System Level Design Review - EDGEedge.rit.edu/content/P09343/public/Published Documents/System...
Transcript of System Level Design Review - EDGEedge.rit.edu/content/P09343/public/Published Documents/System...
System Level Design Review
PO9343 Microwave Data II
Joel Barry
Amanda Kristoff
Mia Mujezinovic
Michael Pecoraro
P09343 Microwave Data II Technical Review Agenda
Meeting Purpose:
This meeting is to review the system and components to confirm it meets the Functionality, Engineering Specifications and Customer Needs of the P09343 Microwave Data II KGOCE MSD team. Materials to be Reviewed: Project Overview (Rev. A) Customer Needs Chart (Rev. B) Customer Specifications Chart (Rev. B) Customer Needs to Specifications Chart (Rev. B) Concept Generation/System Level Design (Rev. A) Risk Assessment (Rev. A) Engineering Design Process (Rev. A) Component and System Progress Report (Rev. A) Two Quarter Schedule/Milestones (Rev. B)
Meeting Date: January 16, 2009
Meeting Location: 78-2150
Meeting time: 12:30 – 2 PM
Timeline:
Start Time
Topic of Review
12:30 Team Introductions
12:35 Project Overview
12:40 Concept Generation/System Level Design
12:50 Risk Assessment
12:55 Engineering Design Process
1:05 Component Progress: Branchline Hybrid
1:10 Component Progress: Knochel Hybrid
1:15 Component Progress: Schiffman Phase Shifter
1:20 System Progress: System A
1:25 System Progress: System B
1:30 Two Quarter Schedule
1:40 Question
Introduction
Mia
Members:
Mia Mujezinovic (Team Lead)
Michael Pecoraro
Amanda Kristoff
Joel Barry
Breakdown of Roles/Tasks:
90 Degree Hybrid: Michael Pecoraro
180 Degree Hybrid: Joel Barry
Schiffman Phase Shifter: Mia Mujezinovic
System A: Mia Mujezinovic
System B: Amanda Kristoff
Vertical Launch: Anaren, Joel Barry, Michael Pecoraro
EDGE Updates: Amanda Kristoff
Resources Utilized:
Ansoft Designer
Ansoft HFSS
IEEE Xplorer Database
Michael Enders (Anaren, customer)
Project Overview
Michael
Customer
Michael Enders RF Engineer Anaren Microwave Incorporated Space and Defense Group [email protected] 315-362-0273
Project
To design, build and verify the operation of two 4x4 Butler Matrices for use in antenna beamforming
Figure 1: Block Diagram of System A Figure 2: Block Diagram of System B
Purpose
Butler Matrices are networks that allow antenna beams to be electrically (as opposed to physically) steered. The necessary condition for electrically steering an antenna beam is that the antenna elements must be given different phase progressions. This is what a Butler Matrix does – depending on which input is used, a different phase progression is realized at the output.
This project is more of a research type project for Anaren. They have two main goals that they would like us to attain:
o Both Butler Matrices should be wideband – covering a bandwidth of 10-12GHz o Anaren has never used a Knöchel hybrid - they would like to see how it performs individually as
well as in a system
Concept Generation / System Level Design
Joel
Layout Stackup
Figure 3: Anaren specified stackup designs for implementation into both Butler Matrices. The only difference between the two stackups is the thickness of the Rogers 3003 dielectric.
The 120 mil stackup will be implemented in the Butler Matrix Designs for some key advantages: o Thicker dielectric yields thicker traces for equal impedance lines ensures traces will be
greater than the minimum manufacturing tolerance of 10 mil o Thicker traces are less susceptible to variation with the width tolerance of 0.5 mil
Topology Selection
By unfolding the crossovers of the Butler Matrix A design, a single copper trace layer that mirrors the symmetry of the 90° coupler.
Butler Matrix B is similarly unfolded to a single copper trace
Vertical Mounting SMA connectors allow for access to all ports without implementation of crossover networks.
Figure 4: Selected topologies for the Butler Matrix A Design (Top) and the Butler Matrix B Design (Bottom).
30 mil
30 mil
1.5 mil
Alternative Stackup
60 mil
60 mil
1.5 mil
Chosen Stackup Copper Ground Plane
Rogers 3003
Arlon 6700 Copper Traces
Rogers 3003
Copper Ground Plane
Risk Assessment
Amanda
Risks that have been addressed:
Understanding the theory behind the components Understanding customer expectations/needs Learning HFSS Understanding ideal simulation tool in Ansoft Designer
Risks that still need to be addressed or considered:
o Finalize and reevaluate specifications from Anaren o Vertical mount (receiving documentation from Anaren, designing and effects at system level needs
to be considered) o Lead time on board manufacturing is 6 weeks
Process
Michael
An extremely simplified process (or flow) by which we have been working to complete this project is seen below:
TheoryIdeal
Simulatio
ns
Designer
Simulatio
nsHFSS
Simulatio
ns
Figure 5: Project Flow Chart
This process applies to each component and system that we will design. To explain the above process a bit further, read below:
Theory - Everything begins with theory – there are textbook entries and papers written on each of the components that we are building (not necessarily the systems, but definitely on Butler Matrices in general). We read them, check them and fully understand them before getting started with any type of simulations. When we feel we are ready to start simulating, we go to the next level:
Ideal Simulations - This type of simulation is done in Ansoft Designer using the special ‘ideal components.’ These components do not take into account any losses and show the absolute best case performance of your component or system. Impedance values, electrical lengths and frequencies are determined by the papers that were read (formulas would be given) and depending on the specific specifications for the project.
Designer Simulations - Once the ideal simulations are completed, one can now include the substrate effects, discontinuities and other losses into the system using Ansoft Designer. These simulations will be, clearly, worse than the ideal simulations. The ideal simulation is, however, your goal; you always check back to see how well you can match it. Designer allows for many variables to be tuned in real time. Intelligent tuning, however, takes into account your ideal values and theoretical dependencies.
HFSS Simulations - Once the designer simulation has been finished, the final step is tuning in HFSS. Using the theoretical dependencies and the ability of designer to see, magnitude-wise, how much these changes affect the output variables, the final tweaking is done.
Again, this process applies to both the component level and system level designs. This is the most efficient way of simulating – otherwise, you would have no real basis for what you are doing – you would be blindly tuning variables and hoping for the best.
Component Progress - Branchline Hybrid
Michael
Ideal
Figure 6: Schematic View of Ideal Branchline Hybrid
Figure 7: S-Parameters of Ideal Branchline Hybrid Figure 8: Output Phase Difference for Ideal Branchline Hybrid
Ideal values were obtained from the paper: A Multisection Broadband Impedance Transforming
Branch-Line Hybrid; IEEE Transaction on Microwave Theory and Techniques. Formulas were given for
theoretical values of impedances – these were solved for a line impedance of 50 ohms.
Results are as expected: over the entire bandwidth of 10-12GHz, there is equal power division between
ports two and three and a 90 degree phase difference between ports two and three.
Po
rt1
Po
rt2P
ort3
Po
rt4
E
E=90deg
F=11GHz
Z=50
E
E=90deg
F=11GHz
Z=b
E
E=
90
de
g
F=
11
GH
z
Z=
a
E
E=90deg
F=11GHz
Z=50
E
E=90deg
F=11GHz
Z=50
E
E=90deg
F=11GHz
Z=50
E
E=90deg
F=11GHz
Z=b
E
E=90deg
F=11GHz
Z=b
E
E=90deg
F=11GHz
Z=b
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=
90
de
g
F=
11
GH
z
Z=
a
Component Progress - Branchline Hybrid
Michael
Designer (Initial 120mil)
Figure 9: Schematic View of Designer Hybrid Figure 10: Layout View of Designer Hybrid
Figure 11: S-Parameters of Designer Hybrid Figure 12: Output Phase Difference for Designer Hybrid
Variable Value [mm] Value [mil]
fifty 1.97623 77.8043307
down 3.77688 148.696063
mid 4.48688 176.648819
a 0.684897 26.9644488
b 2.95945 116.51378
c b
d a
The tune feature was used to tune the lengths and widths of the transmission lines – the values seen in the table to the left were found to give the best performance
Although not as perfect as the ideal results, the simulated results are quite good – we have approximately equal power division between ports 2 and 3 and there is approximately a 90 degree phase difference between ports 2 and 3.
Component Progress - Branchline Hybrid
Michael
Designer (Second Cut 120mil)
Figure 13: Schematic View of Second Cut Hybrid Figure 14: Layout View of Second Cut Hybrid
Figure 15: S-Parameters of Second Cut Hybrid Figure 16: Output Phase Difference for Second Cut Hybrid
Variable Value [mm] Value [mil]
fifty 1.97623 77.8043307
down 3.78 148.818898
mid 4.49 176.771654
a 0.70 27.5590551
b 3.00 118.110236
c b
d a
qwt 2.23 87.7952756
Quarter Wave Transformers (QWTs) were included at each of the four ports. These widths tuned, as well as the other lengths and widths to produce the table found to the left.
Transmission did not change much, neither did the phase. However, the return loss and isolations improved to approximately 22dB down.
Component Progress - Branchline Hybrid
Michael
HFSS (Initial)
Figure 17: 3D View of HFSS Branchline Hybrid Figure 18: Power Efficiency of HFSS Branchline Hybrid
Figure 19: S-Parameters of HFSS Branchline Hybrid Figure 20: Output Phase Difference for HFSS Branchline Hybrid
The ‘Export to HFSS’ option in Ansoft Designer was used to create the HFSS model seen in Figure 12.
This model, when simulated, has major problems. Figure 13 shows the power efficiency of this model
– as can be seen, over half of power over the entire 10-12GHz bandwidth is being lost somewhere.
From the simulated S-Parameters we see that the transmission is no longer 50:50 and that the
reflections and isolations have risen.
The phase difference plot shows that the phase difference is approximately correct, but has worsened
from the Designer simulation.
Component Progress - Knöchel Hybrid
Joel
Ideal
Figure 21: Schematic View of Ideal 180° Hybrid Coupler
Figure 22: S-Parameters of Ideal 180° Hybrid Coupler Figure 23: Output Phase Difference for Ideal 180° Hybrid Coupler
Ideal values were obtained from the paper: Broadband Printed Circuit 0°/180° Couplers and High Power
Inphase Power Dividers; IEEE Transaction on Microwave Theory and Techniques. Designer simulations
were presented for a 50 ohm line impedance.
The results are as expected. Over the entire desired bandwidth of 10-12GHz, there is equal power
division between ports two and three and a 180° phase difference between ports two and three when
power is applied to port four.
Port1
Po
rt2
Po
rt3
Port4
E
E=
18
0d
eg
F=
11
GH
z
Z=
91
.96
E
E=
45
de
g
F=
11
GH
z
Z=
86
.03
E
E=360deg
F=11GHz
Z=50
E
E=90deg
F=11GHz
Z=49.95
E
E=
90
de
g
F=
11
GH
z
Z=
69
.64
E
E=
90
de
g
F=
11
GH
z
Z=
69
.64
E
E=
36
0d
eg
F=
11
GH
z
Z=
50
E
E=90deg
F=11GHz
Z=55.78
E
E=90deg
F=11GHz
Z=55.78
E
E=135deg
F=11GHz
Z=86.03
E
E=
36
0d
eg
F=
11
GH
z
Z=
50
E
E=
18
0d
eg
F=
11
GH
z
Z=
41
.7
E
E=90deg
F=11GHz
Z=77.81
E
E=90deg
F=11GHz
Z=77.81
E
E=
18
0d
eg
F=
11
GH
z
Z=
69
.97
E
E=360deg
F=11GHz
Z=50
Component Progress - Knöchel Hybrid
Joel
Designer (Initial 120 mil)
Figure 24: Schematic View of Designer Hybrid Figure 25: Layout View of Designer Hybrid
Figure 26: S-Parameters of 180° Hybrid Coupler Figure 27: Output Phase Difference for 180° Hybrid Coupler
The tune feature was used to tune the lengths and widths of the transmission lines – the values seen in the table to represent the widths of the different transmission lines based on the ideal impedances that will be tuned.
With the addition of non-ideal corners, junctions and bends, the parasitic impedances of the lines have been changed, resulting in a shifted band of operation and relatively poor performance when compared to the ideal model.
Port1
Port2
Po
rt3
Port4
P=3.93376mm
W=1.97623mm
12
3
W1=1.97953mm
W2=1.97623mm
W3=0.549439mm
P=
7.8
67
51
mm
W=
0.5
49
43
9m
m
P=3.93376mm
W=1.97953mm
12
34
W1=1.97953mmW2=1.05529mm
W3=0.653787mmW4=1.05529mm
P=5.90063mm
W=0.653787mm
P=3.93376mm
W=1.05529mm
W1=0.653787mmW2=0.653787mm
ANG=35degR=1e-020mm
W=1.05529mm
ANG=55degR=1e-020mm
W=1.05529mm
AN
G=
55
de
gR
=1
e-0
20
mm
W=
1.0
55
29
mm
AN
G=
35
de
gR
=1
e-0
20
mm
W=
1.0
55
29
mm
W=
0.5
49
43
9m
m
W=
0.6
53
78
7m
m
P=3.93376mm
W=1.05529mm
12
3
W1=1.63679mm
W2=1.05529mm
W3=1.97623mm
1 2
3
W1=1.05529mm
W2=1.63679mm
W3=1.97623mm
P=
3.9
33
76
mm
W=
1.9
76
23
mm
P=
3.9
33
76
mm
W=
1.9
76
23
mm
P=3.93376mm
W=1.63679mm
P=
7.8
67
51
mm
W=
2.6
33
38
mm
P=3.93376mm
W=0.833586mm
P=3.93376mm
W=0.833586mm
P=
7.8
67
51
mm
W=
1.0
54
33
mm
P=3.93376mm
W=1.63679mm
P=3.93376mm
W=1.97623mm
P=
1.9
66
88
mm
W=
0.6
53
78
7m
m
1 2
3
W1=1.63679mm
W2=0.833586mm
W3=2.63338mm
12
3
W1=0.833586mm
W2=1.63679mm
W3=2.63338mm
12
3
W1=1.97623mm
W2=0.833586mm
W3=1.05433mm
W1=0.833586mm
W2=1.05433mm
Port1
Port2
Port3
Port4
Variable Value [mm] Value [mil]
$wfifty 1.97623 77.8043307
$wZ1 1.06491 41.9255905
$wZ2 1.63679 64.4405511
$wZ3 2.63338 103.676378
$wZ4 0.833586 32.8183464
$wZ5 1.05433 41.5090551
$wZ6 0.653787 25.7396456
$wZ7 1.97953 77.9342519
$wZ8 0.549439 21.6314567
Component Progress - Knöchel Hybrid
Joel
Designer (Second Cut 120 mil)
Figure 28: Schematic View of Designer Hybrid Figure 29: Layout View of Designer Hybrid
Figure 30: S-Parameters of 180° Hybrid Coupler Figure 31: Output Phase Difference for 180° Hybrid Coupler
The tune feature was used to tune the lengths and widths of the transmission lines – the values seen in the table to the left were found to give the best performance
Although not as perfect as the ideal results, the simulated results are quite good relative to the initial trial – we have approximately equal power division between ports 2 and 3 and there is a normalized phase difference error of approximately +/- 0.4° between ports 2 and 3.
Po
rt1
Port2
Port3
Po
rt4
P=619.489mil
W=$wf if ty
P=154.872mil
W=$wZ7
P=$LZ1
W=$wZ1
P=25mil
W=$wZ6
P=$LZ1
W=$wZ1
P=$LZ4
W=$wZ2
P=
$L
Z3
W=
$w
Z3
P=$LZ4
W=$wZ4
P=
$L
Z3
W=
$w
Z5
P=$LZ4
W=$wZ2
P=$LZ4
W=$wZ4
12
3
W1=$wZ7
W2=$wf if ty
W3=$wZ8
12
3
W1
=$
wfift
y
W2
=$
wZ
1
W3
=$
wZ
2
12
3
W1=$wZ4
W2=$wZ2
W3=$wZ3
12
3
W1=$wf if ty
W2=$wZ4
W3=$wZ5
1 2
3
W1=$wZ2
W2=$wZ4
W3=$wZ31
2
3
W1
=$
wZ
1
W2
=$
wfift
y
W3
=$
wZ
2
12
3
4
W1=$wZ7
W2=$wZ1
W3=$wZ6
W4=$wZ1
W1=$wZ4
W2=$wZ5
P=
61
9.4
89
mil
W=
$w
fift
y
P=
61
9.4
89
mil
W=
$w
fift
y
P=619.489mil
W=$wf if ty
P=
$L
Z8
W=
$w
Z8
P=37mil
W=$wZ6
P=$wZ2/2
W=$wZ1
P=$wZ2/2
W=$wZ1
ANG=90deg
R=0.001mil
W=$wZ6
AN
G=
18
0d
eg
R=
12
mil
W=
$w
Z6
P=25mil
W=$wZ6
P=
70
mil
W=
$w
Z6
ANG=45deg
R=12mil
W=$wZ6
Variable Value [mm] Value [mil]
$wfifty 1.97623 77.8043307
$wZ1 0.8382 33
$wZ2 2.0828 82
$wZ3 2.1082 83
$wZ4 0.508 20
$wZ5 0.8636 34
$wZ6 0.9144 36
$wZ7 2.0574 81
$wZ8 0.5588 22
Component Progress - Schiffman Phase Shifter
Mia
The original Schiffman phase shifter design performed in Week 2 was scrapped completely and had to be redesigned. There was a misunderstanding of the purpose of the phase shifter and how it functions within the entire circuit. After a meeting with Michael Enders at Anaren on 1/9/2009, the Schiffman phase shifter was re-designed from scratch.
Ideal
Figure 32: Schematic View of Ideal Schiffman Phase Shifter
Figure 33: Ideal Phase Difference Figure 34: Ideal Reflection and Transmission Losses
The ideal model was developed using Schiffman’s original paper published in the IEEE Microwave Techniques Journal. B.M. Schiffman, “A New Class of Broadband Microwave 90-Degree Phase Shifters,” IRE Trans. Microwave Theory Tech., vol. MTT-6, no. 4, pp.232-237.
The results are: Over 10 – 12GHz, the phase difference between the reference line and the Shiffman phase shifter is 45° flat, with a return loss of -34.88dB at best between Markers 1 and 3.
From the ideal model, even and odd mode impedances are extracted to be used in the real model. Zoe=66Ω and Zoo=42Ω
Port1Port2
Port3Port4
ZE=66ZO=42
E=90degF=11GHz
E
E=(180+45) deg
F=11GHz
Z=50
Component Progress - Schiffman Phase Shifter
Mia
Designer
Because of the nature of the Shiffman phase shifter, the design cannot be modeled in Designer and then verified in HFSS. Attempts to model the phase shifter in Designer first have lead to failures. After speaking with Michael Enders at Anaren, he suggested from his experience that the design be completed in HFSS, and Designer used to obtain starting values.
As a sanity check, the ideal model was converted to a physical model with the chosen substrate of 121.5mil.
Figure 35: Schematic view of semi-ideal phase shifter. Figure 36: Variable values for the phase shifter.
Figure 37: Phase difference of semi-deal shifter. Figure 38: Reflection and transmission.
From the figures above, it can be seen that the substrate does have some effect on the results. The phase difference is a bit flatter, and the return loss becomes a bit worse, with best case return loss of – 31.07dB. The results above are the best possible results, and the model will be exported to HFSS and modified.
Po
rt1
Po
rt2
Po
rt3
Po
rt4
P=387.181mil
W=78.7771mil
S=S
W=W
P=L
Component Progress - Schiffman Phase Shifter
Mia
HFSS
The Schiffman phase shifter is designed primarily in HFSS. The ideal design does not take into account the thin strip of copper connecting the two coupled lines, and is nearly impossible to model in Designer. Also, port locations will become significant in the HFSS design. The design is still in progress, and results so far are unusable.
Figure 39: 3D view of the Schiffman phase shifter, with chamfered edges.
System Progress - System A
Mia
Ideal
Figure 40: Circuit layout of the ideal System A.
Figure 41: Transmission from Port 1 to all outputs. Figure 42: Phase difference between output ports when Port 1 is input.
Po
rt1
Po
rt2
Po
rt3
Po
rt4
Port5
Port6
Port7
Port8
E
E=90deg
F=11GHz
Z=50
E
E=90deg
F=11GHz
Z=b
E
E=
90
de
g
F=
11
GH
z
Z=
a
E
E=90deg
F=11GHz
Z=50
E
E=90degF=11GHz
Z=50
E
E=90degF=11GHz
Z=50
E
E=90deg
F=11GHz
Z=b
E
E=90degF=11GHz
Z=b
E
E=90degF=11GHz
Z=b
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=
90
de
g
F=
11
GH
z
Z=
a
E
E=
90
de
g
F=
11
GH
z
Z=
50
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=90degF=11GHz
Z=a
E
E=
90
de
g
F=
11
GH
z
Z=
50
E
E=
90
de
g
F=
11
GH
z
Z=
50
E
E=
90
de
g
F=
11
GH
z
Z=
50
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=90deg
F=11GHzZ=b
E
E=90degF=11GHz
Z=a
E
E=
90
de
g
F=
11
GH
z
Z=
50
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=90deg
F=11GHzZ=a
E
E=
90
de
g
F=
11
GH
z
Z=
50
E
E=
90
de
g
F=
11
GH
z
Z=
50
E
E=
90
de
g
F=
11
GH
z
Z=
50
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=90degF=11GHz
Z=b
E
E=90deg
F=11GHzZ=a
E
E=90deg
F=11GHzZ=50
E
E=90deg
F=11GHzZ=b
E
E=
90
de
g
F=
11
GH
z
Z=
a
E
E=90deg
F=11GHzZ=50
E
E=90deg
F=11GHz
Z=50
E
E=90deg
F=11GHz
Z=50
E
E=90deg
F=11GHzZ=b
E
E=90deg
F=11GHz
Z=b
E
E=90deg
F=11GHz
Z=b
E
E=
90
de
g
F=
11
GH
z
Z=
b
E
E=
90
de
g
F=
11
GH
z
Z=
a
ZE=66ZO=42E=90degF=11GHz
ZE=66ZO=42
E=90degF=11GHz
E
E=(180+45) deg
F=11GHz
Z=50
E
E=(180+45) degF=11GHz
Z=50
System Progress - System A
Mia
Figure 43: Reflection at all inputs. Figure 44: All S parameters.
The above results are the best possible using all ideal components.
The phase difference at the outputs when Port 1 is excited is 45° ±0.1°.
The best possible return loss is -23.18dB. All four outputs have about -6dB ±0.3dB amplitude, meaning power is being split relatively equal between the four outputs, with each getting a quarter of the total input power
System Progress - System B
Amanda
Designer
Figure 45: Schematic view of System B Figure 46: Layout view of System B
Figure 47: Simulation results for input port 1 Figure 48: Simulation results for input port 2
Figure 49: Simulation results for input port 3 Figure 50: Simulation results for input port 4
Simulation results show that there is a large amount of reflection at S55 and S66, these require the 180 degree Hybrid Coupler to be readjusted S78 also needs improved isolation, these will be re-evaluated after a new revision of the Hybrid Coupler is implemented
Port1
Port4
Port6
Port7
Port10
Port11
Port15
Port16