2.2 2.4 GHz Phased Array

(1)

2.2-2.4 GHz Phased-Array Conceptual

Redesign

by

Russell Hofer

May 2005

(2)

(3)

ACKNOWLEDGEMENTS ... 1

INTRODUCTION... 2

BACKGROUND OF EXISTING ANTENNA ARRAY... 2

REDESIGN REDESIGN OBJECTIVESOBJECTIVES ... ... ... 66 INCREASE ANTENNA BANDWIDTH... 7

INCREASE ANTENNA BANDWIDTH... 7

Increase of Antenna Element’s Bandwidth... 7

Increase Increase of of Bandpass Bandpass Filters’ Filters’ BandwidthBandwidth ... ... 1313 CONCEPTUAL REDESIGN CONCEPTUAL REDESIGN OF ANTENNA ELEMENOF ANTENNA ELEMENT PCBT PCB ... 18... 18

Steering Steering RedesignRedesign ... ... 1818 Schematic Redesign... 24

Schematic Redesign... 24

TRANSMIT TRANSMIT CAPABILITIESCAPABILITIES ... ... 2929 TRANSMIT TRANSMIT CAPABILITIESCAPABILITIES ... ... 3030 CONCLUSIONS CONCLUSIONS ... ... 3939 FURTHER WORK... 40 FURTHER WORK... 40 REFERENCES... 41 REFERENCES... 41 APPENDICES... 42 APPENDICES... 42 Appendix I:

Appendix I: Matlab CodeMatlab Code ... ... 4242

Figure

Figure 9...9... 4242 Figure

(4)

A

Ac

ck

kn

no

ow

wlled

edg

gem

emen

ents

ts

I would like to

I would like to acknowledge Dr. Christopher Allen for his guidance through this project.acknowledge Dr. Christopher Allen for his guidance through this project. In addition, I would like to

In addition, I would like to thank Mr. Dan Depardo thank Mr. Dan Depardo for his willingness to answer questionsfor his willingness to answer questions and share his valuab

and share his valuable experience of antennle experience of antenna and microwave designa and microwave design. . Lastly, Dr. Jim StilesLastly, Dr. Jim Stiles for consultation on general microwave principles and providing what has proven to be for consultation on general microwave principles and providing what has proven to be aa very useful course in microwave

very useful course in microwave design.design.

I would like to extend a thanks to Dr.

I would like to extend a thanks to Dr. Christopher Allen, Dr. Prasad Gogineni, andChristopher Allen, Dr. Prasad Gogineni, and Dr. Kenneth Demarest for participation in my committee.

Dr. Kenneth Demarest for participation in my committee.

Last, but not least, I would like to

Last, but not least, I would like to extend my gratefulness to my wife, Kristie Hofer, for herextend my gratefulness to my wife, Kristie Hofer, for her support and patience.

(5)

Introduction

In 2004, Kansas University and Honeywell FM&T

In 2004, Kansas University and Honeywell FM&T Kansas City Plant jointly built a Kansas City Plant jointly built a 2.3-2.42.3-2.4 GHz Phased-Array Protot

GHz Phased-Array Prototype. ype. Mr. Dan Depardo, RF electronMr. Dan Depardo, RF electronics engineer, was respoics engineer, was responsiblensible for the design, constructi

for the design, construction, and preliminary testion, and preliminary testing. ng. The antenna has proven to performThe antenna has proven to perform adequately per its or

adequately per its original design criiginal design criteria. teria. However, Dwayne Brown, KCP (KanHowever, Dwayne Brown, KCP (Kansas Citysas City Plant) product engineer, has extended requirements that are

Plant) product engineer, has extended requirements that are to be incorporated into theto be incorporated into the

second design iteration

second design iteration. . In this report, the results of tIn this report, the results of the conceptual redesign of the antennahe conceptual redesign of the antenna array are discussed.

array are discussed. Included is a brief backgrouIncluded is a brief background of the existing antenna array, redesind of the existing antenna array, redesigngn objectives, new antenna

objectives, new antenna design, simulation results, and further work/conclusions.design, simulation results, and further work/conclusions.

Background

Background of Existi

of Existi ng Antenna

ng Antenna Array

Array

The following discussion gives a limited overview of the

The following discussion gives a limited overview of the original antenna array design.original antenna array design.

The interested reader is r

The interested reader is referred to [1] foeferred to [1] for a detailed overview. r a detailed overview. The original antennThe original antennaa consists of eight patch elements po

consists of eight patch elements positioned in a linear array. sitioned in a linear array. The array is capable ofThe array is capable of performing electro

performing electronic steering in azimutnic steering in azimuth. h. The bandwidth oThe bandwidth of the antenna encompf the antenna encompasses 2.3-asses 2.3-2.4 GHz.

2.4 GHz. Figure 1 shows an iFigure 1 shows an image of the originmage of the original 2.3-2.4 GHz phasedal 2.3-2.4 GHz phased-array antenna-array antenna prototype [1

prototype [1].].

Figure 1:

2.3-2.4 GH

2.3-2.4 GHz Phased-Ar

z Phased-Array Antenn

ray Antenna Prototype

a Prototype

Each patch antenna element consists of a

Each patch antenna element consists of a low-noise RF amplifier, two-stage RF bandpasslow-noise RF amplifier, two-stage RF bandpass filter (100 MHz B/W),

filter (100 MHz B/W), two variable voltage phase shifters, and two variable voltage phase shifters, and further small-signal RFfurther small-signal RF amplification.

amplification. Figure 2 shFigure 2 shows a basic bows a basic block diagram lock diagram of the antennof the antenna element. a element. Notice thatNotice that each antenna element must be given a steering signal.

each antenna element must be given a steering signal. This signal controlThis signal controls the amount ofs the amount of

phase shift incu

(6)

Output

Patch Antenna

5

th

-Order

BPF (Bandpass Filter)

Φ

₁

Φ

₂

Small-Signal Amplifier

SC (Steering Control Voltage)

LNA (Low Noise Amplifier)

Phase Shifters

3

rd

_-Order

BPF (Bandpass Filter)

(7)

Ideally, each antenna element will have a constant delta phase shift with respect to its neighboring elements given by Equation 1 [2]:

0 cos * / 2π λ θ θ = _d ∆ ₍₁₎ d = interelement spacing λ = wavelength θ0= steering angle

For a broadside beam, there is no relative phase shift per element. The steering control voltage, depicted in Figure 2, is generated by an analog op-amp array. This circuit generates a maximum delta steering voltage between elements of approximately 1 volt. Assuming the phase shifters’ behavior is roughly linear, this results in a maximum delta phase shift between elements of approximately 30 degrees. Also, the distance between

elements with respect to wavelength is approximately 0.23. Using this information in Equation 1, the maximum obtainable steering angle is +/- 20 degrees from broadside. The signals from all antenna elements are combined with two RF power combiners and a

180o hybrid coupler. These signals, one given the designation ‘SUM RF’ and the other

‘DELTA RF’ are routed onto the RF Signal Processor board. Additional functions of the RF signal processing board are to provide additional variable gain, sum level, and delta level detection as follows. Both the ‘DELTA RF’ and the ‘SUM RF’ signal are further amplified with a variable gain amplifier. Both signals are then sampled using a

bi-directional coupler. The sampled signal is used for level detection. The detected signals are given the name ‘SUM LEVEL’ and ‘DELTA LEVEL’. The purpose of the ‘SUM LEVEL’ and ‘DELTA LEVEL’ signals are to determine the strength of the received signal and the direction of the transmitter. The ‘DELTA LEVEL’ signal has the desirable

attribute that it provides a sharp null when the antenna is steered in the direction of the transmitter. This allows for accurate tracking. The RF signal processing board interacts with a user supplied digital control system as depicted in Figure 3.

Figure 3 shows an overview of the original system. As can be seen from Figure 3, the user supplied digital control system provides the RF signal processor with a steering control voltage. The RF signal processor subsequently provides each element with a steering control voltage. The control voltage is directly applied to each element’s set of phase shifters. The digital control system is also responsible for interpreting the ‘SUM LEVEL’ and ‘DELTA LEVEL’ signals. Based upon its interpretation, it may vary the gain of the RF Signal Processor’s variable gain amplifier and/or adjust the steering control.

(8)

RF Signal Processor

Digital

Control

System

“user

supplied”

RF Signal to S-band Receiver

Steering Control (1.0 – 8.0 V)

Gain Control (VGA) (0-2.5V) Delta Level (270mV – 3.0V) Sum Level (270mV – 3.0V) Antenna Array

(9)

Redesign Objectives

The following list details the redesign objectives: • Maximize antenna gain / efficiency

• Increase antenna bandwidth to 200 MHz covering 2200-2400 MHz • Maximize steering angle

• Variable beamwidth (controllable) • Increase dynamic range of antenna

• Add dual mode capabilities – transmit & receive at 10 W • Transmission and reception must operate in duplex mode

The following list details personal objectives to be gained at the end of this project: • Expertise in antenna design

• Learn antenna design CAD software (HFSS) • Learn RF microwave design software (ADS)

• Perform simulations in both antenna design software and RF microwave design software

• Increase familiarity with RF microwave design terminology

Initially, the following schedule was adopted to perform the described objectives:

Research and determine best CAD software to use, obtain license if Oct 2004

needed

Attend course to learn CAD software, pending availability of course Dec 2004

Attend course to learn RF microwave design software Dec 2004

Literature search and review of the topic Feb 2005

Modify element design to include increased bandwidth April 2005

Modify array design & create transceiver design May 2005

(Deliverable is a proof of concept / virtual prototype; conceptual schematic & geometry)

(10)

Increase Antenna Bandwidth

In order to increase the antenna bandwidth, two things must be realized. The antenna patch element bandwidth must operate at 2200-2400 MHz; all electronic components must satisfy the increased bandwidth requirement. This means that the bandwidth of both the first and second stage bandpass filters must be increased.

Increase of Antenna Element’s Bandwidth

Each antenna element consists of a single layer patch antenna. A number of different

approaches are available to increase the bandwidth of patch antennas. The most

common involve the use of cavities and s substrate techniques [3]. However, these techniques increase the complexity, size, and weight of the design. The simplest appro to increase the height of the substrate. Th drawback to this approach is that if the heigh is made too tall, an increase in the production of surface waves will decrease the antenna’s efficiency. tacked ach is e t [4]

Figure 4:

Patch Geometry

The strategy to be employed is to moderately increase the height of the patch antenna in order to achieve the desired increase in

bandwidth. Figure 4 shows the geometry of a typical patch antenna. Figure 4 is close to the original configuration, with the exception that the patch antenna was fed using a coaxial probe. More discussion with regards to

Figure 4 will be given in the proceeding paragraphs.

The basic design equations used are given by [5]. A brief overview of the design strategy is included in this report. For more detailed information, the interested reader is referred to

[5]. The model used throughout the design process consists of the 1st order

transmission-line model. Some of the concepts of the cavity model are also deployed. The

transmission-line model assumes that the patch antenna may be treated as a transmission line with an open circuit at each end of the transmission line. These ‘open circuits’ are the radiating apertures of the patch antenna. Since the end apertures radiate and subsequently are lossy, this is only a rough approximation. The length of the patch antenna is chosen so that the transmission line transforms the total admittance of the open circuit ends to a real impedance. It can be shown from transmission line theory that this is approximately λ/2. It is actually somewhat less to account for fringing/radiating effects.

(11)

With the correct patch length, a real impedance is obtained. Now, the real input impedance can be matched to the line impedance feeding the patch. In the case of a coaxial feed, this is best illustrated by observing Figure 4. When the patch is in resonance, the electric field will be as depicted. As can be seen, the E-field is a maximum at the two edges of the patch antenna. The current will be close to zero at the two edges of the patch and a maximum at the center. Consequently, the impedance will be close to 0 at the center and relatively large at the ends. It can be shown that an approximate relation is given by the following

equation [5]: ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ = = = ₎ ₍ ₀₎_*_cos2 ₀ ( y L y Rin yo y Rin π (2)

Thus, if one moves the coaxial feed lengthwise with respect to the patch, a resonance point may be found. The design strategy employed is as follows. Initially, the patch height is increased. The coaxial feed is centered along the width of the patch antenna. The coaxial feed is moved lengthwise along the patch until resonance is achieved at any arbitrary frequency. The patch length is adjusted to achieve resonance at the desired frequency of 2.3 GHz. An iterative procedure is necessary, as second order effects relate all design parameters; for example, adjusting the length of the patch antenna will slightly alter the

required feed location.

In the original design, the patch height was set to 0.15”. The height was chosen due to the availability of excess material from another research project. The original dielectric

material was Rogers Dielectric 5870, with a relative dielectric constant of 2.33. Rogers Corporation only carries standard sizes up to 0.125” thick. After contacting the

manufacturer, they were able to verify availability of Dielectric 5880, 0.25” thick. This material has a slightly lower loss factor compared to 5880, with a relative dielectric constant of 2.2. It was decided that this would be a good candidate for the design. HFSS was used to model the

antenna’s performance. The original model was

developed using an example from [6]. The design model is depicted in Figure 5. The initial model used

idealizations in order to improve simulation speed. These assumptions were later relaxed during final

simulations. Assumptions include all conductors being

(12)

PEC (perfect electric conductors), the ground plane being infinite in extent, and a reduced size radiation boundary. The simulated antenna was totally enclosed in an air box, as

depicted in Figure 5. The walls of the box are assigned as the radiation boundary. A larger radiation boundary will yield more accurate results. According to HFSS documentation, the following requirements should be met for the radiation boundary to accurately simulate open space: 1) the radiation boundary must be greater than ¼ wavelength from any

radiating surface; 2) boundary orientation must be set perpendicular to incident radiation. These requirements were met to a fair degree; during final prove-in of the simulation, they were improved.

The patch height was increased to 0.25” using Dielectric 5880 as the substrate. The coaxial feed was centered along the width of the patch antenna. A failed attempt was made to move the coaxial feed lengthwise along the patch to reach resonance. After a great deal of consideration, it was determined that the increased probe length was introducing too much inductance into the feed. See Figure 6.

Probe Inductance Probe Inductance

E uivalent Circuit

Figure 6:

Probe Inductance

Two simple methods are available to counteract the introduced inductance. The first method is that a gap may be introduced between the probe and the patch. This will add a capacitance that negates the inductive reactance at the design frequency. Second, the probe width may be increased. This minimizes the inductance of the probe. The first method was too sensitive to design deviations. The second technique was found to be adequate. It was verified that a probe radius of 0.15 cm could achieve a reasonably good 50-ohm match. A probe radius of 0.15 cm corresponds to a 3 mm wide probe. This may be difficult to solder onto the patch due to heat dissipation, but construction should still be feasible. In addition to increasing the height, it can be shown that increasing the patch width lends towards an increase in bandwidth. Unfortunately, it was experimentally determined that an increase in patch width also increases inductance. Consequently, the patch width was also reduced from 6.239 cm to 6 cm in order to minimize inductance to achieve an acceptable impedance match.

The final patch dimensions can be summarized as follows. The dielectric height is 0.25”; the patch width is 6 cm; the probe radius is 0.15 cm; the patch length is 3.913 cm. The patch is fed using an inset of 0.75 cm from the edge (this is the distance from the patch’s edge to the radial center of the coaxial feed). The final simulation was made more

realistic by using the patch’s true conductor (finite thickness copper). The simulation volume was also expanded to include ~2 wavelength width radiation boundary (compared

(13)

to ¼ wavelength in original simulation). The ground plane was reduced from being infinite in extent. The resulting simulation took ~30 minutes per iteration. The simulated results were not drastically different from the ideal simulations. We now proceed to describe the simulation results.

Figure 7 shows the impedance match as a function of frequency vs. VSWR.

Figure 7:

Antenna VSWR

As can be seen from Figure 7, the resonant frequency is centered at 2.3 GHz. The

2:1 VSWR band extends from 2.2-2.4 GHz. This will allow ~ 90% of the incident power to be transmitted into a 50-ohm matched transmission line.

Figure 8 shows a 3D plot of the radiation pattern in dB. As to be expected, the beam pattern is fairly symmetrical with respect to the z-axis. There is ~5 dB attenuation at ~60

degrees compared to the field calculated at broadside. This field is calculated with 1 W of incident power on the antenna’s input port.

(14)

In order to obtain a better estimate of the antenna’s expected performance in a receiver, the realized gain is plotted vs. theta for phi=0,90 degrees. These results are shown in Figure 9. The realized gain is defined as follows:

Pincident U gain

realized _ =4π * ₍₃₎

U = the radiation intensity in the specified direction (watts per steradian) Pincident = the total incident power on antenna’s input port (watts)

This definition of gain indicates the actual power/steradian radiated compared to that of an ideal lossless isotropic radiator. The realized gain pattern shown in Figure 9 does not

include coupling effects between elements. Figure 7 & 8 were also created with no regards to coupling effects. Results shown in Figures 7, 8, and 9 will differ from actual results due to the presence of other neighboring antenna elements. Coupling effects are dependent upon frequency as well as the scan angle of the patch antenna [9]. It is possible that

coupling effects could reduce the realizable steering angle of the patch antenna array. We will briefly consider the effects of coupling as follows. The patch elements are positioned collinearly in the E-plane. The distance between patch edges is ~3.6 cm. This yields a normalized edge separation with respect to wavelength of ~0.27. According to [5, p 765],

this will yield an |S12|2 of ~-20 dB. This result shows that coupling effects are not expected

to be a dominate factor. All subsequent results assume that coupling effects may be ignored. However, it is recommended that a simulation be performed to verify that coupling effects do not significantly alter the obtained results. A simulation should be performed to test the complete antenna array performance while the array is steered both at broadside and at a maximum steering angle.

(15)

‘Broadside’

Radiation Pattern of One Element, psi = 0, 90 degrees

Figure 9:

Realized Antenna Gain

In order to more easily model the gain of the system in subsequent discussion, we will proceed to calculate an approximate relationship for the antenna gain with respect to θ. A

good first order approximation for gain will have the following form [4, p 213]: ))

sin( * cos(

max*

_ gain alpha theta

realized = (4)

alpha=constant

max=maximum realized gain

Curve fitting to Equation 4 was accomplished by setting the maximum value to 6.32 dB as read from Figure 9. In order to obtain alpha, an arbitrary reading from θ=60 degrees was used. Figure 10 was obtained by plotting the described expression in Matlab (see

(16)

2 4 6 8 30 210 60 240 90 270 120 300 150 330 180 0

Approximate Radiation Pattern of One Patch Element

‘Relative Orientation of Element’

‘Broadside’ 3rd -Order Filter 5th-Order Filter Figure 11:

PCB Layout

Figure 10:

Approximate Antenna Gain

Comparison of Figure 9 and Figure 10 reveal that this approximation is reasonable. This approximation will be used in subsequent discussion of the overall system performance to include the effects of the array.

Increase of Bandpass Filters’ Bandwidth

In order to increase the antenna bandwidth, all electronic components must satisfy the increased bandwidth requirement. This means that both the

first and second stage bandpass filters must be modified.

The original PCB layout contains two bandpass

filters. The first bandpass filter is of 3rd -order,

placed directly after the antenna input, while the

other bandpass filter is of 5th-order directly after

the low noise amplifier. It might be noted that placement of a filter before the low noise amplifier

(17)

of the 3rd -order filter is to attenuate signals in the 2400-2500 MHz band (802.11 WLAN traffic) to an acceptable level.

This type of filter is referred to as an Interdigital Connect Filter. The interested reader is referred to [7] for a more detailed discussion of this filter type.

m1 freq= dB(S(2,1))=-5.142 2.330GHz 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 1.0 3.0 -45 -40 -35 -30 -25 -20 -15 -10 -5 -50 0 freq, GHz d B ( S ( 2 , 1 ) ) m1 B on n. It Similar

he original filter design was made using a

. p,

Figure 12:

5

th

Order Filter

R

The first design strategy was to use the built-in wizard contained within ADS design software. The process involves specifying both the upper and lower pass and stop bands. The maximum allowable passband attenuation and the minimum allowable stopband

attenuation are also specified. The filter order may be specified; if it is not, the wizard will calculate the minimum order needed to meet the requirements. Lastly, the response type may be specified as maximally flat, or Chebychev. An example of the best obtainable output for a

5th-order filter is presented in Figure 12. This result was obtained after considerable tuning

of the input parameters. As can be seen, the resulting filter is not acceptable. There is a 5 d notch in the center of the passband. The reas for this unacceptable behavior is unknow appears that Agilent may not have thoroughly tested this design guide for robustness.

results were obtained with the 3rd -order filter.

T

product from Eagleware. Unfortunately, this software was unavailable for immediate use. An alternative solution was to make use of optimization features contained within ADS This solution proved to be a bit tedious to setu but once setup it is extremely flexible and

(18)

Figure 13 shows the optimization circuit for a 5th-order filter. As can be seen, there are a total of four goal objects. The goal objects make reference to ‘stop1’,’stop2’,’avg’, and ‘delta’. ‘Stop1’ refers to the stopband attenuation found at the lower stopband frequency. ‘Stop2’ refers to the stopband attenuation found at the upper stopband frequency. ‘Avg’ refers to the average attenuation within the passband. ‘Delta’ refers to the

maximum-minimum passband attenuation. Specific definitions of these variables are defined underneath the Meas Eqn object in the lower right hand corner. The optimization

object, ‘Optim’, makes references to all of the defined goal objects, determines the search algorithm, weighting, and other options. All goals are each assigned an acceptable level. For instance, if the lower stopband is less than 40 dB, no penalty is incurred. If all goals are found to be within their acceptable level, ‘perfect’ optimization is achieved and optimization stops. In most situations, the designer is trying to obtain the ‘best’ possible solution with conflicting goals; consequently, acceptable levels may be set in such a way that the

optimization does not reach an optimal value. In this scenario (this also describes the current filter optimization), the ability to weight goals relative to their importance level is a critical feature. This was taken advantage of in order to obtain a steep rolloff in the upper stopband

edge of the 3rd -order filter (to remove as much 802.11 interference as possible). When the

optimization cannot reach a ‘perfect’ optimization, the optimizer may be set to a maximum number of iterations. Lastly, another interesting feature of ADS’s optimization is the ability to see the results within a data display in real-time as the optimizer performs iterations. This facilitated troubleshooting/tuning of parameters to obtained the desired results.

Figure 13:

5

th

-Order Optimization Circuit

Goal OptimGoal4 RangeMax[1]= RangeMin[1]= RangeVar[1]= Weight= Max=-40 Min= SimInstanceName="SP1" Expr="stop2" GOAL Goal OptimGoal3 RangeMax[1]= RangeMin[1]= RangeVar[1]= Weight= Max=-40 Min= SimInstanceName="SP1" Expr="stop1" GOAL MLSUBSTRATE2 Subst1 LayerType[2]=ground LayerType[1]=signal Cond[2]= 1.0E+50 T[2]=34 um Cond[1]= 1.0E+50 T[1]=34 um TanD=0.0021 H=20 mil Er=3.38 Diele ctric -i : ER[i], H[i], TAND[i] Meta l-i : T[i], COND[i], TYPE[i]

M e t a l - 2 M e t a l - 1 D i e l e c t r i c - 1 MeasEqn Meas1 stop2=dB(S(2,1)[freq_LSB]) stop1=dB(S(2,1)[freq_USB]) avg=mean(dB(S(2,1)[freq_lower::freq_upper])) delta=abs(max_dev-min_dev) max_dev=max(dB(S(2,1)[f req_lower::freq_upper])) min_dev=min(dB(S(2,1)[freq_lower::f req_upper])) freq_LSB= find_index(freq,2.1e9) freq_USB=find_index(freq,2.5e9) freq_upper=find_index(freq,2.4e9) freq_lower=find_index(freq, 2.2e9) Eqn Meas Goal OptimGoal2 RangeMax[1]= RangeMin[1]= RangeVar[1]= Weight=1 Max= Min=-1 SimInstanceName="SP1" Expr="avg" GOAL Goal OptimGoal1 RangeMax[1]= RangeMin[1]= RangeVar[1]= Weight=1 Max=1.5 Min= SimInstanceName="SP1" Expr="delta" GOAL S_Param SP1 Step=0.02 GHz Stop=40 GHz Start=2.0 GHz S-PARAMETERS Optim Optim1 UseAllGoals=yes UseAllOptVars=yes SaveAllIterations=no SaveNominal=yes UpdateDataset=yes SaveOptimVars=yes SaveGoals=yes SaveSolns=yes Seed= SetBestValues=yes NormalizeGoals=yes FinalAnalysis="SP1" StatusLevel=10 DesiredError=0.0 MaxIters=999 OptimType=Random OPTIM Term Term2 Z=50 Ohm Num=2 Term Term1 Z=50 Ohm Num=1 VAR VAR1

L2=660 mil opt{ 550 m il to 700 mil } L1=126 mil opt{ 80 m il to 160 mil } S2=54 m il opt{ 20 m il to 80 mil } S1=43 m il opt{ 20 m il to 80 mil } W1= 37 mil opt{ 20 mil to 120 mil }

EqnVar ML5CTL_V CLin1 ReuseRLGC=no RLGC_File= Layer[5]=1 Layer[4]=1 Layer[3]=1 Layer[2]=1 Layer[1]=1 W[5]=W1 S[4]=S1 W[4]=W1 S[3]=S2 W[3]=W1 S[2]=S2 W[2]=W1 S[1]=S1 W[1]=W1 Length=L2 Subst="Subst1" ML5CTL_V CLin2 ReuseRLGC=no RLGC_File= Layer[5]=1 Layer[4]=1 Layer[3]=1 Layer[2]=1 Layer[1]=1 W[5]=W1 S[4]=S1 W[4]=W1 S[3]=S2 W[3]=W1 S[2]=S2 W[2]=W1 S[1]=S1 W[1]=W1 Length=L1 Subst="Subst1"

(19)

The results obtained for the 5th-order filter after optimization are shown in Figure 14. For

brevity, the 3rd -order optimization results are not shown (final simulation results for both

filters will be included at the end of this section). As can be seen from Figure 14, the maximum dip in the passband is 1.5 dB—a dramatic improvement over that obtained with the wizard. Also, the rolloff of the filter is steeper.

As a final verification o erformed using

omentum. Momentum is a 2.5d RF electromagnetic simulator packaged with ADS. This

slightly lengthened to compensate.

The final simulation results of both the 3rd

-order and 5th-order filters are given by

Figures 16 and 17, respectively. These results are acceptable, and this completes the discussion of the filter design.

m1 freq=

dB(S(2,1))=-1.52

f the filter design, a more accurate simulation was p M

allows for relatively quick simulation of the designed filters by importing the filter design

into Momentum. Figure 15 shows an example layout of the 5th-order filter imported into

Momentum. Some necessary changes were made to make the layout realizable; for

example, the ground vias were made circular. As a result, the lengthwise dimensions had to be 2 2.330GHz 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 1.0 3.0 -60 -50 -40 -30 -20 -10 -70 0 freq, GHz d B ( S ( 2 , 1 ) ) m1

Figure 14:

5

th

-Order Filter Response using ADS

Optimization

(20)

S21

Frequency

Figure 16:

3

rd

-Order Filter Frequency

Res onse in Momentum

2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.0 2.8 -50 -40 -60 0 -30 -20 -10 M a g [ d B ] 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.0 2.8 -60 -50 -40 -30 -20 -10 -70 0 Frequency Ma g. [dB ] S22

Figure 17:

5

th

-Order Filter Frequency

Response in Momentum

Frequency M a g [ d B ] S21

(21)

Concept

design of Antenna Element PCB

Steering Redesign

The following design objectives will be addressed in this sub-section:

• Maxim ering angle

• Variable beamwidth (controllable)

The original configuration obtained steering through the use of two phase shifters. Each phase shifter only allows for a maximum phase shift ~110 degrees. This yields a total

maximum phase shift of 220 degrees. To increase the steering angle, a larger phase shift is needed. Upon consultation with the manufacturer, SVMicrowave, by utilizing three phase shifters in series, a full 360 degrees of phase shift should be obtainable. The only sacrifice

being made is an extra i ti

Another way to maximiz rol the phase shift.

The original configuration used uit in order to control each

phase shifter’s phase shift. This lim

result, the maximum steering angle is ~20 degrees. Also, the analog control assumes that the delta phase shift is linear with respect to voltage. Unfortunately, the linearity of the phase shifter’s voltage response curve is mediocre at best (see [1] for details). In order to

overcom ility in beam control, the phase shifters in

the new design will be individually control the digital control system. This will

allow for digital linearity compensation and added flexibility via arbitrary phase control. The second objective listed is to add variable beamwidth control. In order vary the

beamwidth, a tape tribution of weights is required. (Each weight is multiplied by the

incoming signal of each element, and then summed together.) The comparison made in most antenna textbooks is that a tapered distribution is mathematically analogous to windowing of a Fourier series. Therefore, a tapered distribution will have a wider

beamwidth and lower sidelobes. The original configuration only changed the phase of each received signal, with no tapering. Tapering of the weights is to be obtained in the new configuration through the utilization of variable gain amplification on each antenna element. These will be controlled by the digital control system. Figure 18 shows the modified block diagram of the antenna element’s PCB to include an additional phase

shifter, digital control of each phase shifter, and variable a ifi on. Figure 19

shows a top system level diagram. La nce, Figure 20 shows a block diagram

of the RF signal processor.

stly, for refere

cati mpl

gain red dis

e these weaknesses, and add more flex le

ib d by

e the steering angle is to more accurately cont an analog voltage divider circ

ited the maximum delta phase shift to ~30 degrees; as a on loss.

nser 1.3 dB of

ize ste

(22)

Figure 18:

Block Diagram of a Modified Antenna Element

Φ

₁

Φ

₂

Φ

₃

SC (Steering Control Voltage)

VGA (Variable Gain Amplifier)

Phase Shifters

BS (Beam Shaping Voltage)

LNA (Low Noise Amplifier)

Output

5 -Order

BPF (Bandpass Filter)

th r

3

d

-Order

BPF (Bandpass Filter)

Patch Antenna

(23)

RF Signal Processor

Digital

Control

System

“user

supplied”

Steering Control Bus (0-10V) 8

Beam Shaping Bus (0-2.5V)

8

Gain Control (VGA) (0-2.5V)

Delta Level (270mV – 3.0V)

Sum Level (270mV – 3.0V)

(24)

Element Circuit #1 SC1 BS1 Antenna Element Circuit #2 SC2 BS2 Antenna Element Circuit #3 SC3 BS3 Antenna Element Circuit #4 SC4 BS4 Antenna Element Circuit #5 SC5 BS5 Antenna Element Circuit #6 SC6 BS6 Antenna Element Circuit #7 SC7 BS7 Antenna Element Circuit #8 SC8 BS8

4 - i

n

p

u

t

p

o

w

e

r

c

o

m

b

i

n

e

r

Power

Combiner

“Σ”

Hybri

Degree

d

180 “∆”

R

F

P

r

e

s

o

c

i

n

g

C

i

r

c

u

i

t

VGA

evel Delta Level Sum L

Key:

SC = Steering Control

BS = Beam Steering

VGA = Variable Gain Amplification

Figure 20:

Block Diagram of RF Signal Processor

4 - i

n

p

u

t

p

o

w

e

r

c

o

m

b

i

n

e

r

(25)

We are now in a position to simulate the generated beam pattern obtained from both the

ante t and the weighted summer. It can be shown that both the element beam

pattern and the array beam pattern may be multiplied together to form the composite beam pattern. In the simulations to follow, the array beam pattern has been normalized to yield

0 dB in the direction of arrival. One may envision a constant 9 dB added to the simulated

plot to take int unt the array gain associated with adding all elements

toge would de the overall antenna gain with respect to an isotropic antenna.

Nevertheless, the follow ots accurately show the shape of the beam pattern. The first

simu se a tapered distribution. The beam pattern will be calculated at

steering angles (relative adside) of 0, 30, 60, and 90 degrees. Figure 21 shows the

results. As can be seen f igure 21, when the beam is steered to 90 degrees, the

elem patterns starts to attenuate the composite gain. Also, the peak radiation is

not at the steering angle since the element’s beam pattern is strongly distorting the

com pattern. For comparison, a maximum gain of 2.1

stee f 60 deg The actual realized steering angle is 48 degrees. On the other

hand, a maximum of gai 6.25 dB is achieved for a steering angle of θ=0 degrees

(relative to broadside). Values for all an bulated in Table 1.

nna elemen s in order ther. This o acco scribe ing pl to bro rom F rees. n of ~ lation will not u

ent’s beam posite beam

ring angle o

dB is achieved for a

gles are ta

‘Relative Orientation of Phased Array’ A

B

pproxima diation Pattern of Composite

eam at V Steering Angles

te Ra arious Broadside’ ‘

Fi

Simulated gain for various steering angles, using a

uniform distribution

(26)

The second simulation will use a tapered distribution. The tapered distribution to be used is the Kaiser distribution, with Beta=3 [8, p 107]. A larger value of beta will increa

main beamwidth and decrease the sidelobe level. As in Figure 17, the beam pattern wil calculated at steering angles (relative to broadside) of 0, 30, 60, and 90 degrees. Figure 22 shows the results. As one would expect, the element’s beam pattern still distorts the ar beam pattern. By comparison to Figure 21, the beamwidth has widened. lso, one

bserves substantially lower side lobes. It should be noted that the peak gain is lower.

Steering Angle (deg) 3-dB Beamwidth (deg) Maximum Gain (dB) Realized Steering Angle (deg) Sidelobe Level (dB) 0 27.5 6.3 0 -10.5 30 30.2 5.0 25.9 -7.0 60 42.1 2.1 47.9 -6.5 90 49.1 0.35 57.8 -6.7 se the l be ray’s A

sequence of the variable gain amplifiers being turned down to obtain a tapered distribution. Still, by close observation, the gain is greater than that obtained in Figure 21 over a broader range of angles. Table 2 contains tabulated values of various parameters for comparison with Table 1.

o

This is a con

Figure 22:

Simulated gain for various steering angles, using a

Kaiser tapered distribution

‘Broadside’

‘Relative Orientation of Phased Array’

Table 1:

Tabulated parameters of simulated gain for various steering angles,

using a uniform distribution

Approximate Radiation Pattern of Composite Beam at Various Steering Angles

(27)

Schematic Redesign

The following design objectives will be addressed in this sub-section: • Increase dynamic range

In this subsection, the Antenna Element’s schematic will be modified to incorporate the discussed changes. Minimal changes will be necessary for the new RF Signal Processor board, so no discussion of the RF Signal Processor board is given (notably, only the

removal of the analog phase shifter voltage section will be necessary). The addition of a variable gain amplifier will improve the dynamic range of the system. The implementation of the schematic into ADS will be briefly discussed. Some components were simplified

with the proper use of engineering judgment) in order to allow implementation in a timely anner.

igu

onnected directly to port 1. Both the 3rd - and 5th-order filters are modeled using the

reviously discussed design. AMP1 and AMP2 use a manufacturer-supplied model. The P242D phase shifters are modeled using their scattering parameters. There is

pproximately 1.3 dB of insertion loss for each. The phase shift may be manually

ontrolled for each phase shifter. Unfortunately, it was not possible to easily convert the

C supplied phase shift control voltage into a phase shift due ns of ADS

oftware. The only feasible method for doing this would have involved custom -programming within the ADS environment. Lastly

mplifier is modeled using a system level amplifier. arameters are incorporated from the datasheet. The

GA2031 described its small-scale nonlinear distorti

hannel Power Rejection). Unfortunat I (Third

rder Intercept) value for insertion into the system level amplifier. (The third order

terce ualto the

ird h Steering Angle (deg) 3-dB Beamwidth (deg) Maximum Gain (dB) Realized Steering Angle (deg) Sidelobe Level (dB) 0 32.9 3.2 0 -26.8 30 35.6 2.0 24.5 -21.6 60 46.4 -0.76 44.5 -20.5 90 55.8 -2.3 52.9 -20.5

Table 2:

Tabulated parameters of simulated gain for various steering

angles, using a tapered distribution

( m

F re 23 shows a top level view of the antenna element’s schematic. The antenna is

c p V a c D tolimitatio s

C , the BGA2031 variable gain

The P1dB point and scattering amplifier specification sheet for on using the term ACPR (Adjacent a

p B

C ely, this is not easily convertible to a TO

O

in pt is the theoretical output power of the desired signal when it would be eq

armonic.) th

(28)

MLSUBSTRATE2 Subst1 Layer LayerType[1]=signal Cond[2]=174000 T[2]=34 u m Cond[1]=174000 T[1]=34 u m TanD=0.0021 H=20 mil Er=3.38 Type[2]=ground

Dielectric-i: ER[i], H[i], TAND[i] Metal- i : T [ i], COND[i], TYPE[i]

Metal-2 Metal-1 Dielectric-1 MSUB MSub1 Rough=.0014 meter TanD=.0021 T=34 um Hu=3.9e+034 mil Cond=174000 Mur=1 Er=3.38 H=20 mil MSub S_Param SP1 Step=10 MHz Stop=3.0 GHz Start=2.0 GH z S-PARAMETERS Term Term1 Z=50 Ohm Num=1 Port VGA_Control Num=1 C CX5 C=100 pF L LX1 R= L=22 nH Port +3_vDC Num=2 DC_Feed DC_Feed3 C CX10 C=100 pF DC_Feed DC_Feed5 R RX2 R=0 R RX1 R=0 C C22 C=0.01 uF BGA2031 X4 Port RF_Output Num=4 L LX2 R= L=22 nH DC_Feed DC_Feed4 C CX7 C=1000 pF C CX6 C=0.1 uF C CX9 C=1000 pF C CX11 C=0.1 uF Term Term2 Z=50 Ohm Num=2 R RX R=22 Ohm C CX8 C=36 pF C CX4 C=36 pF Port Antenna_Input Num=5 5th_order X6 C CX12 C=36.0 pF DC_Feed _Feed1 DC C C21 C=100 pF 3rd_order X7 C C2 C=36.0 pF L L1 R= L=4 nH R R1 R=68 Ohm C C19 C=1000 pF C C1 C 8 =0.1 pF L FB1 R= L=100 nH V_ SR DC C2 dc=5.0 V V C C3 C=36 pF sa_hp_MGA-86576_19930601 AMP1 OFFSET=0 mil SMT_Pad="Pad1" sa_hp_MGA-86576_19930601 AMP2 SMT_Pad="Pad1" OFFSET=0 mil DC_Feed DC_Feed2 C C20 C=1 uF R R2 R=150 Ohm L L2 R= L=4 nH C C5 C=22.0 pF C C4 C=36.0 pF D d D d6 C_Fee C_Fee VP242D X2 VP242D X1 VP242D X3 C CX2 C=0.1 uF C CX3 C=1000 pF C C15 C=0.1 uF C C14 C=1000 pF C C17 C=0.1 uF C C16 C=1000 pF C CX1 C=36 pF Port Phase_Shift_Control C C6 C=36 pF Num=3 in ors. The following set of tests was executed. Results are described within the corresponding figures:

• Gain. The overall system gain was measured. Figure 24.

• Gain Compression. The power level for the system experiencing 1 dB of ga compression was computed. Figure 25.

• Noise Figure. The overall noise figure of the system with the various contribut Figure 26.

• Intermodulation. The system receives two tones at maximum power of -35 dBm. The simulator calculates the resulting mixer terms within the system bandwidth. Figure 27.

(29)

The

65dB of attenuation occurs for a signal falling outside of the passband by 200 MHz. It sho

pro s

overall system gain is ~65 dB within the passband of the receiver. Approximately uld be noted that an additional 20 dB of variable gain is available in the RF signal ces or.

Figure 24:

Overall System Gain vs Frequency

2.2 2.4 2.6 2.8 2.0 3.0 -40 -20 0 20 40 60 -60 80 freq, GHz d B ( G a i n _ T o t a l . . S ( 2 , 1 ) )

Antenna Element PCB, S21 Gain

Antenna Element PCB, S21 Phase

2.2 2.4 2.6 2.8 2.0 3.0 -100 0 100 -200 200 freq, GHz p G a i n_ T o t a l . . S ( 2 , 1 ) ) h a s e (

(30)

m t to its minimum setting, Pin can be as high as –35 dBm without gain

mpression.

The P1dB output power is 11.69 dBm. With the 3rd -stage VGA amplifier set to maximu

setting, Pin can be as high as –53 dB without gain compression. With the 3rd -stage

amplifier se co

Figure 25: Gain Compression of System

P1dB q= m(HB1. 2.300 output power m2 fre dB HB.Vout)=11.685 GHz 0.5 1.0 1.5 2.0 0.0 2.5 2 4 6 8 10 0 12 freq, GHz O / P P o w e r a t 1 d B c o m p ( d B m ) m3 Pin= linear=13.071-53.000 m4 Pin= dBm(HB2.HB.Vout[::,1])=11.971 -53.000 -90 -80 -70 -60 -50 -100 -40 -20 0 20 -40 40 Pin (dBm) P o u t ( d B m ) m3 m4

Output Pow er v s Pin

-90 -70 -60 -50 -40 -30

-100 -20

P1dB compression calculated using ADS Gain Compression Simulator (P1dB = 11.685 dBm) P1 -80 -80 -60 -40 -20 0 -100 20 Pin ( dBm)

Pout vs 1st, 2nd, and 3rd Stage Amplifi ers

With a sett ing of -33dB gain for 3rd stage amplf ier, Pin can be as high as -35 dB without gain compression.

dB compression calculated by manually var in of Pin Output Power Linear Fit 2nd Stage Amplifier

1st Stage Amplifier (LNA)

3rd-Stage Amplifier (Variable Gain)

With a setting of -33 dB gain for the 3rd stage VGA

amplifier, Pin can be as high as –35 dBm without gain compression. Pin for ) d B m P o u t (

(31)

nf freq 2.200GHz 2.225GHz 2.250GHz 2.275GHz 2.300GHz 2.325GHz 2.350GHz 2.375GHz 2.400GHz nf(1) nf(2) 3000.000 3000.000 3000.000 3000.000 3000.000 3000.000 3000.000 3000.000 3000.000 2.494 2.445 2.705 2.412 2.121 2.408 2.853 2.672 2.867 Noise_Figure1..port1.NC.freq 2. Noise_Figure1..port1.NC.name 300GHz AMP1.a1 AMP2.a1 R1 X1.S2P2 X6.CLin1 X6.CLin2 X7.CLin1 X7.CLin2 _total Noise_Figure1..port1.NC.type Ampl ifie r Ampl ifie r R S_Port Line ine Noise_Figure1..port1.NC.vnc PC_ PC_L PC_Line PC_Line _total 200.1pV 1.132pV 85.55fV 16.53fV 917.5fV 544.9fV 129.0pV 41.55pV 241.7pV

Figure 26: Noise Figure of System

Noise Figure Contributions

The system noise figure is 2.12 dB at 2.3 GHz. As to be expected the major contributor is e first stage LNA. The noise figure of the LNA is ~2 dB.

(32)

freq 0.0000 Hz 50.00MHz 100.0MHz 150.0MHz 200.0MHz 250.0MHz 300.0MHz 350.0MHz 400.0MHz 450.0MHz 500.0MHz 1.950GHz 2.000GHz 2.050GHz 2.100GHz 2.150GHz 2.200GHz 2.250GHz 2.300GHz 2.350GHz 2.400GHz 2.450GHz 2.500GHz 2.550GHz 2.600GHz Mix Mix(1) Mix(2) 0 -1 -2 -3 -4 -5 -6 -7 -8 -9 -10 10 9 8 7 6 5 4 3 2 1 0 -1 -2 -3 0 1 2 3 4 5 6 7 8 9 10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 m1 freq= dBm(Vout)=16.0992.400GHz m2 freq= dBm(Vout)=-10.703 2.350GHz m1 freq= dBm(Vout)=16.0992.400GHz m2 freq= dBm(Vout)=-10.703 2.350GHz -120 -100 -80 -60 -40 -20 0 -140 20 2.20 2.25 2.30 2.35 2.40 2.45 2.50 2.55 2.60 2.15 2.65 f req, GHz d B m ( V o u t ) m1 m2

m signal level for the rec eiver). t.

The input signals are 2.4 and 2.45 Ghz tones at -35 dBm (the maximu As can be seen, there is ~26 dB of attenuation in the third order produc

Figure 27: Intermodulation System Test Mixing Products Included

in Simulation

Mixing Product In-Band

Signal Output at Edge of Passband

The input signals are 2.4 and 2.45 GHz tones at –35 dBm (the maximum signal level for the active antenna). As can be seen, there is ~26 dB of attenuation in the third order product.

(33)

Trans

The following design objectives will be addressed in this sub-section: • Add dual mode capabilities – transmit & receive at 10 W • Transmission and reception must operate in duplex mode

The ideal scenario would be to use the same antenna for both transmit and receive.

Unfortunately, this is not possible. This is a result of the requirement that the antenna be able to send and receive simultaneously at the same frequency. The closest realization would have made use of an isolator. Due to the relatively poor mismatch of the antenna, reflected signal from the transmit circuit would overwhelm the receiver. No filtering could be used to attenuate the reflected signal since it is the same frequency as the receive

section.

Upon consultation with the customer, it was decided to make the transmitter separate from the receiver. Consequently, the device would only be able to transmit

that 8 W of transmit power was adequate. An additional requirement to tr range of 1 mW thru 8 W was added.

Fortunately, the receive antenna would be able to reuse a large portion of receiver design. Figures 28 and 29 show a block diagram of the prelimina

concept. Notice that the composite output ranges from 0.23 mW thru 8 W. The low side of 0.23 m will allow tapering of the elements when operating at the minimal power of 1 mW. Unfortunately, the same is not true when operating at 8 W.

A brief description of the signal flow in Figures 28 and 29 follows. The modulated signal is amplified to a level such that the phase shifters in each antenna element will receive a signal level of ~0 dB of signal power after going through a splitter. This is a requirement of the phase shifters for optimal performance. The signal is then filtered to remove any spurious signals. Finally, the signal is amplified to 1 W (or the desired transmit power level) at each antenna element. As a result, the total radiated power is ~8 W (not including

losses in the antenna structures). The c r is to provide a ‘driver amplifier’ that meets

the listed specification: 50 dB dynamic range, -10 dB minimum gain, 40 dB maximum gain, 1 W maximum output. Further requirements of the ‘driver amplifier’ are listed in subsequent ragra

To verify the calculations in Figures 28 and 29, a simplified schematic was incorporated into ADS. The implementation of the schematic into ADS will be briefly discussed. Some

omponents were simplified in order to allow implementation in a timely manner. These

simulate insertion loss. Although not explicitly used in the simulation, the power splitter’s model has an isolation of 20 dB. It is recommended that the cables connecting the splitter to the active elements be close to the same length (~3/100*λ or ~3 mm). This is not crucial since path length differences may be accounted for by proper adjustment of the phase

mit Capabilities

. It was also decided ansmit through a the existing ry design W ustome pa phs. c

will be briefly discussed. The power splitter was modeled by configuring multiple two-way splitters in series. An attenuator is placed within the power splitter circuit in order to

(34)

shifters. The power amplifier was reviously discussed amplifier prop

modeled using a system amplifier. In addition to the erties, the following were also incorporated. The

f 30.2 dBm, and a 40 by comparison of a

sim rmance,these

spe ic erfilterandVP242D

ha s nadesign.

p

amplifier has a NF (Noise Figure) of 5 dB, a 1 dB compression point o dBm TOI (Third Order Intercept). These specifications were obtained

ilar off the shelf component. In order to ensure adequate perfo

cif ations should be met or exceeded. Lastly, both the 5th-ord

(35)

Figure 28: Top-Level Block Diagram of Active Antenna Array Driver Amplifier 1 W Maximum Power 40 dB Maximum Gain 50 dB Dynamic Range -10 dB Minimum Gain Composite Pow -6.4 dBm (0.23 39 dBm (8 W) Dynamic Range

Signal Level = 0 dBm thru 30 dBm Adjust Signal to 9.7 dBm

Signal Level = -9.7 dBm through 20.3 dBm Must Limit Signal Level to exactly 0 dBm for Active_Antenna Element (Phase Shifter limitation)

Signal Level -13.1 dBm thru 30.0 dBm (1 W)

'Active Antenna Array'

'Transmitter' Signal Level = -25 dBm er Radiated mW) = 43.1 dB S1P_Eqn Active_Antenna_Element1 S1P_Eqn Active_Antenna_Element2 S1P_Eqn Active_Antenna_Element3 S1P_Eqn Active_Antenna_Element4 S1P_Eqn Active_Antenna_Element5 S1P_Eqn Active_Antenna_Element6 S1P_Eqn Active_Antenna_Element8 S1P_Eqn Active_Antenna_Element7 Splitter X1 Isolation=20 dB Insertion_Loss=0.7 dB 2 3 4 6 5 7 8 9 1 S1P_Eqn

(36)

Figure 29: Top-Level Schematic of Active Antenna Element

May add optional 5th order filter to remove spurious signals outside of 2.2 - 2.4 GHz

Insertion Loss = ~1.5 dB

Signal Level -15.4 dBm thru 30.0 dBm (1 W)

Driver Amplifier 1 W Maximum Power 40 dB Maximum Gain 50 dB Dynamic Range -10 dB Minimum Gain Signal Level -5.4 dBm Signal Level -3.9 dBm Signal Level 0 dBm

'Active Antenna Element'

RF Power = 0 dBm Nominal Insertion Loss = 1.3 dB Typical Return Loss = 18 dB RF Power = 0 dBm Nominal

Insertion Loss = 1.3 dB Typical Return Loss = 18 dB RF Power = 0 dBm Nominal

Insertion Loss = 1.3 dB Typical Return Loss = 18 dB Amplifier Driver_Amplifier1 5th_order 5th_order_filter S1P_Eqn Phase_Shifter3 S1P_Eqn Phase_Shifter2 S1P_Eqn Phase_Shifter1

(37)

in the corresponding figures:

•

• Gain Comp e system experiencing 1 dB of gain

comp

• _{with the various}

• Interm the modulator at maximum

power, -28 dBm ixer terms within the

system

The following set of tests was executed. Results are described with

Gain. The overall system gain was measured. Figure 30. ression. The power level for th

ression was computed. Figure 31.

Noise Figure. The overall noise figure of the system contributors. Figure 32.

odulation. The system receives two tones from . The simulator calculates the resulting m bandwidth. Figure 33.

(38)

With a maximum gain setting, the overall system gain is ~55 dB. With a modulated signal of –25 dBm, this will produce an output signal of 30 dBm, or 1 W. This is the output of one element. The composite power output will be 8 W. Similarly, for

minimum gain setting, a modulated signal of –25 dBm will produce an output signal of -15 dBm, or 0.03 mW. This is the output of one element. The composite power output will be 0.24 mW. 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.0 3.0 10 20 30 40 50 0 60 freq, GHz d B ( S ( 2 , 1 ) ) 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.0 3.0 -150 -100 -200 -50 0 50 100 150 200 freq, GHz , 1 ) p h a s e ( S ( 2 ) 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.0 3.0 -40 -30 -20 -10 0 10 -50 20 freq, GHz d B ( M i n_ G a i n _ T r a n s m i t t e r . . S ( 2 , 1 ) ) 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.0 3.0 -150 -100 -50 0 50 100 150 -200 200 freq, GHz p h a s e ( M i n_ G a i n_ T r a n s m i t t e r . . S ( 2 , 1 ) )

Figure 30:

Overall Active Antenna Gain vs Frequency

Maximum Gain Setting

(39)

The input power from the modulator is set to a constant of 25 dBm. The VGA (Variable is

he Gain Amplifier) of the patch is varied. With the VGA set to 35 dB, the output power ~30 dBm (1 W). Under this scenario, the transmitter is outputting maximum power. T amplifier is experiencing 0.844 dB of gain compression.

.9

0 10 2 0 0 30 40 -1 -10 0 10 20 30 -20 40 VGA (dB) P o u t ( d B m )

m3

m4

Output Power vs Pin

Element_Gain 35.000 compression 0.844 Gain of VGA in Transmitter Element

(40)

freq 2.200GHz 2.225GHz 2.250GHz 2.275GHz 2.300GHz 2.325GHz 2.350GHz 2.375GHz 2.400GHz nf nf(1) nf(2) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0. 0. 5.034 5.036 5.035 5.032 5.035 5.037 5.036 5.035 5.036 000 000 ...itter_No e_Figure1..port2.NC.freqis 2.300GHz ...ter_Noise_Figure1..port2.NC.name _total AMP1 X2.AMP1 X2.X3.S2P2 X1.PWR13.CMP1 X2.X2.S2P2 X2.X1.S2P2 X2.X4.CLin1 X1.PWR11.CMP1 X2.X4.CLin2 X1.PWR10.CMP1 X1.ATTEN1.CMP1 X3.X1.S2P2 X6.X1.S2P2 ...itter_Noise_Figure1..port2.NC.type _total Ampl ifi er Ampl ifi er S_Port S_Port S_Port S_Port PC_Line S_Port PC_Line S_Port S_Port S_Port S_Port ...itter_Noise_Figure1..port2.NC.vnc 428.6nV 426.1nV 38.74nV 12.06nV 10.44nV 10.39nV 8.912nV 8.586nV 7.110nV 4.819nV 4.234nV 2.262nV 1.625nV 1.625nV

Noise Figure of Antenna Transmitter

Noise Figure Contributions

Figure 32:

The system noise figure is 5 dB. The primary contributor is the first stage driver

amplifier before the power splitter. Since the system signal level is at 25 dBm, the noise gure should not introduce a significant amount of noise.

(41)

freq 0.0000 Hz 50.00MHz 100.0MHz 150.0MHz 200.0MHz 250.0MHz 300.0MHz 350.0MHz 400.0MHz 450.0MHz 500.0MHz 1.950GHz 2.000GHz 2.050GHz 2.100GHz 2.150GHz 2.200GHz 2.250GHz 2.300GHz 2.350GHz 2.400GHz 2.450GHz 2.500GHz 2.550GHz 2.600GHz Mix Mix(1) Mix(2) 0 -1 -2 -3 -4 -5 -6 -7 -8 -9 -10 10 9 8 7 6 5 4 3 2 1 0 -1 -2 -3 0 1 2 3 4 5 6 7 8 9 10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 m1 freq= dBm(Vout)=27.50 2.400GHz m2 freq= dBm(Vout)=-5.362.350GHz m1 freq= dBm(Vout)=27.50 2.400GHz m2 freq= dBm(Vout)=-5.362.350GHz

The input signals are 2.4 and 2.45 GHz tones at -28 dBm (the maximum signal level for the transmitter). As can be seen, there is ~33 dB of attenuation in the third order product.

2.20 2.25 2.30 2.35 2.40 2.45 2.50 2.55 2.60 2.15 2.65 -120 -100 -80 -60 -40 -140 -20 0 20 40 f req, GHz d B m ( V o u t ) m1 m2

Figure 33: Intermodulation of Active Antenna Mixing Products Included

in Simulation

Mixing Product In-Band

Signal Output at Edge of Passband

(42)

onclusio

or the presented conceptual redesign, it has been shown that all major redesign bjectives have been met. The antenna bandwidth was increased to operate over the

200-2400 MHz band. Simulations show that the new patch element meets this

andwidth yielding a VSWR less than or equal to 2 in the passband. This corresponds to pproximately 90% of the power being transmitted at the edge of the passband. The

andpass filters were modified to accommodate the extended bandwidth. The antenna eamforming circuit has been modified to include digital control over both arbitrary hase and attenuation of each element. Simulations incorporating both the element and rray beam pattern show that the beam may be steered to a realizable angle of 60 degrees

om broadside. At a steering angle of 60 degrees, however, the main return axis is ttenuated by ~6 dB. This is due to the limitations of the patch element’s beam pattern.

he receive antenna circuit was modified to accommodate an increased dynamic range. simulation of the antenna revealed the capability of receiving a signal power as high as 35 dBm without overdriving the system. The gain of the receive system

e ~65 dB with an additional 20 dB of variable gain within the

ther tests include a noise figure (2.1 dB) and an intermodulation test (~26 dB of

ppression). Next, a block diagram design for an active transmit antenna was presented. iven that the discussed driver amplifier specification is met, the a

apable of transmitting within a range of 0.25 mW thru 8 W. As a

e block diagram, a simulation in ADS was performed. The circuit demonstrated

cceptable performance in a suite of tests to include gain, gain compression, noise figure nd intermodulation. In conclusion, these results show that both a new receive antenna

nd an ive transmit antenna should be capable of meeting the proposed redesign

bjectiv .

C

ns

F o 2 b a b b p a fr a T A – was shown to RF signal processor. b O su

G ctiveantenna will be

second validation of c th a a a act es o

(43)

Further Work

The following outline briefly summarizes anticipated future work with the second iteration of the antenna array.

1. Build prototype of antenna element

a. Verify / Order required components b. Calibration strategy

c. Complete board layout i. Antenna element

ii. Signal processing board iii. Patch antenna

d. Fabricate antenna

e. Calibrate and test antenna

f. Interconnect user supplied digital controller g. Create desired adaptive beamforming algorithm 2. Build prototype of active antenna element (transmitter)

a. Complete design of driver amplifier b. Verify / order required components

c. Calibration strategy d. Complete board layout

i. Antenna element

ii. Signal processing board iii. Fabricate antenna

design

e. Calibrate and test antenna

f. Interconnect user supplied digital controller g. Create desired adaptive beamforming algorithm 3. Future enhancements

a. Possible to use same antenna for transmit and receive i. Use different operating frequencies

ii. Isolate by using time division multiple access b. Increase number of elements / size of antenna

(44)

References

ive, Volume I, Norwood, MA: Artech House, 1981, p 160.

] Radiation Patterns Using

[4] Stutzm enna Theory & Design. New

Sons, Inc., 1998.

[5] Bal s nalysis and Design, John Wiley and Sons,

New h Frequency Structure Simulator v9

Us orporation, Pittsburgh, PA, March 2004.

[7] Mat e icrowave Filters, Impedance-Matching Networks,

[8] Van

John Wiley & Sons, Inc., ]

l. AP-13, pp. 342-345,

[1] Depardo, Dan, “2.3-2.4 GHz Phased-Array Prototype”, 2004.

[2] Ulaby, Fawwaz; Moore, Richard; Fung, Adrian, Microwave Remote Sensing

Active and Pass

[3 Huie, C. Keith, Microstrip Antennas: Broadband

Photonic Crystal Substrates, 2002.

an, Warren L. & Gary A. Thiele. Ant

York, NY: John Wiley &

ani , C.A. Antenna Theory: A

York, 1997.[6] Ansoft Hig

r’s Guide, Rev 2.0, Ansoft C

e

Young and Jones, M

tha i,

and Coupling Structures, Artech House, 1980, pp. 614.

Trees, Harry L. Optimum Array Processing, Part IV of Detection, Estimation, and Modulation Theory. New York:

2002.

[9 P. W. Hannan and M. A. Balfour, “Simulation of a phased-array antenna in

waveguide,” IEEE Trans. Antennas Propagat., vo May 1965.

(45)

Appendices

Appendix I: Matlab Code

Figure 9

%Wr i t t en by:

poi nt 1_db=6.

maxi mum=10^( al pha=acos( p

zer o=20*l og10( el ement ) >0 el ement _db=el ement _ db. * zer o pol ar ( t het a, el ement _db)

Figure 17, 18

%Wr i t t en by: Russel l Hof er %Mar ch 2005

%Pr ogr am Descr i pt i on: Consummat e beam pat t er n at var i ous st eer i ng

% angl es Uni f orm or Taper ed Wei ght i ng

N = 8; % El ement s i n ar r ay

d_cm = 2. 9 % El ement spaci ng i n cent i met er s

d_m = d_cm/ 100; % El ement spaci ng i n met er s

c = 3e8; % Speed of l i ght

f = 2. 3e9; % Nomi nal f r equency of ar r ay

l ambda = c/ f ;

d = d_m/ l ambda; % spaci ng wr t wavel engt h

st eer i ng = 90 % st eer i ng angl e i n degr ees ( 90 =

br oadsi de)

Bet a = 0 % Bet a = 0 => not t aper ed,

posi t i ve val ue => t aper ed

D=d*[ - ( N- 1) / 2: 1: ( N- 1) / 2] ; % el ement l ocat i ons

ang = pi *[ - 1: 0. 001: 1] ;

u = cos( ang) ;

%cal cul at e an appr oxi mat e beam pat t er n f or an el ement Russel l Hof er

%Mar ch 2005

%Pr ogr am Descr i pt i on: Appr oxi mat e El ement Beam Pat t er n t het a=- pi / 2: . 01: pi / 2;

32 %maxi mum val ue of r eal i zed gai n ( db) poi nt 2_db=1. 17 %r eal i zed gai n at t het a=60 degr ees

poi nt 1_db/ 20) poi nt 2=10^( poi nt 2_db/ 20)

oi nt 2/ maxi mum) *2/ ( 3^. 5) el ement =maxi mum*c os( al pha*si n( t het a) ) ; el ement _db=20* l og10( el ement )

(46)

t het a=pi *[ - 1/ 2: . 001: ( 1/ 2- . 001) ; ] %t empor ar y var i abl e used t o cal cul at e er n

axi mum val ue of r eal i zed gai n ( db) oi nt 2_db=1. 17 %r eal i zed gai n at t het a=60 degr ees

^( poi nt 2_db/ 20) ;

s( poi nt 2/ maxi mum) *2/ ( 3^. 5) ; ( t het a) ) ;

t ] ; %necessar y t o al i gn coor di nat e syst em

0] %st eer i ng r el at i ve t o br oadsi de

degr ees = br oadsi de

ng/ 180*pi ) *D' ) %St eer i ng, phase por t i on %Beamf or mi ng, ampl i t ude

%Use bessel di st r i but i on i f

1- ( 2*n_t i l da/ N) ^2) AB( i +1) =bessel i ( 0, expr essi on)

= AS. *AB %Wei ght vect or i ncl udi ng st eer i ng

%Nor mal i ze wei ght s s o max i s 1/ N

r n t o f or m el ement beam pat t

poi nt 1_db=6. 32 %m p

maxi mum=10^( poi nt 1_db/ 20) poi nt 2=10

al pha=aco

el ement =maxi mum*cos ( al pha*si n 001: 1] *0, el emen el ement =[ [ 0: . of el ement count =0 r st eer i ng=[ 0, 30, 60, 9 f o

st eer i ng=90- st eer i ng %t r ansf or m st eer i ng, 90

unt =count +1 co

AS = exp( j *2*pi *cos( st eer i AB = ones( N, 1) / N r t i on po Bet a~=0 i f t aper ed f or i =0: ( N- 1) n_ t i l da=i - ( N- 1) / 2; expr essi on=Bet a*sqr t ( end

end W

W=W/ max( abs( AB) ) / N Au = exp( j *2*pi *D' *u) ; B = W' *Au

B = B. *el ement %I ncor por at e el ement beam pat t e

consummat e beam pat t er n G = 20*l og10( ( abs( B) ) ) ; f i gur e( 1) i f count ==1 h=pol ar db( ang, G, - 50, ' r ' ) ; el sei f count ==2 h=pol ar db( ang, G, - 50, ' g' ) ; el sei f count ==3 pol ar db( ang, G, - 50, ' b' ) ; h= el sei f count ==4 ; h=pol ar db( ang, G, - 50, ' c' ) end l d on ho end