2.2 2.4 GHz Phased Array

Final Project

2.2-2.4 GHz Phased-Array Conceptual Redesign  by Russell Hofer May 2005

Tabl Table e of Cont ents ACKNOWLEDGEMENTS ............................................................................................. 1 INTRODUCTION............................................................................................................. 2 BACKGROUND OF EXISTING ANTENNA ARRAY................................................ 2 REDESIGN OBJECTIVES ................................................... .......................................... 6 INCREASE ANTENNA BANDWIDTH......................................................................... 7 Increase of Antenna Element’s Bandwidth...................................................................................... 7 Increase of Bandpass Filters’ Bandwidth .................. ........................... ................... ................... .................. ................... ................... .................. ............ ... 13

CONCEPTUAL REDESIGN OF ANTENNA ELEMENT ELEMENT PCB ............................... 18 Steering Redesign ................................. ................ .................................. ................................... .................. .................................. ................. .................................... ................... 18 Schematic Redesign...................................................................................................................... 24

TRANSMIT CAPABILITIES ........................................................ ....................................................................................... ............................... 29 TRANSMIT CAPABILITIES ........................................................ ....................................................................................... ............................... 30 CONCLUSIONS ..................................................... ............................................................................................................. ........................................................ 39 FURTHER WORK......................................................................................................... 40 REFERENCES................................................................................................................ 41 APPENDICES................................................................................................................. 42  Appendix I: Matlab Code ................... ............................ ................... ................... .................. ................... ................... .................. ................... ................... ................ ....... 42

Figure 9.................... 9.................................. ........................... ........................... ............................ ........................... ........................... ............................ ............................. ............................. ................. ... 42 Figure 17, 18....................... 18..................................... ............................ ........................... ........................... ............................ ............................ ............................ ............................ .................... ...... 42

1 / 44

 Ac  A c k n o w l edg ed g emen em ents ts

I would like to acknowledge Dr. Christopher Allen for his guidance through this project. In addition, I would like to thank Mr. Dan Depardo for his willingness to answer questions and share his valuable valuable experience of antenna antenna and microwave design. design. Lastly, Dr. Jim Stiles for consultation on general microwave principles and providing what has proven to be a very useful course in microwave design. I would like to extend a thanks to Dr. Christopher Allen, Dr. Prasad Gogineni, and Dr. Kenneth Demarest for participation in my committee. Last, but not least, I would like to extend my gratefulness to my wife, Kristie Hofer, for her support and patience.

2 / 44

Introduction In 2004, Kansas University and Honeywell FM&T Kansas City Plant jointly built a 2.3-2.4 GHz Phased-Array Prototype. Prototype. Mr. Dan Depardo, RF electronics electronics engineer, was responsible responsible for the design, construction, construction, and preliminary testing. testing. The antenna has proven to perform adequately per its original original design criteria. criteria. However, Dwayne Brown, KCP (Kansas (Kansas City Plant) product engineer, has extended requirements that are to be incorporated into the second design iteration. iteration. In this report, the results of the the conceptual redesign of the antenna array are discussed. Included is a brief background background of the existing antenna array, redesign redesign objectives, new antenna design, simulation results, and further work/conclusions.

Background of Existi ng Antenna Array The following discussion gives a limited overview of the original antenna array design. The interested reader is referred referred to [1] for for a detailed overview. The original antenna antenna consists of eight patch elements positioned positioned in a linear array. The array is capable of  performing electronic electronic steering in azimuth. azimuth. The bandwidth of of the antenna encompasses encompasses 2.32.4 GHz. Figure 1 shows an image image of the original original 2.3-2.4 GHz phased-array phased-array antenna  prototype [1]. [1].

Figure 1: 2.3-2.4 GHz GHz Phased-Array Phased-Array Antenna Antenna Prototype Each patch antenna element consists of a low-noise RF amplifier, two-stage RF bandpass filter (100 MHz B/W), two variable voltage phase shifters, and further small-signal RF amplification. Figure 2 shows shows a basic block block diagram of the antenna antenna element. Notice that each antenna element must be given a steering signal. This signal controls controls the amount of  phase shift incurred incurred by the RF signal signal at each element.

3 / 44

Patch Antenna

3rd -Order BPF (Bandpass Filter)

LNA (Low Noise Amplifier)

5th-Order BPF (Bandpass Filter)

SC (Steering Control Voltage)

Φ1 Φ2

Phase Shifters

Small-Signal Amplifier

Output

Figure 2: Block Diagram of an Antenna Element

4 / 44 Ideally, each antenna element will have a constant delta phase shift with respect to its neighboring elements given by Equation 1 [2]:

∆θ  = 2π d / λ  * cosθ 0  

(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.

5 / 44

Antenna Array

RF Signal Processor 

Sum Level (270mV – 3.0V)

Delta Level (270mV – 3.0V)

Steering Control (1.0 – 8.0 V)

Digital Control System

Gain Control (VGA) (0-2.5V)

“user supplied”

Figure 3: Top Level Block Diagram

RF Signal to S-band Receiver

6 / 44

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 needed

Oct 2004

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) Produce a hardware prototype (if possible)

May 2005

7 / 44

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

Figure 4: Patch Geometry [4]

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 1 st order transmissionline 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.

8 / 44

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]: ⎛ π 

 ⎞

 Rin( y =  yo) =  Rin( y = 0) * cos ⎜  y 0 ⎟   ⎝  L  ⎠ 2

(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

Figure 5: Patch Geometry

9 / 44 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

10 / 44 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.

Figure 8: 3-D Polar Plot from Patch Antenna Element

11 / 44

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: realized  _ gain = 4π  *

U  Pincident 

 

(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 |S 12|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.

12 / 44

‘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]: realized  _ gain = max* cos(alpha * sin(theta))  

(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 Appendix I for the actual code).

13 / 44 ‘Relative Orientation of Element’ 90

8

120

60 6

4

150

Approximate Radiation Pattern of One Patch Element

30

2

‘Broadside’

180

0

210

330

240

300 270

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.

5th-Order Filter 

3rd -Order Filter

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 5 th-order directly after the low noise amplifier. It might be noted that  placement of a filter before the low noise amplifier does not follow conventional design. The purpose

Figure 11: PCB Layout

14 / 44 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. 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 B notch in the center of the passband. The reason m1 m1 for this unacceptable behavior is unknow n. It freq=2.330GHz dB(S(2,1))=-5.142 appears that Agilent may not have thoroughly tested this design guide for robustness. Similar results were obtained with the 3 rd -order filter. 0

-5

-10

-15

The original filter design was made using a  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 p,  but once setup it is extremely flexible and effective.

-20    )    )    1  ,    2    ( -25    S    (    B    d -30

-35

-40

-45

-50 1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

2.6

2.8

3.0

freq, GHz

th

Figure 12: 5  Order Filter

Response using Built-in ADS

15 / 44 Figure 13 shows the optimization circuit for a 5 th-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.

Term Term1 Num=1 Z=50 Ohm

Goal OptimGoal1 Expr="delta" SimInstanceName="SP1" Min= Max=1.5 Weight=1 RangeVar[1]= RangeMin[1]= RangeMax[1]=

S_Param SP1 Start=2.0 GHz Stop=40 GHz Step=0.02 GHz

OPTIM

Term Term2 Num=2 Z=50 Ohm

GOAL

S-PARAMETERS

ML5CTL_V CLin2 Subst="Subst1" Length=L1 W[1]=W1 S[1]=S1 W[2]=W1 S[2]=S2 W[3]=W1 S[3]=S2 W[4]=W1 S[4]=S1 W[5]=W1 Layer[1]=1 Layer[2]=1 Layer[3]=1 Layer[4]=1

Layer[5]=1 RLGC_File= ReuseRLGC=no

ML5CTL_V CLin1 Subst="Subst1" Length=L2 W[1]=W1 S[1]=S1 W[2]=W1 S[2]=S2 W[3]=W1 S[3]=S2 W[4]=W1 S[4]=S1 W[5]=W1 Layer[1]=1 Layer[2]=1 Layer[3]=1 Layer[4]=1

Layer[5]=1 RLGC_File= ReuseRLGC=no

Var  Eqn

Metal-1 Dielectric-1 Metal-2

Optim Optim1 OptimType=Random MaxIters=999 DesiredError=0.0 StatusLevel=10 FinalAnalysis="SP1" NormalizeGoals=yes SetBestValues=yes Seed= SaveSolns=yes SaveGoals=yes SaveOptimVars=yes UpdateDataset=yes SaveNominal=yes SaveAllIterations=no UseAllOptVars=yes

VAR VAR1 W1= 37 mil opt{ 20 mil to 120 mil } S1=43 m il opt{ 20 m il to 80 mil } S2=54 m il opt{ 20 m il to 80 mil } L1=126 mil opt{ 80 m il to 160 mil } L2=660 mil opt{ 550 m il to 700 mil }

Meta l-i : T[i], COND[i], TYPE[i] Diele ctric -i : ER[i], H[i], TAND[i]

MLSUBSTRATE2 Subst1 Er=3.38 H=20 mil TanD=0.0021 T[1]=34 um Cond[1]= 1.0E+50 T[2]=34 um Cond[2]= 1.0E+50 LayerType[1]=signal LayerType[2]=ground

GOAL

GOAL

Goal OptimGoal3 Expr="stop1" SimInstanceName="SP1" Min= Max=-40 Weight= RangeVar[1]= RangeMin[1]= RangeMax[1]=

Goal OptimGoal4 Expr="stop2" SimInstanceName="SP1" Min= Max=-40 Weight= RangeVar[1]= RangeMin[1]= RangeMax[1]=

th

UseAllGoals=yes

GOAL Goal OptimGoal2 Expr="avg" SimInstanceName="SP1" Min=-1 Max= Weight=1 RangeVar[1]= RangeMin[1]= RangeMax[1]= Meas Eqn

MeasEqn Meas1 freq_lower=find_index(freq, 2.2e9) freq_upper=find_index(freq,2.4e9) freq_USB=find_index(freq,2.5e9) freq_LSB= find_index(freq,2.1e9) min_dev=min(dB(S(2,1)[freq_lo wer::f req_upper])) max_dev=max(dB(S(2,1)[f req_lo wer::freq_upper])) delta=abs(max_dev-min_dev) avg=mean(dB(S(2,1)[freq_lower::freq_upper])) stop1=dB(S(2,1)[freq_USB]) stop2=dB(S(2,1)[freq_LSB])

Figure 13: 5 -Order Optimization Circuit

16 / 44 The results obtained for the 5 th-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. m1

0

-10

m1 freq=2.330GHz dB(S(2,1))=-1.52 2

-20

   )    ) -30    1  ,    2    (    S    (    B -40    d

-50

-60

-70 1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

2.6

2.8

3.0

freq, GHz

th

Figure 14: 5 -Order Filter Response using ADS

Optimization As a final verification of the filter design, a more accurate simulation was performed using Momentum. Momentum is a 2.5d RF electromagnetic simulator packaged with ADS. This allows for relatively quick simulation of the designed filters by importing the filter design into Momentum. Figure 15 shows an example layout of the 5 th-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 slightly lengthened to compensate. The final simulation results of both the 3 rd 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.

th

Figure 15: 5 -Order Filter Layout in Momentum

17 / 44

S21

0 -10    ] -20    B    d    [ -30   g   a    M-40

-50 -60 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 Frequency rd  Figure 16: 3 -Order Filter Frequency

Res onse in Momentum

S21 S22 0 -10

   ]

-20

Ma    B-30 g.    d    [ [dB ]   g-40

  a    M

-50 -60 -70 2.0

2.1

2.2

2.3

2.4

2.5

2.6

2.7

Frequency Frequency th

Figure 17: 5 -Order Filter Frequency

Response in Momentum

2.8

18 / 44

Concept ual Redesign of Antenna Element PCB Steering Redesign

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

• •

Maximize steering 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 1.3 dB of inser tion loss. Another way to maximize the steering angle is to more accurately control the phase shift. The original configuration used an analog voltage divider circuit in order to control each  phase shifter’s phase shift. This limited the maximum delta phase shift to ~30 degrees; as a 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 overcome these weaknesses, and add more flexibility in beam control, the phase shifters in the new design will be individually control led by 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 tapered distribution 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 gain a mplifi cation. Figure 19 shows a top system level diagram. Lastly, for reference, Figure 20 shows a block diagram of the RF signal processor.

19 / 44

Patch Antenna

3r d -Order BPF (Bandpass Filter)

LNA (Low Noise Amplifier)

5th-Order BPF (Bandpass Filter)

SC (Steering Control Voltage)

Φ1 Φ2

Phase Shifters

Φ3 BS (Beam Shaping Voltage)

VGA (Variable Gain Amplifier)

Output

Figure 18: Block Diagram of a Modified Antenna Element

20 / 44

RF Signal Processor 

RF Signal to S-band Receiver

8

Steering Control Bus (0-10V) Sum Level (270mV – 3.0V)

Delta Level (270mV – 3.0V)

Digital Control System

Beam Shaping Bus (0-2.5V) Gain Control (VGA) (0-2.5V)

“user supplied”

Figure 19: Block Diagram of a Modified System

8

21 / 44 SC1 BS1

SC2 BS2

SC3 BS3

SC4 BS4

SC5 BS5

SC6 BS6

SC7 BS7

SC8 BS8

Antenna Element Circuit #1

Antenna Element Circuit #2

Antenna Element Circuit #3

Figure 20: Block Diagram of RF Signal Processor 

  r   e   n    i    b   m   o   c   r   e   w   o   p    t   u   p   n    i      4

Antenna Element Circuit #4

Antenna Element Circuit #5

Antenna Element Circuit #6

Antenna Element Circuit #7

Antenna Element Circuit #8

  r   e   n    i    b   m   o   c   r   e   w   o   p    t   u   p   n    i      4

RF Signal to S-band Receiver

Power Combiner  “Σ”

Hybrid 180 Degree “∆”

Delta Level

   t    i   u   c   r    i    C Sum Level   g   n    i   s   s   e   c   o   r    P    F    R

VGA

Key: SC = Steering Control BS = Beam Steering VGA = Variable Gain Amplification

22 / 44 We are now in a position to simulate the generated beam pattern obtained from both the antenna element 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  plots in order to take into account the array gain associated with adding all elements together. This would describe the overall antenna gain with respect to an isotropic antenna.  Nevertheless, the following plots accurately show the shape of the beam pattern. The first simulation will not u se a tapered distribution. The beam pattern will be calculated at steering angles (relative to broadside) of 0, 30, 60, and 90 degrees. Figure 21 shows the results. As can be seen from Figure 21, when the beam is steered to 90 degrees, the element’s beam 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 posite beam pattern. For comparison, a maximum gain of 2.1 dB is achieved for a steering angle of 60 degrees. The actual realized steering angle is 48 degrees. On the other hand, a maximum of gain of ~6.25 dB is achieved for a steering angle of θ=0 degrees (relative to broadside). Values for all an gles are tabulated in Table 1.

‘Broadside’

A pproximate Radiation Pattern of Composite Beam at Various Steering Angles ‘Relative Orientation of Phased Array’

Figure 21: Simulated gain for various steering angles, using a

uniform distribution

23 / 44

Steering Angle (deg)

3-dB Beamwidth (deg)

Maximum Realized Steering Gain (dB) Angle (deg)

Sidelobe Level (dB)

0 30 60 90

27.5 30.2 42.1 49.1

6.3 5.0 2.1 0.35

-10.5 -7.0 -6.5 -6.7

0 25.9 47.9 57.8

Table 1: Tabulated parameters of simulated gain for various steering angles,

using a uniform distribution 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 se the main beamwidth and decrease the sidelobe level. As in Figure 17, the beam pattern will be 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 ray’s  beam pattern. By comparison to Figure 21, the beamwidth has widened. Also, one o bserves substantially lower side lobes. It should be noted that the peak gain is lower. This is a consequence 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.

‘Broadside’

Approximate Radiation Pattern of Composite Beam at Various Steering Angles ‘Relative Orientation of Phased Array’

Figure 22: Simulated gain for various steering angles, using a

Kaiser tapered distribution

24 / 44 Steering Angle (deg)

3-dB Beamwidth (deg)

Maximum Realized Steering Gain (dB) Angle (deg)

Sidelobe Level (dB)

0 30 60 90

32.9 35.6 46.4 55.8

3.2 2.0 -0.76 -2.3

-26.8 -21.6 -20.5 -20.5

0 24.5 44.5 52.9

Table 2: Tabulated parameters of simulated gain for various steering

angles, using a tapered distribution 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 manner. Figure 23 shows a top level view of the antenna element’s schematic. The antenna is connected directly to port 1. Both the 3rd - and 5th-order filters are modeled using the  previously discussed design. AMP1 and AMP2 use a manufacturer-supplied model. The VP242D phase shifters are modeled using their scattering parameters. There is a pproximately 1.3 dB of insertion loss for each. The phase shift may be manually controlled for each phase shifter. Unfortunately, it was not possible to easily convert the DC supplied phase shift control voltage into a phase shift due to limitations of ADS software. The only feasible method for doing this would have involved custom C-programming within the ADS environment. Lastly , the BGA2031 variable gain amplifier is modeled using a system level amplifier. The P1dB point and scattering  parameters are incorporated from the datasheet. The amplifier specification sheet for BGA2031 described its small-scale nonlinear distorti on using the term ACPR (Adjacent Channel Power Rejection). Unfortunat ely, this is not easily convertible to a TOI (Third Order Intercept) value for insertion into the system level amplifier. (The third order intercept is the theoretical output power of the desired signal when it would be eq ual to the third harmonic.)

25 / 44

C C19 C=1000 pF

C C1 8 C =0.1 pF

V_DC L SRC2 FB1 Vdc=5.0 V L=100 nH R=

R R1 R=68 Ohm

Port  Antenna_Input Num=5

C CX12 C=36.0 pF

3rd_order  X7

Term Term1 Num=1 Z=50 Ohm

Port Phase_Shift_Control Num=3

C C2 C=36.0 pF

D C_Feed D C_Feed6

DC_Feed DC _Feed1

L C L1 C3 L=4 nH C=36 pF R= sa_hp_MGA-86576_19930601

C C21 C=100 pF

5th_order  X6

C C16 C=1000 pF

C C15 C=0.1 uF

C C14 C=1000 pF

C CX2 C=0.1 uF

R R2 R=150 Ohm

C C20 C=1 uF

C C4 C=36.0 pF

L L2 L=4 nH R=

 AMP1 SMT_Pad="Pad1" OFFSET=0 mil

C C17 C=0.1 uF

DC_Feed DC_Feed2

C C5 sa_hp_MGA-86576_19930601 C=22.0 pF  AMP2 SMT_Pad="Pad1" OFFSET=0 mil

C CX3 C=1000 pF C CX4 C=36 pF

C C6 C=36 pF

BGA2031 X4 R RX2 R=0

C CX1 C=36 pF

VP242D X1

VP242D X2

VP242D X3

C C22 C=0.01 uF

R RX R=22 Ohm

C CX8 C=36 pF

Port RF_Output Num=4

R RX1 R=0

Term Term2 Num=2 Z=50 Ohm

Port VGA_Control Num=1

Metal-1

S-PARAMETERS

Dielectric-1

MSub

Metal-2 Metal- i :

T [ ]i , COND[i],

Diele ctric-i: ER[i] , H[i],

TYP E[i]

TAN D[i]

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

S_Param SP1 Start=2.0 GH z Stop=3.0 GHz Step=10 MHz

MSUB MSub1 H=20 mil Er=3.38 Mur=1 Cond=174000 Hu=3.9e+034 mil T=34 um TanD=.0021 Rough=.0014 meter 

L LX1 L=22 nH R=

L LX2 L=22 nH R=

C CX5 C=100 pF

DC_Feed DC_Feed5

C CX10 C=100 pF

DC_Feed DC_Feed4 DC_Feed DC_Feed3

Port +3_vDC Num=2

C CX7 C=1000 pF

C CX6 C=0.1 uF

C CX9 C=1000 pF

C CX11 C=0.1 uF

Figure 23: Top-Level Schematic of Antenna Element 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 in compression was computed. Figure 25. •  Noise Figure. The overall noise figure of the system with the various contribut ors. 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.

26 / 44

 Antenna Element PCB, S21 Gain 80

   )    )    1  ,    2    (    S  .  .    l   a    t   o    T _   n    i   a    G    (    B    d

60 40 20 0 -20 -40 -60 2.0

2.2

2.4

2.6

2.8

3.0

freq, GHz

 Antenna Element PCB, S21 Phase    )    )    1  ,    2    (    S  .  .    l   a    t   o    T _   n    i   a    G    (   e   s   a    h   p

200

100

0

-100

-200 2.0

2.2

2.4

2.6

2.8

3.0

freq, GHz

The overall system gain is ~65 dB within the passband of the receiver. Approximately 65dB of attenuation occurs for a signal falling outside of the passband by 200 MHz. It should be noted that an additional 20 dB of variable gain is available in the RF signal  processor.

Figure 24: Overall System Gain vs Frequency

27 / 44

P1dB output power  12

40

   )   m 10    B    d    (   p   m   o   c    B    d    1    t   a   r   e   w   o    P    P    /    O

Output Power Linear Fit

Output Pow er v s Pin

m3 m4

20

8

   )   m    B    d    (    t   u   o    P

m2 freq=2.300GHz dBm(HB1.HB.Vout)=11.685

6 4

m4 Pin= -53.000 dBm(HB2.HB.Vout[::,1])=11.971

0

m3 Pin=-53.000 linear=13.071

-20

2 0 0.0

0.5

1.0

1.5

2.0

-40

2.5

-100

freq, GHz

-90

-80

-70

-60

-50

-40

Pin (dBm)

P1dB compression calculated using ADS Gain Compression Simulator (P1dB = 11.685 dBm)

P1dB compression calculated by manually var in of Pin

2nd  Stage Amplifier 1st Stage Amplifier (LNA) 3rd-Stage Amplifier (Variable Gain)

Pout vs Pin for 1st, 2nd, and 3rd Stage Amplifi ers 20 0

   ) -20   m    B    d    ( -40    t   u   o -60    P

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

With a setting of -33 dB gain for the 3  stage VGA amplifier, Pin can be as high as –35 dBm without gain compression.

-80 -100 -100

-90

-80

-70

-60

-50

-40

-30

-20

Pin ( dBm)

Figure 25: Gain Compression of System

The P1dB output power is 11.69 dBm. With the 3rd -stage VGA amplifier set to maximu m setting, Pin can be as high as –53 dB without gain compression. With the 3 rd -stage amplifier set to its minimum setting, Pin can be as high as –35 dBm without gain compression.

28 / 44

nf 

freq

nf(1) 2.200GHz 2.225GHz 2.250GHz 2.275GHz 2.300GHz 2.325GHz 2.350GHz 2.375GHz 2.400GHz

Noise_Figure1..port1.NC.freq 2.300GHz

nf(2) 3000.000 3000.000 3000.000 3000.000 3000.000 3000.000 3000.000 3000.000 3000.000

Noise_Figure1..port1.NC.name  AMP1.a1  AMP2.a1 R1 X1.S2P2 X6.CLin1 X6.CLin2 X7.CLin1 X7.CLin2  _total

2.494 2.445 2.705 2.412 2.121 2.408 2.853 2.672 2.867 Noise_Figure1..port1.NC.type  Ampl ifie r   Ampl ifie r  R S_Port PC_ Line PC_Line PC_Line PC_Line  _total

Noise_Figure1..port1.NC.vnc 200.1pV 1.132pV 85.55fV 16.53fV 917.5fV 544.9fV 129.0pV 41.55pV 241.7pV

 Noise Figure Contributions

Figure 26:  Noise Figure of System

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

29 / 44 Mixing Products Included in Simulation Mix

freq

Mix(1)

Mix(2)

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

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=2.400GHz dBm(Vout)=16.099 m1

20

Mixing Product In-Band

m2 freq=2.350GHz dBm(Vout)=-10.703 m2

0

Signal Output at Edge of Passband

-20

-40

   )    t   u   o    V -60    (   m    B    d -80

-100

-120

-140 2.15

2.20

2.25

2.30

2.35

2.40

2.45

2.50

2.55

2.60

2.65

f req, GHz

The input signals are 2.4 and 2.45 Ghz tones at -35 dBm (the maximu m signal level for the rec eiver).  As can be seen, there is ~26 dB of attenuation in the third order produc t.

Figure 27: Intermodulation System Test

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.

30 / 44

Trans mit Capabili ties 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 . It was also decided that 8 W of transmit power was adequate. An additional requirement to tr ansmit through a range of 1 mW thru 8 W was added. Fortunately, the receive antenna would be able to reuse a large portion of the existing receiver design. Figures 28 and 29 show a block diagram of the prelimina ry design concept. Notice that the composite output ranges from 0.23 mW thru 8 W. The low side of 0.23 mW 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 customer 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 paragra phs. 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 components were simplified in order to allow implementation in a timely manner. These will be briefly discussed. The power splitter was modeled by configuring multiple twoway splitters in series. An attenuator is placed within the power splitter circuit in order to 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

31 / 44 shifters. The power amplifier was modeled using a system amplifier. In addition to the  previously discussed amplifier prop erties, the following were also incorporated. The amplifier has a NF (Noise Figure) of 5 dB, a 1 dB compression point o f 30.2 dBm, and a 40 dBm TOI (Third Order Intercept). These specifications were obtained by comparison of a similar off the shelf component. In order to ensure adequate perfo rmance, these specifications should be met or exceeded. Lastly, both the 5 th-ord er filter and VP242D  phase shifters use the same model as that used in the receive antenna design.

32 / 44

'Active Antenna Array'

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)

S1P_Eqn  Active_Antenna_Element1

S1P_Eqn  Active_Antenna_Element2

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

S1P_Eqn  Active_Antenna_Element3 2

3

4

6

1

5

S1P_Eqn  Active_Antenna_Element4

Composite Pow er Radiated -6.4 dBm (0.23  mW) 39 dBm (8 W) Dynamic Range = 43.1 dB

7

S1P_Eqn Modulator 

Signal Level = -25 dBm

 Amplifier  Driver_Amplifier  Driver Amplifier  1 W Maximum Power  40 dB Maximum Gain 50 dB Dynamic Range -10 dB Minimum Gain

8

9

Splitter  X1 Insertion_Loss=0.7 dB Isolation=20 dB

S1P_Eqn  Active_Antenna_Element5

S1P_Eqn  Active_Antenna_Element6

S1P_Eqn  Active_Antenna_Element7

S1P_Eqn  Active_Antenna_Element8

Figure 28: Top-Level Block Diagram of Active Antenna Array

33 / 44

'Active Antenna Element'

Signal Level -3.9 dBm

Signal Level 0 dBm

S1P_Eqn Phase_Shifter1 RF Power = 0 dBm Nominal Insertion Loss = 1.3 dB Typical Return Loss = 18 dB

S1P_Eqn Phase_Shifter2 RF Power = 0 dBm Nominal Insertion Loss = 1.3 dB Typical Return Loss = 18 dB

S1P_Eqn Phase_Shifter3 RF Power = 0 dBm Nominal Insertion Loss = 1.3 dB Typical Return Loss = 18 dB

Signal Level -5.4 dBm

5th_order  5th_order_filter  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)

 Amplifier  Driver_Amplifier1 Driver Amplifier  1 W Maximum Power  40 dB Maximum Gain 50 dB Dynamic Range -10 dB Minimum Gain

Figure 29: Top-Level Schematic of Active Antenna Element

34 / 44 The following set of tests was executed. Results are described with in the corresponding figures:

• •

Gain. The overall system gain was measured. Figure 30. Gain Compression. The power level for the system experiencing 1 dB of gain compression was computed. Figure 31. •  Noise Figure. The overall noise figure of the system with the various contributors. Figure 32. • Intermodulation. The system receives two tones from the modulator at maximum  power, -28 dBm. The simulator calculates the resulting mixer terms within the system bandwidth. Figure 33.

35 / 44

200

60

150 50

100

   )    )    1  ,    2    (    S    (   e   s   a    h   p

40    )    )    1  ,    2    ( 30    S    (    B    d 20

50 0 -50 -100

10

-150 -200

0 2.0

2.1

2.2

2.3

2.4

2.5

2.6

2.7

2.8

2.9

3.0

2.0

2.1

2.2

2.3

2.4

2.5

2.6

2.7

2.8

2.9

3.0

freq, GHz

freq, GHz

Maximum Gain Setting

20

   )    )    1  ,    2    (    S  .  .   r   e    t    t    i   m   s   n   a   r    T _   n    i   a    G _   n    i    M    (    B    d

200

10

   )    )    1  ,    2    (    S  .  .   r   e    t    t    i   m   s   n   a   r    T _   n    i   a    G _   n    i    M    (   e   s   a    h   p

0

-10

-20

-30

-40

150 100 50 0 -50 -100 -150 -200

-50 2.0

2.1

2.2

2.3

2.4

2.5

freq, GHz

2.6

2.7

2.8

2.9

3.0

2.0

2.1

2.2

2.3

2.4

2.5

2.6

2.7

2.8

freq, GHz

Minimal Gain Setting

Figure 30: Overall Active Antenna Gain vs Frequency 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.

2.9

3.0

36 / 44

Gain of VGA in Transmitter Element Output Power vs Pin 40

m3 m4

30

Element_Gain 35.000

   ) 20   m    B    d    ( 10    t   u   o 0    P

compression 0.844

-10

.9

-20 -10

0

10

20

30

40

VGA (dB)

Figure 31: Gain Compression of Antenna Transmitter Element

The input power from the modulator is set to a constant of 25 dBm. The VGA (Variable Gain Amplifier) of the patch is varied. With the VGA set to 35 dB, the output power is ~30 dBm (1 W). Under this scenario, the transmitter is outputting maximum power. T he amplifier is experiencing 0.844 dB of gain compression.

37 / 44

nf 

freq

nf(1) 2.200GHz 2.225GHz 2.250GHz 2.275GHz 2.300GHz 2.325GHz 2.350GHz 2.375GHz 2.400GHz

...itter_Nois e_Figure1..port2.NC.freq 2.300GHz

nf(2) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

...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

5.034 5.036 5.035 5.032 5.035 5.037 5.036 5.035 5.036 ...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 Contributions

Figure 32:  Noise Figure of Antenna Transmitter

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 figure should not introduce a significant amount of noise.

38 / 44

Mixing Products Included in Simulation Mix

freq

Mix(1)

Mix(2)

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

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

m1 freq=2.400GHz dBm(Vout)=27.50

40

Mixing Product In-Band m1

m2 freq=2.350GHz dBm(Vout)=-5.36

20

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

Signal Output at Edge of Passband

m2

0

-20    )    t   u   o    V    (   m    B    d

-40

-60

-80

-100

-120

-140 2.15

2.20

2.25

2.30

2.35

2.40

2.45

2.50

2.55

2.60

2.65

f req, GHz

Figure 33: Intermodulation of Active Antenna

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.

39 / 44

Conclusio ns For the presented conceptual redesign, it has been shown that all major redesign o bjectives have been met. The antenna bandwidth was increased to operate over the 2200-2400 MHz band. Simulations show that the new patch element meets this  bandwidth yielding a VSWR less than or equal to 2 in the passband. This corresponds to a pproximately 90% of the power being transmitted at the edge of the passband. The  bandpass filters were modified to accommodate the extended bandwidth. The antenna  beamforming circuit has been modified to include digital control over both arbitrary  phase and attenuation of each element. Simulations incorporating both the element and array beam pattern show that the beam may be steered to a realizable angle of 60 degrees fr om broadside. At a steering angle of 60 degrees, however, the main return axis is attenuated by ~6 dB. This is due to the limitations of the patch element’s beam pattern. The receive antenna circuit was modified to accommodate an increased dynamic range. A 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 was shown to  be ~65 dB with an additional 20 dB of variable gain within the RF signal processor. Other tests include a noise figure (2.1 dB) and an intermodulation test (~26 dB of su ppression). Next, a block diagram design for an active transmit antenna was presented. Given that the discussed driver amplifier specification is met, the a ctive antenna will be capable of transmitting within a range of 0.25 mW thru 8 W. As a second validation of the block diagram, a simulation in ADS was performed. The circuit demonstrated acceptable performance in a suite of tests to include gain, gain compression, noise figure and intermodulation. In conclusion, these results show that both a new receive antenna and an active transmit antenna should be capable of meeting the proposed redesign o bjectives.

40 / 44

Further Work The following outline briefly summarizes anticipated future work with the second  design 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 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 i. May not be feasible given design constraints

41 / 44

References [1] [2] [3] [4] [5] [7] [8] [9]

Depardo, Dan, “2.3-2.4 GHz Phased-Array Prototype”, 2004. Ulaby, Fawwaz; Moore, Richard; Fung, Adrian, Microwave Remote Sensing  Active and Passive, Volume I, Norwood, MA: Artech House, 1981, p 160. Huie, C. Keith, Microstrip Antennas: Broadband Radiation Patterns Using Photonic Crystal Substrates, 2002. Stutzman, Warren L. & Gary A. Thiele.  Ant enna Theory & Design. New York, NY: John Wiley & Sons, Inc., 1998. Balanis, C.A.  Antenna Theory: Analysis and Design, John Wiley and Sons,  New York, 1997.[6]  Ansoft High Frequency Structure Simulator v9 User’s Guide, Rev 2.0, Ansoft Corporation, Pittsburgh, PA, March 2004. Matthaei, Young and Jones,  M icrowave Filters, Impedance-Matching Networks, and Coupling Structures, Artech House, 1980, pp. 614. Van Trees, Harry L. Optimum Array Processing, Part IV of Detection, Estimation, and Modulation Theory. New York: John Wiley & Sons, Inc., 2002. P. W. Hannan and M. A. Balfour, “Simulation of a phased-array antenna in waveguide,” IEEE Trans. Antennas Propagat., vol. AP-13, pp. 342-345, May 1965.

42 / 44

 Appendices  Appendix I: Matlab Code

Figure 9 %Wr i t t en by: 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; poi nt 1_db=6. 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 maxi mum=10^( poi nt 1_db/ 20) poi nt 2=10^( poi nt 2_db/ 20) al pha=acos( poi 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 ) 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; d_cm = 2. 9 d_m = d_cm/ 100; c = 3e8; f = 2. 3e9; l ambda = c/ f ; d = d_m/ l ambda; st eer i ng = 90 br oadsi de) Bet a = 0 posi t i ve val ue => t aper ed D=d*[ - ( N- 1) / 2: 1: ( N- 1) / 2] ; ang = pi *[ - 1: 0. 001: 1] ; u = cos( ang) ;

% El ement s i n ar r ay % El ement spaci ng i n cent i met er s % El ement spaci ng i n met er s % Speed of l i ght % Nomi nal f r equency of ar r ay % spaci ng wr t wavel engt h % st eer i ng angl e i n degr ees ( 90 = % Bet a = 0 => not t aper ed, % el ement l ocat i ons

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

43 / 44 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 el ement beam pat t er n poi nt 1_db=6. 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 maxi mum=10^( poi nt 1_db/ 20) poi nt 2=10^( poi nt 2_db/ 20) ; al pha=aco s( poi nt 2/ maxi mum) *2/ ( 3^. 5) ; el ement =maxi mum*cos ( al pha*si n( t het a) ) ; el ement =[ [ 0: . 001: 1] *0, el ement ] ; %necessar y t o al i gn coor di nat e syst em of el ement count =0 f or st eer i ng=[ 0, 30, 60, 90] st eer i ng=90- st eer i ng count =count +1

%st eer i ng r el at i ve t o br oadsi de %t r ansf or m st eer i ng, 90 degr ees = br oadsi de

AS = exp( j *2*pi *cos( st eer i ng/ 180*pi ) *D' ) AB = ones( N, 1) / N por t i on

%St eer i ng, phase por t i on %Beamf or mi ng, ampl i t ude

i f  Bet a~=0 %Use bessel di st r i but i on 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 ( 1- ( 2*n_t i l da/ N) ^2) AB( i +1) =bessel i ( 0, expr essi on) end end W = AS. *AB W=W/ max( abs( AB) ) / N

%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

Au = exp( j *2*pi *D' *u) ; B = W' *Au B = B. *el ement 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, ' el sei f count ==2 h=pol ar db( ang, G, - 50, ' el sei f count ==3 h=pol ar db( ang, G, - 50, ' el sei f count ==4 h=pol ar db( ang, G, - 50, ' end hol d on end

r' ); g' ) ; b' ) ; c' ) ;

%I ncor por at e el ement beam pat t e r n t o f or m