Aperture Efficiency of Amplitude Weighting Distributions for Array Antennas

By: Glenn D. Hopkins, glenn.hopkins@gtri.gatech.edu Justin Ratner, justin.ratner@gtri.gatech.edu Anya Traille, anya.traille@gtri.gatech.edu Vic Tripp, vic.tripp@gtri.gatech.edu Sensors and Electromagnetic Applications Laboratory Georgia Tech Research Institute Georgia Institute of Technology

AbstractArray antennas offer a wide range of opportunities in the variation of their directivity patterns through amplitude and phase control. Peak sidelobe levels may be reduced via amplitude control or weighting across the array aperture. Several authors have made significant contributions in detailing processes for synthesizing these aperture amplitude distributions for the purpose of sidelobe level control. One of the basic trade-offs when implementing amplitude weighting functions is that a trade between low sidelobe levels and a loss in main beam directivity always results. Some of the commercially available pattern calculation programs that can implement sidelobe level control do not provide calculations of the aperture efficiencies given different amplitude weightings. Calculation of the aperture efficiency can be somewhat confusing, particularly with regards to the difference between tapering via attenuation versus redistribution. The purpose of this paper is to define these terms, to provide a review of the proper normalization technique that is important in obtaining accurate aperture efficiency estimation. Descriptions of the amplitude tapers and their utility will be presented. A design example will be presented which will compare theoretical efficiencies with those obtained via Finite Element Method simulation. TABLE OF CONTENTS 1. 2. 3. 4. 5. 6.
12

1. INTRODUCTION
Through the use of individual amplitude and phase control, array antennas offer a wide range of directivity pattern shape implementations to the antenna designer. High directivity antennas have defined main beams whose widths are inversely proportional to their aperture extents. High directivity antennas also have sidelobes, which are often undesirable as they may permit reception of energy from undesired directions. The energy from the undesired directions may contain interfering sources such as multipath or even deliberate jammers. Techniques for reducing the levels of the peak sidelobes (those near the main lobe) are well understood. Near-in sidelobes may be reduced relative to the peak of the main beam by simply tailoring the amplitude distribution across the array aperture. The amplitude distribution is often referred to as the amplitude weighting. The following section will present a review of techniques for synthesizing amplitude weighting functions to achieve varied levels of sidelobe reduction. Use of these amplitude weighting functions have a wellknown effect on the peak of the main beam of the directivity pattern. The amplitude tapering for sidelobe reduction reduces the spatial efficiency (or aperture efficiency) of the antenna. Along with the reduction of peak directivity, amplitude tapering also results in a broadening of the main beam. Although not the focus of this discussion, the resultant change in beamwidth should be taken into consideration in the design process. While reduced sidelobes are usually desirable, reduced peak directivity is usually not, therefore an accurate understanding of the aperture efficiency is a valuable design tool. Some commercially available pattern calculation programs do not provide calculations of the aperture efficiencies given different amplitude weighting functions. Calculation of the aperture efficiency can be somewhat confusing, particularly 1

INTRODUCTION AMPLITUDE WEIGHTING FUNCTIONS DIRECTIVITY PATTERN CALCULATIONS NORMALIZATIONS AND APERTURE EFFICIENCIES A DESIGN EXAMPLE CONCLUSIONS

1 1

1-4244-0525-4/07/$20.00 2007 IEEE.

IEEEAC paper #1163, Version 2, Updated Dec 19, 2006

with regards to the difference between tapering via attenuation versus redistribution. The purpose of this paper is to define these terms, to present the proper normalization technique that is critical to obtaining accurate aperture efficiency estimations, and to validate this approach through a design example.

Taylor published his synthesis technique for linear/rectangular [2] and circular apertures [3] in 1955 and 1960, respectively. Figure 2 presents the same directivity pattern as that of Figure 1, except that a Taylor 25 dB n = 3 amplitude weighting has been employed to reduce the nearin sidelobes. The n parameter is used to define how many near-in sidelobes are held constant at the desired amplitude level. For further detail on this parameter refer to Taylor [23].

2. AMPLITUDE WEIGHTING FUNCTIONS

Equal illumination at every element in an array, referred to as uniform illumination, results in directivity patterns with three distinct features. Firstly, uniform illumination gives the highest aperture efficiency possible of 100% or 0 dB, for any given aperture area. Secondly, the first sidelobes for a linear/rectangular aperture have peaks of approximately 13.1 dB relative to the main beam peak; and the first sidelobes for a circular aperture have peaks of approximately 17.6 dB relative to the main beam peak. Thirdly, uniform weighting results in a directivity pattern with the familiar sinc(x) or sin(x)/x where x=sin() angular distribution, as shown in Figure 1. Uniformly illuminated circular apertures exhibit similar sidelobe structures based on Bessel functions, J1(x)/x.

Figure 2. The directivity plot of a linear array of eight elements with a Taylor 25 dB n = 3 amplitude for near-in sidelobe reduction from commercially available software. The prior two directivity plots are from a commercially available package (formerly EnsSyn from Boulder Microwave). The patterns presented are believed to be correct however note that the peak amplitudes of the two patterns are the same. They are both normalized to zero dB. As these patterns are not normalized to a consistent peak, this tool is of limited to no value in assessing the efficiency loss trade-off with sidelobe reduction levels.

Figure 1. The directivity plot of a linear array of eight elements with uniform amplitude from commercially available software. Since the 1940s, numerous researchers have contributed varying approaches for synthesizing amplitude distributions for the purpose of sidelobe reduction. A good summary of these techniques is presented by Hansen [1]. The amplitude synthesis approaches are known by their authors names and the most notable are Dolph-Chebyshev, Elliot, Taylor OneParameter, Taylor n , and Villeneuve n . For this discussion, we will use the Taylor distributions as they are arguably more commonly used for array antenna pattern synthesis. 2

3. DIRECTIVITY PATTERN CALCULATIONS

Even though the calculations are not overly complex, understanding the proper normalization technique can become confusing when viewing them from the perspectives as presented by Hansen [4] and Raffoul and Hilburn [5]. In this section, we will define aperture efficiency. We will also present graphically different amplitude distributions and their corresponding directivity patterns. As a review, Equation 1 presents the calculation for the voltage directivity pattern for a linear array of N elements of

isotropic radiators, where x is the inter-element spacing, and an is the amplitude of element n. Note that this equation is for the simplest array case of uniform phase for that of a broadside fixed beam array. These calculations can easily be extended for phase scanned cases.

E ( ) = a e
n =1 n

j(

nx sin )

(1)

The term uniform illumination is often used to describe the array amplitude distribution when the amplitudes at all of the elements are equal. If the voltage amplitudes all equal one Volt, the peak voltage, Epeak, for the ideal linear array of isotropic elements occurs when is zero and has a value given by Equation 2.

E Peak = a n
n =1

Figure 3. Plot of the uniform amplitude distribution for the eight element array. (2)

If a n =1 E Peak = 1 = N
n =1 N

For an array with uniform amplitude distribution, an = 1 Volt for all n, the EPeak equals the number of elements, N Volts. The corresponding power value is proportional to N2. The following series of plots were calculated using Equation 1, using software written in MATLAB. All of the following patterns assume an ideal array of eight elements with cos() element patterns and an inter-element spacing of 0.695 . The plots of Figure 3 and 4 show the uniform amplitude function and resulting directivity pattern for the eight element array. Note that the voltage peak (or peak directivity if each amplitude were truly 1 Volt) in Figure 4 is equal to the number eight in dB. Routines written to calculate the amplitude weighting coefficients for array sidelobe reduction usually provide the an for each element in Voltage form. Not always, but often, these routines are written to provide a maximum value of 1, as shown in the plot of Figure 5 the amplitude distribution of an array with a Taylor 25 dB n = 3 amplitude distribution.

Figure 4. Plot of the directivity pattern for the eight element array with uniform amplitude distribution.

Figure 5. Plot of the Taylor 25 dB n = 3 amplitude distribution

The Taylor Voltages calculated for this amplitude illumination function are (from the outer elements to the center) 0.3930, 0.5918, 0.8432, and 1.0000. Figure 6 presents the directivity pattern for the array with the amplitude distribution shown in Figure 5.

Obviously, the case where the amplitude taper is achieved through attenuation suffers increased losses. The attenuation efficiency, attenuation, of an array can be calculated using Equation 3. Note that this term only contains the inefficiency of the ohmic losses. Equation 3 gives a result in power form, but not in dB. Equation 3 also assumes the individual element weights, an, have a maximum value of 1.

attenuation =

(a )
n n =1

(3)

Figure 6. Plot of the directivity pattern for the array with the Taylor amplitude distribution of Figure 5. Notice that the peak gain of Figure 6 was reduced by approximately 2.5 dB, as compared to that of Figure 4. This might lead one to believe that the aperture efficiency for this sidelobe function was a full 2.5 dB. For some cases that is correct, however for many cases it is incorrect. There are two methods for implementing the amplitude tapers. The different methods result in widely different calculations of overall efficiencies.

The attenuation method, the one used for the plots of Figures 5 and 6, is analogous to achieving the amplitude taper by increasingly resistively attenuating the field energy for radiators toward the periphery of the array to achieve Voltages less than 1. As one might expect, this method is the least efficient, and the main beam gain loss is the greatest. This approach is further illustrated in Figures 7 and 8 where the amplitude distributions and resulting directivity patterns are presented for the range of Taylor weightings from 20 to 65 dB. The attenuation method predicts very significant main beam pattern losses. However, the losses observed are not only from the attenuation inefficiency. An additional loss of aperture efficiency is always incurred with both the attenuation and redistribution methods of achieving amplitude tapers.

4. NORMALIZATIONS AND APERTURE EFFICIENCY

As we introduced in the previous section, there are two physical methods for generating amplitude distributions for array antennas. Amplitude tapers can be created by either redistributing the power among the elements or by attenuating the power for the outer elements. With the attenuated method, power removed at the outer elements is attenuated in ohmic losses. This is the case in active arrays where switched attenuators are used to control the amplitude taper. With the redistribution method, the amplitudes are renormalized so that the sum of the tapered amplitudes is the same as the sum of a uniform illumination of the same number of elements. With the amplitude redistribution, total power radiated by the array is the same as the uniform case. Amplitude redistribution is usually employed in beam formers when a permanent amplitude taper is desired. 4

Figure 7. Plot of the amplitude distributions for Taylor functions of 20 to 65 dB using the attenuation normalization

aperture effectively becomes smaller as the energy is concentrated near the center of the array. .

Figure 8. Plot of the directivity patterns with amplitude distributions for Taylor functions of 20 to 65 dB using the attenuation normalization. These patterns suffer both attenuation and spatial aperture efficiencies. Note the very high main beam loss and low aperture efficiency In contrast to the attenuation method, significantly higher aperture efficiencies can be obtained by redistributing or renormalizing the energy within the amplitude distribution. This can be thought of as conservation of energy as any Voltage removed from outer elements is reallocated to more central elements. The redistributed normalization process can be effected by simply translating the average of the amplitude weights back to one using Equation 4. This is the same as dividing by the attenuation efficiency calculated in Equation 3.

Figure 9. Plot of the directivity pattern for the same configuration as in Figure 6 but calculated using redistribution normalization.
2

aperture

N an = = nN1 N ( an ) 2
n =1

(5)

n , redistributed normalized

a N a
n =1 n N n

(4)

With the redistribution normalization, the directivity pattern of Figure 6 is calculated to have an increased peak gain of approximately 2 dB, as shown in Figure 8. Figure 6 demonstrates both ohmic and spatial inefficiencies, yet Figure 8 demonstrates only the spatial inefficiency. Figure 9 presents the redistribution normalized amplitude functions for Taylor 20 through 65 dB tapers. Figure 10 presents the corresponding directivity patterns all of which demonstrate less loss and higher aperture efficiencies. The spatial aperture efficiency, aperture, can be calculated using Equation 5. This is the efficiency term that is incurred for every array that uses amplitude tapering, whether it is from attenuation or redistribution. This efficiency term is caused by the inefficient use of the available aperture. This loss can be thought of as directivity loss as the antenna 5

Figure 10. Plot of the amplitude distributions for Taylor functions of 20 to 65 dB using the redistribution normalization

these are to be implemented in a redistributed beam former. The array pattern synthesis method used to calculate the pattern and peak directivities must correctly renormalize the amplitudes to predict the appropriate losses. If the pattern synthesis either normalizes to the main beam peak or normalizes to a uniform distribution, incorrect peak loss values will be predicted and the efficiencies of the amplitude taper may not be accurately known.

5. DESIGN EXAMPLE
To demonstrate the aperture efficiency calculation approach, it was desired to compare the theoretical predictions of redistributed taper efficiency to full-wave solution models of linear array antennas. For this comparison, a Taylor 25 dB, n = 3 amplitude taper was again used. As shown in Figure 5, the element weightings, an, were calculated to be (from the outer elements to the center) 0.3930, 0.5918, 0.8432, and 1.0000. From Equation 3, the theoretical attenuated taper efficiency for this distribution is 2.56 dB. Likewise from Equation 5, the redistributed efficiency for this distribution is only 0.45 dB. The antenna architecture selected for the comparison was a resonant linear array of narrow-wall waveguide slots. This architecture was based on the prior art by Stevenson summarized by Johnson and Jasik [6]. The narrow-wall slotted waveguide array was designed to operate at 9.5 GHz using a standard WR-90 waveguide cross-section. The slots were spaced at 0.695 of a freespace wavelength. The amplitude redistribution of energy is implemented by varying the conductances of the individual slot radiators. The conductances can be easily modified by changing the tilt angles of the slots. This provided the authors a convenient method to implement both uniformly illuminated and amplitude tapered designs for direct comparison. The MATLAB-predicted uniformly illuminated linear array pattern of Figure 4 predicted an unrealistically high peak directivity of approximately 17.5 dBi. It was high because of assumptions made in the individual radiator pattern gain and in the omission of real world losses. The design example here includes significantly more real world effects via the use of full-wave Finitie Element Method (FEM) simulations. All of the following simulations were completed using the High Frequency Structure Simulator (HFSS), commercially available from Ansoft. Figure 12. A comparison of the aperture efficiencies between the attenuated and redistribution normalization methods. Note also that routines often used to calculate amplitude coefficients output values normalized to a peak of 1 Volt, as in the case of the distributions shown in Figure 7 -- even if 6 Figure 13 presents the HFSS CAD model of the eight element linear slotted waveguide array with uniform amplitude illumination. To achieve uniform illumination, all of the slot tilt angles are the same and they were calculated using the method described in [6] to be at 25 with respect to the vertical axis. As shown in the pattern presented in Figure 14, this antenna exhibited a peak gain of 14.93 dBi with peak sidelobe levels approximately 12 dB

Figure 11. Plot of the directivity patterns with amplitude distributions for Taylor functions of 20 to 65 dB using the redistribution normalization. This pattern suffers only the spatial aperture efficiency of the amplitude taper. Note the lower main beam loss and higher aperture efficiency as compared to the patterns of Figure 8. Figure 12 presents a comparison of the attenuation and aperture efficiencies. Regarding the methods of achieving the tapers, the attenuated method discards the energy while the redistribution method does not lose any energy because of deliberate attenuation. Note that the efficiency terms should be considered to reflect the array hardware to be implemented.

below the main beam peak. An ideal linear array would have peak sidelobe levels at 13.1 dB; therefore there are apparent amplitude and phase errors in the implementation of the model. These could be caused by several factors, including not having the radiators tuned exactly at resonance, not having the correct distance between the radiators, not having the correct tuned distance between the last radiator and the shorted waveguide end wall, and even not converging the FEM solution sufficiently.

Figure 15. The HFSS CAD model of the eight element linear array of narrow-wall waveguide slots with the Taylor 25 dB amplitude taper.

Figure 13. The HFSS CAD model of the uniformly illuminated eight element linear array of narrow-wall waveguide slots.

Figure 16. Plot of the directivity pattern of the waveguide array with the 25 dB Taylor taper simulated in HFSS. The theoretical spatial efficiency loss, aperture, calculated from Equation 5 was only 0.45 dB. The loss predicted by the FEM simulation was approximately 1.4 dB. If the sidelobe reduction had been effected via attenuation, the efficiency loss would have been greater than 2.5 dB. The differences between the ideal and the simulated pattern for the case of Figure 15 are similar to those for Figure 13. With the tapered case, however, several significant additional differences occur because of the differing slot tilt angles. Foremost, the individual element patterns are polarized across the narrow-wall slots, and the slots with increased tilt angle radiate more power into the orthogonal linear polarization. The directivity pattern data shown herein only presents the desired linear direction (referred to as the co-pol). The co-pol for this design is parallel to the long axis of the waveguide. Any power in the orthogonal polarization (referred to as cross-pol) is considered lost and is not included in these pattern comparisons. As a first 7

Figure 14. Plot of the directivity pattern of the uniformly illuminated waveguide array simulated in HFSS. Figure 15 presents the HFSS CAD model of the eight element linear slotted waveguide array with the 25 dB Taylor amplitude taper. To achieve the amplitude taper, four different slot tilt angles were employed. These angles were calculated and then simulated at 36, 29, 20, and 12 with respect to the vertical axis. As shown in the pattern presented in Figure 16, this antenna exhibited a peak gain of 13.53 dBi with peak sidelobe levels approximately 20 dB below the main beam peak.

order analysis, the co-pol amplitude of the slots tilted at 25 for the uniform case is reduced approximately 0.85 dB by the energy lost in the cross-pol. The co-pol loss increases with the slots tilted at 36 to 1.84 dB. This difference is approximately 1dB, the same order of magnitude as the discrepancy between the theory and the FEM design. It is believed that a combination of the slot tilt and the nonoptimized design account for the 1 dB difference between the theory of Equation 5 and the FEM model.

[6]

Johnson, R. C., and H. Jasik, Edistors, Antenna Engineering Handbook, 2nd ed. New York McGraw-Hill, 1984.

6. CONCLUSIONS
This paper presented a discussion of methods used to calculate the aperture efficiencies of array antennas. Equations to simulate basic array antenna patterns and methods for sidelobe suppression via amplitude tapering functions were reviewed. The concepts of attenuated and redistributed distributions were developed. Equations to calculate attenuation and spatial aperture efficiencies were presented. The importance of understanding and using the correct normalization techniques was discussed. Design examples of uniform and tapered amplitude linear arrays were presented and the full-wave analysis calculated patterns supported the theoretical efficiency equations.

REFERENCES
[1] Hansen, R. C., Phased Array Antennas, Wiley Series in Microwave and Optical Engineering, John Wiley & Sons, New York, 1998. Taylor, T. T., Design of Line-Source Antennas for Narrow Beamwidth and Low Side Lobes, IRE Transactions on Antennas and Propagation, January, 1955, pp. 16-28. Taylor, T. T., Design of Circular Aperture for Narrow Beamwidth and Low Sidelobes, IRE Transactions on Antennas and Propagation, January, 1960, pp. 17-22. Hansen, R. C., Aperture Efficiency of Villeneuve n Arrays, IEEE Transactions on Antennas and Propagation, Vol AP-33, No. 6, June 1985, pp. 666-669. Raffoul, G. W., and J. L Hilburn, Radiation Efficiency of an X-Band Waveguide Array, IEEE Transactions on Antennas and Propagation, March 1974, pp. 355-357.

[2]

[3]

[4]

[5]

BIOGRAPHIES
Glenn D. Hopkins is a Senior Research Engineer in the Sensors and Electromagnetic Applications Laboratory of the Georgia Tech Research Institute. In his 19 years with GTRI, Mr. Hopkins has studied and developed a wide range antenna and RF circuit technologies. His primary research interests involve array antennas and unique beam forming and pattern control architectures. He has coauthored 15 refereed publications and two U.S. Patents. Justin T. Ratner is a cooperative education, undergraduate student at Georgia Tech and is employed by the Sensors and Electromagnetic Applications Laboratory of the Georgia Tech Research Institute. In his 2 semesters with GTRI, Mr. Ratner has helped develop slotted waveguide arrays, terrain and atmospheric simulations, and radar beam steering algorithms. Anya Traille obtained the B.S. Degree in electrical engineering from Georgia Institute of Technology, Atlanta, GA in the Fall of 2004. She started her Masters Degree in electrical engineering in Fall 2006 and is currently a Research Engineer at GTRI SEAL and also works for the Georgia Electronic Design Center where her current research focus is on 3D Integration and Nanostructured Antennas. Victor K. Tripp is a Principal Research Engineer in the Sensors and Electromagnetics Applications Laboratory of the Georgia Tech Research Institute. He has 35 years of experience in antennas and electromagnetics, 27 of those years with GTRI. Mr. Tripp has specialized in vehicle-mounted antennas, broad-band lowprofile antennas, and antenna errors. He is coauthor of 48 refereed conference and journal papers, and co-inventor on 7 patents. Mr.Tripp is currently working with antenna optimization by genetic algorithms coupled with the method of moments.