In recent years, autonomous driving has become one of the hottest topics. With the development of autonomous driving technology, millimeter-wave radar, as one of its key sensors, has also received widespread attention. However, at present, overseas giant companies dominate the global millimeter-wave radar industry, among which the Continental ARS540 leads the global millimeter-wave radar technology with nine leading absolute advantages.
Figure: Continental ARS540
First, the ARS540 is the millimeter-wave radar with the highest resolution, with a horizontal resolution of 1.2° in azimuth, while most millimeter-wave radars have a horizontal resolution of only 5°, making the ARS540 five times that of ordinary millimeter-wave radars.
Secondly, the ARS540 is a millimeter-wave radar that can truly measure the height of targets, with a vertical resolution of 2.3°.
Figure: Height Measurement of ARS540
Third, the ARS540 has an effective distance of up to 300 meters at a horizontal field of view (FOV) of 100°, while other millimeter-wave radars have an effective distance of 250 meters, but the FOV is reduced to only 8-10°, far below the ARS540’s 100°.
Fourth, the ARS540 is a millimeter-wave radar capable of outputting images, which Continental refers to as Radar Detection Image Output (RDI), with performance comparable to 8-line laser radars.
Fifth, the ARS540 uses micro-Doppler technology, which can detect vulnerable road users (VRUs) such as the elderly and children.
Sixth, the ARS540 employs Synthetic Aperture Radar (SAR) technology to improve near-range resolution; within a range of 10 meters, the ARS540 can use SAR technology to enhance angular resolution.
Seventh, the ARS540 is a millimeter-wave radar that does not filter out static targets. Currently, most millimeter-wave radars filter out static targets to reduce false alarm rates; however, the ARS540 maintains ultra-high resolution while being able to detect object heights, significantly reducing the false alarm rate and thus retaining static targets.
Eighth, the ARS540 is capable of trajectory tracking and prediction in complex traffic scenarios.
Ninth, the ARS540 can detect hidden vehicles. The detection rate of hidden vehicles by the ARS540 has improved from 40% in the previous generation advanced product ARS430 to 80%, while most millimeter-wave radars can only achieve 20%.
Detecting Hidden Vehicles (Fleet)
The absolute first performance of the ARS540 is backed by its complex design.
Figure: Structural Diagram of ARS540
In terms of structure, the ARS540 uses a cascaded design with four cascaded NXP 77GHz millimeter-wave radar transceivers (MMIC) MR3003; each MR3003 has 3 transmitters and 4 receivers, totaling 12 transmitters and 16 receivers for four chips. Currently, most millimeter-wave radars use single-chip transceivers, typically with only 3 transmitters and 4 receivers, resulting in only 12 virtual channels, while the ARS540 boasts up to 192 virtual channels, greatly improving resolution. It can be referred to as an image radar.
Figure: Annotation of Image Radar; Image Source: NXP
Currently, most millimeter-wave radars are still in the first stage, using a single MMIC. Most forward main millimeter-wave radars are mainly monopolized by Continental and Bosch. Mid to low-end models use Continental’s ARS4B, ARS408, or ARS410, all of which rely on an older NXP transceiver MR2001. For cost-saving, Tesla Model 3 uses the cheapest ARS4B, with an effective distance of only 170 meters, while many Chinese domestic brands use ARS410 with an effective distance of 250 meters. Foreign high-end models use ARS510, which is based on the latest MR3003. Bosch promotes MRR1 PLUS in China, using Infineon’s separated transceiver RRN7745P/RTN7735P. The long-range LRR4 uses a single MR2001.
Figure: MR2001 System Diagram. MR2001 is the previous generation product of MR3003 and is also the most widely used long-range millimeter-wave radar transceiver, used by Bosch, Continental Group, Aptiv, and Autoliv.
Figure: Internal Framework of MR3003
From the above two images, we can see that MR3003 has undergone significant changes compared to MR2001. Firstly, MR3003 integrates ADC (analog-to-digital converter), while MR2001 requires passing through a low-pass filter to filter out noise before sending the analog data to the MCU for processing, which may result in signal loss and affect accuracy. MR3003 integrates the ADC conversion, eliminating the need for ADC in the MCU, using MIPI CSI2 or LVDS output, allowing for high bandwidth output, laying the foundation for image output and improving the signal-to-noise ratio, which means better resolution and accuracy. Secondly, a functional safety module has been added. Thirdly, it integrates a PLL (phase-locked loop) circuit instead of the external VCO used in MR2001, resulting in lower costs and higher reliability. Finally, MR2001 has a 4-transmitter 3-receiver antenna design, while MR3003 has a 3-transmitter 4-receiver antenna design.
In terms of MCU, the ARS540 also uses the new generation S32R274, while the typical counterpart for MR2001 is MPC5773.
Figure: Internal Framework of S32R274
S32R274 adopts a tri-core design, with two e200z7260 cores responsible for computation and one e200z420 core dedicated to safety, achieving safety applications up to ASIL-D level. In terms of communication interfaces, it offers maximum flexibility, far superior to the previous generation, with added support for Ethernet, featuring Ethernet MAC, supporting over 100 Mbps Ethernet RGMII, and enabling image output. It has a pair of Flexray bus channels supporting 128-bit information buffering. There are three flexible CAN channels supporting CAN-FD. Additionally, it supports high-speed serial communication up to 320 Mbps via Zipwire. The default output is still CAN, with Ethernet output as an alternative.
Radar Signal Processing Diagram; Image Source: NXP
In terms of algorithms, the focus is on enhancing CPI and DOA. Chirp refers to a technique in communication technology related to encoding pulse signals, where the carrier frequency increases linearly during the pulse duration, producing a sound reminiscent of birds chirping. FFT stands for Fast Fourier Transform, and CFAR stands for Constant False-Alarm Rate. In radar signal detection, when the intensity of external interference changes, the radar can automatically adjust its sensitivity to maintain a constant false alarm probability, and receivers with this characteristic are termed constant false alarm receivers. The DOA (Direction of Arrival) estimation algorithm is core and typically requires high computational power, with NXP or Freescale’s Power architecture being suitable, while ARM architecture tends to be costlier. DOA estimation involves applying spatial Fourier transformation to the received signals, obtaining the spatial spectrum, and estimating the direction of signal arrival.
The ARS540 employs a dual MCU design, with two S32R274 chips. Each S32R274 connects to two MR3003 chips. In a sense, the ARS540 is equivalent to four ARS510 radars in one box. This complexity is far greater than a single radar, as radar operates at ultra-high frequencies, making mutual interference likely; many nuances cannot be resolved through computer simulation alone and require long-term experience accumulation, necessitating expertise in both RF electronics and materials/mechanical structures. The ARS540 is also the first mass-produced millimeter-wave radar to adopt a cascaded design. Not only is the design challenging, but the manufacturing process is as well, as the thickness of the shell material can affect radar performance, posing significant challenges for consistency during production, which often relies on experiential knowledge to resolve. China has many excellent designers, but lacks experienced frontline workers.
As millimeter-wave radar performance increases, it begins to threaten the status of laser radar, especially 4-line laser radars. In the future, if 6 or 8 units can be cascaded, millimeter-wave radar will pose a threat to 16-line laser radars. Meanwhile, millimeter-wave radar also holds overwhelming advantages in cost-performance ratio and automotive standards, which means that in the future, laser radar must focus on improving resolution to maintain its status.
*This article is contributed by the author of Gaishi Auto and represents the author’s personal views.
Digital Modulation Series: Characteristics of IQ Modulators
Six Questions on RF Pulse Spectrum Testing
How to Accurately Test 75 Ohm Systems?
How to Accurately Test Phase Noise
Brief Overview of RF Pulse Spectrum and Desensitization Effects
Spectrum Analysis Series: How to Choose a Display Detector
Spectrum Analysis Series: Why Do We Need a Preselector?
Spectrum Analysis Series: A Brief Discussion on Resolution Bandwidth (RBW)
Spectrum Analysis Series: How to Understand the Spectrum Leakage Effect in FFT?
Spectrum Analysis Series: How to Understand the Relationship Between Time Window and RBW in FFT
Spectrum Analysis Series: Why Does Reducing VBW Not Reduce the Noise Floor of the Spectrum Analyzer?
Digital Modulation Series: How to Understand IQ?
Digital Modulation Series: Basic Theory of IQ
Digital Modulation Series: Brief Overview of IQ Modulation and Demodulation
Noise Figure Series: Sources, Definitions, and Impacts of Noise
Noise Figure Series: Antenna Noise Temperature Testing Methods
Noise Figure Series: Gain Method for Noise Figure Testing
Noise Figure Series: A Brief Analysis of Noise Figure Testing in AGC Circuits
Noise Figure Series: A Brief Analysis of Noise Figure Testing in Multi-Channel Receiving Units
Noise Figure Series: The Impact of Image Frequency Suppression Filters on RF Receiver Output Noise
“5x Rule” When Choosing an Oscilloscope
Submission & Column Cooperation Please add the editor’s WeChat | WeChat ID: 18675536035
Please click the bottom right corner to give me a little appreciation!
