Automotive Applications in WIPL-D Software

EM simulations have a significant role in automotive industry. WIPL-D software continuously improves its variety of tools which allow various applications in this growing industry. The range of EM simulations has been extended with introduction of CAD tools (allow easy import of CAD files, as well as modeling and positioning of devices in conjunction to complex CAD geometries) and GPU simulation module (which extended the range of frequencies where applications can designed and simulated).

GPS Antenna Mounted on Car Roof

This EM simulation demonstrates a basic use where a simple patch antenna is mounted on the Citroen shell.

Figure 1. Citroen car shell simulation scenario

The antenna is placed on roof, symmetrically so that number of unknown coefficients is halved and simulation time is reduced. The car shell and magnified area where patch antenna is positioned are presented in Fig 1. Linearly polarized patch is 65 mm wide, on 2 mm thick dielectric plate (Er= 2.2). Radiation pattern of standalone Fig 1 (1.59 GHz operating frequency), while mounted pattern is presented in Fig 2.

Figure 2. Mounted patch antenna pattern

The next step is to extend a simulation by using circularly polarized patch and a full shell model, since the patch antenna is no longer symmetrical. The dimension of the slit is 6.4 mm (Figure 3).

Full model of car shell now requires 44,000 unknowns, almost two times more than in previous simulation. Simulation now typically lasts 15 minutes per frequency on GPU based platforms. However, WIPL-D suite offer features to reduce number of unknown coefficients on models parts which are far away from the antenna itself - antenna placement reduction. When reduction is set to maximum, EM simulation requires only 17,000 unknowns which yields in one minute run time on GPU platforms. Figure 3 demonstrates that antenna placement reduction almost has no effect on result and compares return loss of the antenna in free space and mounted on car roof.

Figure 3. Patch return loss, free space and mounted

Bluetooth and GSM Interference

The following example demonstrates interference within car shell of devices working with various wireless technologies can be efficiently simulated. The scenario includes Citroen car shell with added car seats, GSM mobile device on front seat and Bluetooth devices on the command board.

Figure 4. Complex multiple antenna scenario

Mobile has two antennas, GSM and Bluetooth. Radiation pattern of the GSM antenna in free space and inside car is shown in Figs 4-5 (1.8 GHz).

Figure 5. GSM antenna pattern inside car shell

Figure 6 presents coupling between Bluetooth antennas on Bluetooth device on command board and Bluetooth antenna inside mobile phone.

At 1.8 GHz (GSM) simulation that includes car shell with seats and control board added, as well as two generic devices, requires 92,000 unknowns which is simulated typically in one hour on GPU based platforms. At 2.4 GHz EM simulation requires 156,000 unknowns and simulation time is typically 4 hours per frequency.

Figure 6. Coupling among devices (free space, in car)

FM Antenna Immersed into Glass Window

The next example demonstrates how FM receiving wire antenna (108 MHz) can be immersed into car window. The window glass is modeled as dielectric (Er=3.5). The wire is immersed into window along with heating wires.

Figure 7. Mercedes shell with rear window included

EM simulation requires 39,000 unknowns and lasts around 15 minutes on GPU platforms. Radiation pattern of the FM antenna is illustrated in the following figure.

Figure 8. FM antenna radiation pattern

Vehicle to Vehicle Communication

One of the emerging technologies in automotive industry is vehicle to vehicle communication, as well as general interaction of the vehicles with the environment. The main applications are toll and safety systems, as well as auto pilot and parking sensors systems. This leads to expansion of number of antennas mounted on vehicles.

Typical EM challenge in such cases is design of optimal antenna and positioning it. One of the arising frequency bands is 5.9 GHz. It offers physically small antennas that do not disturb car design. But cars at such high frequencies represent electrically large objects. In that sense unique features offered by WIPL-D software suite allow full wave simulation at these RF frequencies. Among those are higher order basis functions, optimum quadrilateral mesh of CAD models and GPU enhanced simulation.

In the following example we demonstrate placing 3 short monopole antennas on generic car model and an advanced scenario where we analyze 2 cars in one simulation.

Figure 9. Vehicle to vehicle scenario

The location of three antennas is as in the following images.

Figure 10. Antennas locations

In the first simulation, we run a single car model with 3 antennas installed. We investigate return loss of the antennas (only antenna #1 has weak return loss due to its mounting) and coupling between antennas.

Figure 11. Antennas return loss

Simulation was performed on desktop PC (quad core CPU, 24 GB RAM) enhanced with 2 GPU cards and several hard discs. Symmetry was applied and problem required 90,000 unknown coefficients in MoM system matrix. Simulation time per frequency is around 1 hour.

Figure 12. Antennas coupling

Radiation pattern at 5.9 GHz of the third antenna (mounted on the roof) is shown below.

Figure 13. Third antenna radiation pattern

The final scenario includes 2 vehicles and 6 antennas in total. No symmetry can be applied so total number of unknowns is 370,000.

Figure 14. Scenario final model

The EM simulation is now extremely challenging. It was performed on GPU cluster with 7 nodes, each node containing quad core CPU and 2 Nvidia GeForce GTX cards. Number of unknowns was reduced by putting bottom of cars and parts insignificant for EM simulation in the shadow region, where number of unknowns is significantly lowered.

Number of unknowns was also reduced via antenna placement reduction, which means orders of current approximation are lowered on parts away from antennas. All the previously described features are automated. As the last possibility to spare unknown coefficients, parts of the model were selected manually (back side of the car, interior side of vehicle doors ect) and reduction was put specifically on these parts. Such modification requires user intervention and understanding of EM simulation.

Such simulation lasts 7 hours (matrix fill in – 4,800 s, matrix inversion – 14,927 s, and calculation of radiation pattern with high resolution 6,152 s).

Figure 15. First antenna radiation pattern (complex scenario)

Figure 16. Second antenna radiation pattern (complex scenario)

PDF application note

If you have a specific problem in mind and you are not sure if WIPL-D software can handle that problem, contact us. We will analyze your needs and try to help, and if our products satisfy your requirements, we will make you the best possible offer for purchase.