Slide 1

Dr. Er ic K . Wal tonThe Ohio State Univers i ty ; E lectroSc ience

LaboratoryColumbus , Ohio 43212

Wal ton .1@osu.edu

Dr. Wladimiro Vi l l arroe lAGC Automot ive Amer icas R&D

Yps i lant i , MI 48197WVi l larroe l@us .agc-automot ive .com

Multi-Parametric Antenna Test Visualization for Optimization

Session Information Here

Slide 2

INTRODUCTIONThere are a large number of antennas needed for

modern automotive communications systems AM, FM, FM diversity TV & TV diversity Remote keyless entry/start Cellular; Bluetooth Automatic toll systems Smart highway information systems GPS and GPS information systems (traffic information) Radar systems (backup, side impact, lane departure, intelligent

cruise control)). Manufacturers are looking for ways to reduce the total number of

such antennas by using combinations of a smaller number of antennas.

This paper will discuss a software approach that permits the engineer to visualize the antenna performance effects of variations in geometry for a group of antennas based on either test data or simulation.

Slide 3

ACKNOWLEDGEMENT

MUCH OF THIS WORK IS TAKEN FROM AN UNDERGRADUATE INTERN PROJECT BY

MR. RYAN TOKOLAA 2009 UNDERGRADUATE INTERNAT THE OHIO ST. UNIV. ELECTROSCIENCE

LAB.

THE PROJECT WAS SUPPORTED BY AGC AMERICA INC., YPSILANTI, MI.

Slide 4

AGC AMERICA

Asahi Glass Company (AGC) A core Mitsubishi company World-class manufacturer and innovator in the

fields of glass and fluorine chemistry AGC Automotive Americas R&D, Inc.

AGC subsidiary in Ypsilanti, Michigan Dedicated to the development of new

technologies and new products for processed automotive glass

4

Slide 5

Why Designing Car Antennas is Difficult

On-glass (conformal) antennas are becoming more popular Low-cost Easy to manufacture Unobtrusive

Limited available set of possible locations Must have minimal visual obstruction on side and rear

windows Limited to fade band of windshield

5

Slide 6

Why Designing Car Antennas Is Difficult

6

Even a simple antenna has a very large number of possible geometric distributions

Today's codes can predict the performance of an antenna, but not “design” an antenna

Slide 7

EXAMPLE PARAMETRIC VARIABLES

FEED POINT LOCATION

INTERCONNECTION

LOCATION

ELEMENTLENGTHS

FOR EACH PARAMETER, THERE ARE DIFFERENCES IN THE GAIN PATTERN AND IMPEDANCES

THUS THERE ARE A VERY LARGE NUMBER OF POSSIBLE COMBINATIONS AND PERMUTATIONS

WE NEED A WAY TO CHOOSE THE BEST.

THERE IS ALSO AN ANTENNA IN THE FRONT OR SIDE

Slide 8

VISUALIZATION MOTIVATION

Simply plugging a cost function into an optimization algorithm is insufficient. Human interaction (judgment) is required to: Refine the cost function Analyze the solution space and determine which

optimization algorithms are appropriate Identify regions of interest for optimization

8

Slide 9

Software Outline

The software has the following sequence:1. Using

a. Theoretical modeling data setsb. Experimental measurement data sets.

2. Create a data set of antenna gain performance as a function ofa. Polarizationb. Frequencyc. Increments in antenna geometry

3. Graphically Display the Consequencesa. Gain vs. Azimuth vs. Antenna wire locationsb. Polarization vs. Azimuth vs. Antenna wire locationsc. Overall dual antenna gain vs. Azimuth

i. As diversityii. As phase combined

d. Cost function behaviori. Vs. wire and interconnect locationsii. Vs. antenna 1 and antenna 2

Slide 10

GUI TO DEFINE PARAMETERSFOR A THEORETICAL MODEL (ESP5)

WE DEFINE ANTENNA WIRES WITH PARAMETRIC GEOMETRICAL INCREMENTS

Slide 11

Setup and Simulation

Each endpoint may be swept in one or two directions

Example: [min = -4, step = 2, max = 4] results in five simulations with the wire endpoint at {-4, -2, 0, 2, 4}

step = 0 if not swept“xxxx” indicates unnecessary info (vert. wires have only one x-coordinate, horiz. wires have only one y-coordinate)

Slide 12

DISPLAY THE WIRE LAYOUTFOR EACH ANTENNA

Slide 13

DEFINE THE VEHICLEAND INSERT THE ANTENNA

USER CAN DOUBLE CHECK THE LAYOUTPRIOR TO THE (OVERNIGHT) CALCULATION

INSERTEDANTENNAS

VEHICLE WIRE GRID MODEL WITH INITIAL ANTENNA

GEOMETRY

E X A MP L E F

OR

S E DA N

Slide 14

SUB-SCALE MEASUREMENTS

SUB-SCALE MEASUREMENTSFOR COMPARISON

GROUNDPLANE

NOTSHOWN

Slide 15

15 15

Example Simulation for a Van:

GMC Van Mesh Wireframe with Front & Rear Antenna Apertures

Y

X

Z

Y

X

ZCreated using make car mesh of VAAR

EXAMPLE FOR VAN

Slide 16

Measurement of a full scale van

WE CAN COMPARE MEASUREMENT AND SIMULATION

Slide 17

Data Visualization

User-defined functions have access to four arrays of data for each specific parameter combinations: FGth (Vertically polarized gain of front antenna) FGph (Horizontally polarized gain of front antenna) RGth (Vertically polarized gain of rear antenna) RGph (Horizontally polarized gain of rear antenna)

Each of these is a 1x360 array giving the 360° azimuthal antenna radiation pattern in dBi

17

Slide 18

GUI FOR SETTING UPPARTICULAR VISUALIZATION

Slide 19

Data Visualization

User-defined variables (a-f) can contain FGth, FGph, RGth, RGph, or any previous variable (b can contain a as an argument)

Any MATLAB functions (max, mean, etc.) can be used

19

Slide 20

20

Comparing Data Visualizations

Slide 21

EXAMPLE VISUALIZATION( AZIMUTH PLOTS)

FGph

a=FGphb=RGphc=max(a,b)d=FGthe=RGthf=max(d,e)Plot:Max(c,f)

a=FGthPlot(a)

a=RGphPlot(a)

Slide 22

EXAMPLE: GAIN VS. FREQ. & AZIMUTH

Slide 23

VISUALIZATIONS FORMEAN REAR GAIN (φ POL) AND MIN OF MAX

FREQ

WIRE LOCATION

a=RGphPlot(mean(a))

a=FGph; b=RGph; c=max(a,b)d=FGth; e=RGth; f=max(d,e)Plot(min(max(c,f))

Slide 24

24

Optimum Antenna Configuration

Mean front antenna gain phi & theta of wire location vs.

frequency

Minimum front antenna gain phi & theta of wire location

vs. frequency

Maximum front antenna gain phi & theta of wire location

vs. frequency

Slide 25

25 25

Conclusion (What I’ve Learned & Where I’m Going) Ryan’s code has shown a considerable amount of thought

and effort using programming and antenna theory to visualize & then optimize an antenna configuration on a vehicle.

Future studies Study various types of cost functions and there impact on

antenna configuration. Modify the code to allow the user to create only one antenna. Modify the code to allow the user to have an antenna with static

geometry. Finally we plan to discuss future project options with AGC

personnel.

Comments & Questions

Slide 26

References

[1] Kraus J. D. and R. J. Marhefka, Antennas for All Applications, 3rd Ed., McGraw-Hill, N.Y., N. Y., 2002.

[2] Abou-Jaoude, R., and Walton, E. K., “Numerical Modeling of On-Glass Conformal Automobile Antennas”, IEEE Trans. Antennas Prop., vol. 46, pp 845-852, June 1998

[3] Tokola, Ryan and Walton E. K., “Visualization Software for Vehicle Antenna Design,” OSU ElectroScience Laboratory Department of Electrical and Computer Engineering, Columbus, Ohio, Technical Report 60003388-1 and 60024198-1, January 29, 2010

[4] Newman, E.H., “A User’s Manual for The Electromagnetic Surface Patch Code ESP Version 5”, Technical Report 716199-1 1, The Ohio State University ElectroScience Laboratory, Department of Electrical Engineering, 1995

Slide 27

Slide 28

Slide 29