automotive with secure, dynamic, real-time solutions

Safety and reliability are also critically important for any automotive applications. In fact, the auto industry has several strict standards for meeting the requirements necessary for safe and consistent operation in a vehicle. The ISO 26262 functional safety standard, for example, requires components to not only comply with the physical demands of an automotive environment—including reliable operation across extreme temperature ranges—but also have accompanying software that operates in a continuous safe fashion. Lattice Diamond® software, which is used to design the functionality for its MachXO and ECP families is ISO 26262 certified.

In addition to these certifications, Lattice’s Nexus-based FPGAs offer inherent protections against soft errors (caused by environmental factors like alpha particle radiation, etc.) which can cause calculation inaccuracies. Nexus-based products, which are manufactured on a 28 nm process utilizing FD-SOI (Fully Depleted Silicon on Insulator) technology, are up to 100x more reliable against soft errors than other standard FPGAs.

Digital security is also becoming an increasingly important part of automotive designs, particularly as cars start to become computers on wheels and become subject to some of the same kinds of hacks we’ve seen with our digital devices. As it turns out, FPGAs are already being used to perform robust cyber resiliency functions in applications like servers and some of those same practices are moving into the automotive world as well.

As the automotive industry continues with its disruptive transition to Electric Vehicles, increased ADAS / Autonomous capabilities, and adoption of zonal architectures, FPGAs offer an ideal platform. Adapting to a wide range of ever-changing requirements serving as a key component of auto design, FPGA’s innate programmability allows them to evolve with changing needs and deliver critical—albeit oftentimes invisible—capabilities to the operation of modern vehicles.

Bob O’Donnell is the president and chief analyst of TECHnalysis Research, LLC a market research firm that provides strategic consulting and market research services to the technology industry and professional financial community. You can follow him on Twitter @bobodtech.