The technological iteration route of automotive millimeter wave radar chips

The rapid development of intelligent cars and autonomous driving technology has made millimeter wave radar inconspicuous, and the widespread application of millimeter wave radar has driven the technological evolution of MMIC.

From the expensive gallium arsenide (GaAs) process in the early days to the mainstream CMOS and SiGe processes today, and then to the future promising FD-SOI process, the continuous upgrading of technology paths has effectively reduced the cost and improved the performance of millimeter wave radar.

We analyze the core requirements of in vehicle millimeter wave radar, provide a detailed introduction to different chip solutions, and look forward to their future development directions in technology and market.

Part 1 Core Requirements for Millimeter Wave Radar in the Automotive Industry
With the popularization of L2 and higher level autonomous driving technology, vehicles require more precise environmental perception capabilities. Millimeter wave radar, with its all-weather capability and high-resolution characteristics, has become one of the key sensors for achieving autonomous driving.

Governments in multiple regions around the world are pushing for mandatory installation regulations for ADAS (Advanced Driver Assistance Systems), further stimulating the market demand for millimeter wave radar. For example, the European Union has listed Automatic Emergency Braking (AEB) as a mandatory feature for new vehicles.

Millimeter wave radar is mainly used for distance measurement, speed perception, and target recognition, and its functional requirements cover the following points:
High resolution and wide field of view: Improve detection accuracy to support more complex driving scenarios, such as pedestrian avoidance in urban environments.
High detection range: Long distance capability can meet the driving needs of highways, such as advance planning of target lanes.
Real time response and anti-interference: Ensure fast and reliable data processing in a multi-sensor environment.
In the economical driving scenario of L0-L2, low cost has become an important consideration for millimeter wave radar, which requires highly integrated chips to reduce production costs.
In L3-L4 advanced autonomous driving, the 4D imaging radar needs to support thousands of virtual channels to achieve ultra-high resolution point clouds, which puts higher demands on the computing power of the chip.
The high integration and multi-channel design of millimeter wave radar chips make thermal management complex, especially in high-temperature environments where stable performance needs to be maintained.

Part 2 Technical Path and Solution of Millimeter Wave Radar Chip
The traditional automotive radar chip solution mainly adopts low resolution radar technology, which can provide basic distance and speed measurement functions.

RF front-end chip: The RF front-end of traditional radar chips usually uses materials such as gallium arsenide (GaAs) or silicon germanium (SiGe), which have high electron mobility and RF performance, and can achieve high-frequency signal transmission and reception. The RF front-end of radar chips operates in the 77GHz frequency band and can provide a certain distance resolution.

Signal processing chip: The signal processing chip is mainly responsible for processing the echo signals received by the RF front-end, including filtering, amplification, analog-to-digital conversion (ADC), and digital signal processing (DSP) functions.
In traditional radar chips, the computing power of signal processing chips is relatively limited, and simple signal processing algorithms such as fast Fourier transform (FFT) are mainly implemented using microcontroller units (MCU) or digital signal processors (DSP) for spectrum analysis, in order to obtain target distance and velocity information.
However, due to the limitations of its computing power, traditional radar chips may face certain difficulties in processing multi-target detection and recognition in complex environments, and their resolution is relatively low, making it difficult to meet the high-precision requirements of advanced autonomous driving for environmental perception.

The technological evolution of millimeter wave radar has gone through three main stages:
The GaAs Process Era (1990-2007)
With the advantages of high power output and high-frequency operation capability, but low integration and high cost, it is only used in a few high-end models. The first generation Mercedes Benz ARS100 radar uses GaAs technology to serve the luxury car market.

The SiGe process era (2007-2017)
Combining the low-cost advantage of silicon materials with the high-performance characteristics of GaAs, the integration level has been greatly improved, and the price of radar systems has decreased by more than 50%. The mainland ARS4 series radar uses 2 MR2001Tx and 4 MR2001Rx chips to achieve the basic functions of L2 level autonomous driving.

CMOS process and future FD-SOI process (2017 present)
◎ CMOS process: Further reduce the cost of radar systems and support single-chip MMIC to achieve all radar functions.
FD-SOI process: Suitable for future high-end 4D imaging radar applications by reducing power consumption and improving reliability.
Infineon's latest CTR8191 transceiver adopts 28nm CMOS technology, supports 4T4R configuration, and achieves high signal-to-noise ratio and low power consumption performance.

Cost driven: Targeting the L0-L2 market, emphasizing cost-effectiveness, highly integrated MMIC, typically configured with 2-4 transmit channels and 4 receive channels.
◎ TI AWR2944: Supports single-chip integration of MMIC and SoC, applied in short-range scenarios such as door radar and cabin liveness detection.
◎ Gatlan Alps series: Based on SiGe technology, it supports 3D radar and 4D radar functions and is widely used in the mid to low end ADAS market.

Performance driven: Targeting the L3-L4 market, supporting multi-channel 4D imaging radar, emphasizing high resolution, high detection range, and point cloud density.
Arbe Phoenix Solution: Adopting a 48T48R virtual channel architecture, it supports object detection up to 350 meters and ultra fine point cloud density.
Uhnder S80: The world's first automotive grade digital imaging radar chip, supporting 4 cascaded chips to achieve 3072 virtual channels, with excellent anti-interference ability.

Innovative technology: Packaging and loading with central computing architecture to improve signal transmission efficiency, reduce power consumption, and further reduce costs by reducing the number of PCB layers.
The NXP SAF85xx adopts LoP technology, supports efficient thermal management and reduces electromagnetic interference (EMI), and achieves centralized processing through domain control. The radar head only retains the RF front-end, simplifying hardware and reducing costs,
The NXP SAF86xx is designed specifically for central computing radar, optimizing raw data transmission performance and suitable for future autonomous driving perception architectures.
International companies such as NXP, Infineon, TI, and ST, with years of technological accumulation and market layout, have occupied the main market share of in vehicle millimeter wave radar. These companies are characterized by high performance and a wide product portfolio, such as NXP's SAF85xx and Infineon's RXS816x series, both of which support multi-channel high integration to meet L2+to L4 autonomous driving needs.
In addition, TI is in a leading position in the field of central computing architecture, and its products such as AWR2544 further promote the development of millimeter wave radar towards high performance.
In recent years, domestic enterprises such as Gatland Microelectronics, Sijie Microelectronics, and Millimeter Sensing Technology have developed rapidly and have entered the global millimeter wave radar industry chain.
Gatland's Andes series products target the high-end 4D imaging radar market, achieving a combination of low cost and high performance with 22nm technology;
Shijie Microelectronics and Muye Microelectronics focus on SOC and high integration solutions to help expand into the mid to low end market.
Some vehicle manufacturers are accelerating the layout of millimeter wave radar supply chains and promoting vertical integration strategies.
For example, Tesla enhances its system optimization capabilities through self-developed sensing components, while BYD accelerates its independent research and development process by collaborating with local chip companies.

The rapid development of millimeter wave radar chips has propelled the technological leap of intelligent driving from basic ADAS to L4 advanced autonomous driving.

Future development directions include:
CMOS and FD-SOI processes will further promote the high integration development of single-chip MMICs, continuously reducing the cost of millimeter wave radar systems.
High channel 4D imaging radar will become mainstream, promoting the deep integration of millimeter wave radar with sensors such as lidar and cameras, and enhancing overall environmental perception capabilities.
Driven by the explosive growth of the domestic intelligent driving market, local chip manufacturers such as Gatland and Silicon Microelectronics will usher in opportunities for rapid rise.

Summary
Millimeter wave radar is in a dual acceleration stage of cost optimization and technological improvement. Through technological innovation and market promotion, millimeter wave radar is expected to become the core sensor in the field of autonomous driving.

Source: OFweek