Highly Accurate and Robust Automotive Radar System Design. Markus Kopp Lead Application Specialist ANSYS Inc.

Transcription

1 Highly Accurate and Robust Automotive Radar System Design Markus Kopp Lead Application Specialist ANSYS Inc.

2 Introduction This presentation is an overview of a proposed design methodology for automotive radar systems. This presentation is meant to illustrate the advantages of simulation, as well as what can be reasonably simulated using state of the art electromagnetic field solvers. This presentation will concentrate on how to address some of the challenges faced when designing radar systems at, the very high, 77 GHz frequency. All results shown in this presentation were created using ANSYS HFSS RF.

3 HFSS RF Mechanical CAD (MCAD) Both interfaces Electrical are included Layout (ECAD) within the new unified desktop environment

4 HFSS RF HFSS RF using the new unified design environment also allows RF and μwave frequency domain circuit design Analysis types Linear Harmonic Balance Transient Loadpull Envelope PXF Analyses Oscillator TV Noise Phase Noise PlanarEM Simulations

5 Creating an Antenna Array Single Antenna Element Design

6 Initial Design Initial Design is implemented in HFSS for Layout interface Initial Designs are synthesized using built in calculators Transmission Line (TRL) Tool Antenna Estimator Tool Both tools use stackup definition provided by engineer

7 Initial Antenna Design Initial Design is created in layout Then solved using a planar MoM solver For simple structures MoM solver is extremely fast

8 Initial Planar EM Results (MoM) After some tweaking excellent S11 results are obtained However, Mom solver prefer 2D infinitely thin conductors At 77 GHz the finite thickness of the metal may change performance

9 3D thick Metal Results Re-simulating same antenna with mm (half ounce) thick metal produces different S11 results HFSS 3D clearly shows that more tweaking is needed Fine tuning will be done in HFSS-3D using adjoint derivatives

10 Rapid Antenna Tuning using Adjoint Derivatives Functionality (only available in HFSS 3D) Extremely efficient method used for tuning, sensitivity studies, and optimization Computes the derivatives of SYZ parameters with respect to project and design variables Eliminates need to solve multiple variations with small differences and numerical noise Allows real-time tuning of reports to explore effects of small design changes

11 Final Antenna Element Performance

12 Creating an Antenna Array Antenna Array Design

13 Creating an Antenna Array Creating an array once the unit antenna has been design is a straight forward process Antenna elements are arranged in a specific pattern, and then a net resultant antenna performance can be calculated This final antenna performance metric can be obtained using an antenna factor calculation or using an infinite array approximation or a finite sized antenna array can be simulated. Both the Antenna Factor calculation or the infinite array calculation are likely to produce erroneous results with regard to side lobes and back lobes A finite array calculation will provide the best results Simulating a finite array can be time consuming especially when the exact array spacing is still being determined

14 Finite Array Domain Decomposition (FA-DDM) Utilizes Replicated DDM Unit Cell to Address Array Concerns Geometry and Mesh copied directly from Unit Cell Model Unit Cell geometry expanded to finite array through a simple GUI Adaptive Meshing Process imported from Unit Cell Simulation Dramatically reduces the meshing time associated with finite array analyses. Mesh periodicity reinforces array s periodicity.

15 Finite Array Calculation using FA-DDM Efficient solution for repeating geometries (array) with domain decomposition technique (DDM) Direct solver with 12 cores 5:05: GB RAM Patterns from 8X8 Array Finite Array DDM with 12 cores 00:44: GB

16 Unit Cell Model for FA-DDM Unit cell model uses the HFSS Linked Boundaries Master and slaves boundaries are applied on opposite faces With and depth of unit cell determines antenna array element spacing Optimal array spacing (it s not lambda) can be determined efficiently and quickly Slave Master

17 Finite Array Capability Initial Array was modeled using FA-DDM Array spacing was optimized using FA-DDM Final 1 x 10 Array shows good performance

18 Explicit 1 X 10 Array Solution Final 1 X 10 Array was also explicitly simulated in HFSS 3D

19 Comparison of Explicit and FA-DDM Results Results between explicit HFSS 3D simulation and FA-DDM show excellent agreement FA-DDM has clear time advantage however and lends itself to rapid tuning of element spacing

20 Creating an Antenna Array Power Divider Design

21 Initial Divider Design Initial Design was again created in HSS Layout and analyzed using MoM solver for efficiency reasons However, similarly to antenna, finite thickness of metal makes a difference so Power divider was tuned in similar manner

22 Final 10 way Power Divider Design Final 10 way Power divider was designed To have the following power distribution , 0.125, 0.25,0.5, 1,1,0.5, 0.25,0.125,

23 Final 10 way Power Divider Design

24 Creating a Transmit/Receive Module Designing a 77 GHz Automotive Radar Module

25 Effect of Feed Network on Antenna Pattern The feed network will effect the antenna array performance. Simulating our array with attached feed network can show how detrimental the feed network is to our overall array performance No Feed With Feed

26 Radar Tx/RX Module Final Module consisted of 1X 10 Transmit Array 1 X 10 Receive Array Matching feed networks for TX and RX sides Radome of 1mm thickness Duroid er=9.8, 5 mil thick substrate

27 Antenna Pattern for full TX/RX Module

28 Effect of Radome on Array Performance Radome housing reduces back lobes but also flattens and widens main lobe Radome and antenna spacings can be optimized to reduce this effect

29 Final Module Design These optimized results are inclusive of plastic Radome, finite ground planes, feed network and TX/RX antenna structures

30 Full System Simulation Placing an Antenna Module inside a Car

31 Modeling an Antenna Array in its Deployment Environment Placing an Antenna Module in its deployment environment can be a very large and time consuming simulation. Using advanced modeling techniques and hybrid solvers can make these very large and time consuming simulations be manageable and efficient. Creating a hybrid FEM-MoM approach to solving very large simulations can be highly efficient and yield highly accurate results.

32 TX/RX Module in Deployment Environment Car Hood (Perfect Electric Conductor) At 77 GHz this simulation is extremely large! Conventional simulation methods are not efficient. TXRX Module TXRX Module Car Bumper (Plastic) Using a Hybrid Finite Element Method of Moments Approach can be used to solve this model in an efficient and accurate manner

33 Hybrid Finite Element-Integral Equation Method Finite Element Based Method HFSS Efficient handle complex material and geometries Volume based mesh and field solutions Airbox required to model free space radiation Conformal radiation volume with Integral Equations Integral Equation Based Method HFSS-IE Efficient solution technique for open radiation and scattering Surface only mesh and current solution Airbox not needed to model free space radiation This Finite Element-Boundary Integral hybrid method leverages the advantages of both methods to Finite achieve Elements the most vs. accurate Integral and Equations robust solution for radiating and scattering problems

34 Finite Element-Boundary Integral = FEBI True solution to the open boundary condition Surface currents directly computed by IE solver Very accurate far fields No minimum distance from radiator Advantage over ABC Reflection-free boundary condition Ability to absorb incident fields is not dependent on the incident angle Arbitrary shaped boundary Outward facing normal's can intersect Can contain separated volumes FE-BI does come with a computational cost Ability to create air box with smaller volume than ABC or PML can significantly offset this cost Air volumes that much smaller than ABC/PML boundaries will be solvable in less RAM with FEBI

35 FEBI Compare with Friis Transmission Two rectangular waveguide radiators contained in two separately spaced finite element and IE boundary domains Parametrically sweep separation and compare to theoretical Friis formula for free space transmission P r /P I = [(1- S 11 2 ) 2 G 2 ]/[16(πd/λ) 2 ] S11, ~0.3 and G ~4.5 computed from single radiator analysis

36 Transmit Antenna Location Investigation Proposed TX Antenna Location (Entire module not shown for clarity) High Location Low Location Using The FEBI method it is possible to investigate placement of Module within a deployment platform (Car, Truck, etc.) Effect of various plastic and metal obstructions can be evaluated Optimal Transmit antenna location can be determined

37 Transmit Antenna Location Investigation Low Antenna Pattern High Antenna Pattern

38 Transmit Antenna Location Investigation Low Antenna Pattern High Antenna Pattern

39 Obstructions at a Distance TX Array Large or infinite Metal Obstruction FEBI also allows engineers to place large or infinitely large obstructions at considerable distance from Antenna array. This can then be used to determine antenna performance in presence of obstruction or in a full system simulation where EM field solvers are combined with driving circuitry. This combined EM/circuit simulation is possible in HFSS RF but beyond the scope of todays presentation.

40 Obstructions at a Distance No Obstruction Large Obstruction at 5 m Distance

41 Obstructions at a Distance No Obstruction Large Obstruction at 5 m Distance

42 Final Thoughts Electromagnetic simulation can aid in the design of advanced radar modules helping to reduce time to market, design variability and manufacturing issues Advanced methods such as FEBI can be used to integrate Radar Modules into their deployment environment aiding design teams to ensure that Radar modules perform according to specifications Any questions?