Design and Analysis of Chirped Automotive Radar Systems
Thanks to the German inventor Christian Huelsmeyer, radar has been around since 1904, the year he patented his telemobiloscope. This simple radar device could detect ships up to 3000 m away and operated at wavelengths from 40-50 cm with a separate receiver antenna. The device could only be used to determine the presence of a distant object, but later versions could be used to calculate range.
Today’s semi-autonomous vehicles, and upcoming fully autonomous vehicles, require various automotive radar, camera, and LIDAR systems. High-resolution LIDAR systems with lower costs and ranges reaching hundreds of meters are still being developed, but cameras and radar systems are already on the market in anticipation of greater use of driverless cars. The majority of today's automotive manufacturers assume that all three types of systems will be required in varying capacity for fully autonomous vehicles.
How Does Chirped Automotive Radar Work?
Modern radar systems send a modulated RF pulse from a transmitting antenna and detect an echo in the receiving antenna. If the distant object is stationary, then a Doppler shift will not occur, and you can calculate the distance to the object using the signal bandwidth. This is preferable to using an unmodulated signal, which required measuring the time delay between the moment the pulse was transmitted and when the echo pulse was received. If the source and the distant object are moving with nonzero relative velocity, the pulse will experience a Doppler shift, both when it is emitted from the source and reflected from the distant object. The magnitude of the Doppler shift can then be used to determine the relative velocity between the source and the distant object.
Determining the Doppler shift requires comparing the received signal with a reference frequency using a mixer, which then produces a beat frequency that is equal to the difference between the transmitted and received frequencies. Although you could certainly lay this out on a PCB yourself and extract the beat frequency using a phase-locked loop, this functionality is typically integrated into an IC, providing all the functionality you need in a single package.
Using a chirped pulse provides more information in that it can be used to determine range and velocity from multiple targets simultaneously. 1 or 4 GHz linear chirp is typically used in 77 GHz automotive radar systems, which makes data processing easier while still providing accuracy that is comparable to exponential chirp. When combined with beamforming using a phased array antenna, you now have a system that can be used to determine range, velocity, and angle-of-approach simultaneously. The useful range in this type of system is only limited by the sensitivity of the antenna and the noise floor in the transceiver.
Multiple automotive radar modules can provide complete directional detection
One challenge in automotive radar design arises when multiple vehicles with radar operating in a similar frequency range interfere with each other. If we have two radar-equipped cars (Car A and Car B), the radar signal from Car B can mask detection of any true targets that would otherwise be detectable by Car A. Furthermore, Car B’s radar will cause multiple false targets to be seen by Car A. Car A will then have the same effect on Car B. Because the antennas and other circuitry on the PCB cannot be shielded if it is to be used as intended, the signal originating from Car B’s radar must be removed at the PCB or IC level.
Once the chirped echo signal is received, it is normally converted to a digital signal and processed using windowed Fourier transforms to group the signal into a range-velocity grid. The radar signal from a neighboring radar-equipped vehicle can be removed using a carefully-designed anti-aliasing filter prior to analog-to-digital conversion. If you are using a pre-packaged automotive radar transceiver, you should check to see that this important signal processing step is performed.
Transceiver Layout for Chirped Automotive Radar Systems
A system-level diagram for a chirped radar system can be a bit complicated, but thankfully, you do not need to layout every portion of the system as chirped automotive radar ICs are now available on the market. This integrates most or all of the required components for a usable chirped radar system at ~77 GHz into a single package. These transceivers are available in an integrated module that includes the required antennas. If you actually look at these transceiver modules, they use a set of 3 transmit and 4 receive series-fed patch array (SFPA) antennas, where each antenna is surrounded by some shielding, i.e., a coplanar waveguide configuration.
If you’re an antenna or radar designer and you are experimenting with some other design aside from SFPA, then you’ll want to procure the transceiver chip on its own and design a small RF module with microstrip antennas on the surface layer. Because transceiver module manufacturers provide an integrated module, you can look at these modules to get a good idea of how to build your own board to support customized antennas for your particular application. You’ll need to include antenna switches to provide directionality if this functionality is not integrated into your transceiver.
You’ll have to miniaturize your phased array for automotive radar
The transceiver should be set as close as possible to the antenna array, although you won’t be able to eliminate transmission line effects at the relevant frequency; at 77 GHz, the free-space wavelength of the signal is 3.9 mm, or ~1 mm in a microstrip trace, so nearly every trace in the board will be acting like a transmission line. The transceiver itself may be matched to a particular impedance at its output frequency and power value due to any nonlinearity of your transceiver, as the input/output impedance can vary with input power. The transceiver manufacturer may recommend you design your antenna and any impedance matching network to match a particular impedance to ensure maximum power transfer.
Once the beat signal is extracted and converted to a digital number, you’ll now need to worry about sending a high speed digital signal over a connector to an ECU. The connectors used in these transceiver boards are typically shielded, preventing any interference between the RF sections of the board and the output. The transceiver should be placed as close as possible to the connector and the digital outputs should run directly to the connector. At least one of your leads to the connector should be grounded to provide a return path with minimal inductance. These steps should minimize EMI between this minimal digital section and the rest of your RF components.
Don’t Stop Innovating at Automotive Radar
The new systems involved in driverless vehicles don’t end with camera, LIDAR, and radar. V2X (vehicle to everything) wireless connectivity via 2G-4G networks, as well as upcoming 5G, over IEEE 802.11p (5.9 GHz) is already being integrated into newer vehicles. V2V networking (i.e., over VANETs) is also critical to allow coordinated driving among multiple autonomous vehicles, ultimately improving safety, fuel efficiency, and congestion. BroadR-Reach Ethernet at 100Base-T or 1000Base-T is important for in-vehicle connectivity between ADAS, control, connectivity, processing, and other systems. There is plenty of room for innovation in these areas, both at the systems and PCB level.
Any self-driving car design team will need advanced PCB design software with layout, simulation, and signal integrity features for automotive radar systems and other applications. Altium Designer gives you a full suite of design and simulation features in a single program. The simulation and signal integrity features work directly with your design data and are accessible in a single intuitive interface.