Professional Documents
Culture Documents
LO
Multi-Channel RF_LINK IN
IN
OUT
LO
LO
OUT
M1 {Mixer@Data Flow Models}
Amplifier
Amplifier
A2 {Amplifier@Data Flow Models} F1 {CustomFIR@Data Flow Models}
(Dataflow)
ConvGain=0 IN
EnableNoise=YES
NoiseFigure=0
M1 {Mixer@Data Flow Models}
Sideband=Lower
ConvGain=0IN
SOIout=1.0e17 W
EnableNoise=YES
M1 {Mixer@Data
TOIout=1.0e17
OUT
LO
OUT
Flow Models}
W ConvGain=0
GainUnit=voltageAmplifier
Gain=1
A2 {Amplifier@Data Flow Models}
GainUnit=voltage
Gain=1
Amplifier
A2 {Amplifier@Data Flow Models}
F1 {CustomFIR@Data Flow Models}
RF Architecture Layer
(Spectrasys)
Dpath length
a d
ttotal
delay between phasefront
arrival at first element, vs.
arrival at Nth element
t1 t2
a d
D Phase @Nth element : e-jNDa
• Distances normalized to the spacings (l/2 at Fcenter)
• Constant phase shift at all frequencies in BW
• Easier to implement in DSP/RF, but narrowband
ttotal
delay between phasefront
arrival at first element, vs. Steer array to the side by a=10o :
arrival at Nth element
D/A
X SW/
DPX
Precoder A/D X
FBB
𝐹11 𝐹12
𝐹21 𝐹22
⋮
⋯
⋱
𝐹1𝑁
𝐹2𝑁
⋮
D/A
X SW/
• Dedicated ADC/DAC per channel
• High hardware complexity
DPX
𝐹𝑀1𝐹𝑀2 ⋯ 𝐹𝑀𝑁 A/D X
• High power consumption (DC, heat)
…
…
Combiner RF Chains
NS WBB • “N” independent RF up/downconversions
𝑤11 𝑤12 𝑤1𝑁
𝑤21 𝑤22
⋮
⋯
⋱
𝑤2𝑁
⋮ D/A
X
• Large beamforming gains
𝑤𝑀1𝑤𝑀2 ⋯ 𝑤𝑀𝑁 SW/
A/D X
DPX
• Enables multi-stream and multi-user
• Flexible transmission modes
D/A
X
…
• Single baseband signal
RF precoder : FRF • Suitable for single stream (ie - not MIMO)
…
• Reduced power consumption and complexity
RF combiner : WRF
NS • Single up/downconversion for mm-wave
• Signal processing gain in RX (SNR “improves”)
A/D
X • Analog phase shifters
…
D/A
X + Phases applied at RF (for each beam direction)
…
Precoder
FBB Multiple beams summed at RF (each antenna)
RF Chains
…
…
𝐹11 𝐹12 𝐹1𝑁
⋯
• Reduced hardware complexity
𝐹21 𝐹22 𝐹2𝑁 FRF
⋮ ⋱ ⋮
𝐹𝑀1𝐹𝑀2 ⋯ 𝐹𝑀𝑁
NS
MIMO • Similar performance to DBF architecture
encoder
D/A
X
• Being investigated by 5G
…
weights
positions
“Ringing”
No
window Unwanted
energy waste
(or leakage)
2-D
“Taylor”
window Sidelobe
Suppression
from -12.8dBc
to -27.5dBc
Wider beamwidth
from 12.8o to 16.1o © Keysight Technologies 2016 Page 15
Effect of element spacing:
8x8 array with elements at 0.5l, 1.0l, 2.0l, 4.0l separation
Raw result
(No taper)
With Taylor
taper
•
L=0 dB [taper(1)] RFAmp_1 {RFAMP}
•
Port_3 {*OUT}
•
NF=3 dB10 [nf]
Port_6 {*OUT}
L=0 dB [taper(5)] RFAmp_5 {RFAMP} A=45 ° [phase(5)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_7 {*OUT}
L=0 dB [taper(6)] RFAmp_6 {RFAMP} A=122.94 ° [phase(6)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_8 {*OUT}
L=0 dB [taper(7)] RFAmp_7 {RFAMP}A=200.885 ° [phase(7)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_9 {*OUT}
L=0 dB [taper(8)] RFAmp_8 {RFAMP}A=278.824 ° [phase(8)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf]
MultiSource_1 {MultiSource}
Port_10 {*OUT}
S1=CW: 1000 MHz at 0 dBm
L=0 dB [taper(9)] RFAmp_9 {RFAMP} A=90 ° [phase(9)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_11 {*OUT}
L=0 dB [taper(10)] RFAmp_10 {RFAMP}A=167.94 ° [phase(10)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_12 {*OUT}
L=0 dB [taper(11)] RFAmp_11 {RFAMP} ZO=50 Ω
A=245.885 ° [phase(11)]
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_14 {*OUT}
L=0 dB [taper(13)] RFAmp_13 {RFAMP} A=135 ° [phase(13)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_15 {*OUT}
L=0 dB [taper(14)] RFAmp_14 {RFAMP}A=212.94 ° [phase(14)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_16 {*OUT}
L=0 dB [taper(15)] RFAmp_15 {RFAMP} ZO=50 Ω
A=290.885 ° [phase(15)]
G=20 dB10 [gain]
NF=3 dB10 [nf] © Keysight Technologies 2016 Page 28
Port_17 {*OUT}
L=0 dB [taper(16)] RFAmp_16 {RFAMP} ZO=50 Ω
A=368.824 ° [phase(16)]
G=20 dB10 [gain]
NF=3 dB10 [nf]
RF System: improvements to support array architectures
Simulator now abstracted for “Phased Array” analysis type
Port_2 {*OUT}
A=0 ° [phase(1)] ZO=50 Ω
• •
Port_3 {*OUT}
Port_6 {*OUT}
L=0 dB [taper(5)] RFAmp_5 {RFAMP} A=45 ° [phase(5)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_7 {*OUT}
L=0 dB [taper(6)] RFAmp_6 {RFAMP} A=122.94 ° [phase(6)]
ZO=50 Ω
G=20 dB10 [gain] Tx
NF=3 dB10 [nf] Rx
Port_8 {*OUT} RFAmp_1 {RFAMP}
ArrayPort1 {ArrayPort}
L=0 dB [taper(7)] RFAmp_7 {RFAMP}A=200.885 ° [phase(7)]
ZO=50 Ω RxTx=Tx G=20 dB10 [gain]
G=20 dB10 [gain] NF=3 dB10 [nf]
Freq=1000 MHz ArraySplit1 {ArraySplit} ArrayAttn1 {ArrayAttn} ArrayPhase1 {ArrayPhase} ArrayAnt2 {ArrayAnt}
NF=3 dB10 [nf] InsertionLoss=0.0688 dB InsertionLoss=0 dB OP1dB=60 dBm InsertionLoss=0 dB Configuration=Uniform Rectangular Array
Mode=SubArray Window=None [taperOption] OIP3=70 dBm CalcMode=Auto NumElementsX=4
Port_9 {*OUT} NumRows=4 Theta=30 ° NumElementsY=4
L=0 dB [taper(8)] RFAmp_8 {RFAMP}A=278.824 ° [phase(8)]
ZO=50 Ω NumCols=4 Phi=60 ° DistanceUnit=Wavelengths
G=20 dB10 [gain] DistanceX_in_Wavelengths=0.5
NF=3 dB10 [nf] DistanceY_in_Wavelengths=0.5
MultiSource_1 {MultiSource} ActiveLoading=None
Port_10 {*OUT} RxTx=Rx
S1=CW: 1000 MHz at 0 dBm
L=0 dB [taper(9)] RFAmp_9 {RFAMP} A=90 ° [phase(9)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf] Advantages
•
Port_11 {*OUT}
L=0 dB [taper(10)] RFAmp_10 {RFAMP}A=167.94 ° [phase(10)]
G=20 dB10 [gain]
NF=3 dB10 [nf]
ZO=50 Ω
Easy to draw
L=0 dB [taper(11)] RFAmp_11 {RFAMP}
Port_12 {*OUT}
ZO=50 Ω
A=245.885 ° [phase(11)]
• Topology easy to change
• Fast and “tunable”
G=20 dB10 [gain]
NF=3 dB10 [nf]
Port_13 {*OUT}
Port_14 {*OUT}
L=0 dB [taper(13)] RFAmp_13 {RFAMP} A=135 ° [phase(13)]
G=20 dB10 [gain]
ZO=50 Ω • Monte Carlo and failure analysis in minute(s)
NF=3 dB10 [nf]
Port_15 {*OUT}
• Can use X-parameters, measured filters, etc
•
L=0 dB [taper(14)] RFAmp_14 {RFAMP}A=212.94 ° [phase(14)]
ZO=50 Ω
G=20 dB10 [gain]
NF=3 dB10 [nf] Line-up remains compatible with regular Spectrasys
Port_16 {*OUT}
L=0 dB [taper(15)] RFAmp_15 {RFAMP} ZO=50 Ω
A=290.885 ° [phase(15)]
G=20 dB10 [gain]
NF=3 dB10 [nf] © Keysight Technologies 2016 Page 29
Port_17 {*OUT}
L=0 dB [taper(16)] RFAmp_16 {RFAMP} ZO=50 Ω
A=368.824 ° [phase(16)]
G=20 dB10 [gain]
NF=3 dB10 [nf]
RF System: improvements to support array architectures
Differences in modeling TX vs. RX configurations
SW1 {SWITCH_Linear2} L
NF=2.5 dB10
InsertionLoss=1 dB
Window=Taylor
SideLobeLevel=-20 dB10
InsertionLoss=1 dB
Mode=SubArray
NumRows=16 [NumEl]
Transmit Case
IL=0.5 dB
1
Lim_1 {MOD_Limiter}
IL=0.2 dB10
Pmax=10 dBm 1
NumBars=2 NumCols=16 [NumEl]
TX 1 input stimulus, “N” channels out:
Rx
ArrayPort2 {ArrayPort}
ArrayAnt1 {ArrayAnt} RxTx=Tx [ArrayState]
SW2 {SWITCH_Linear2} ArrayPhase1 {ArrayPhase}
Configuration=Uniform Rectangular Array Freq=10 GHz [FRxTx]
IL=0.5 dB InsertionLoss=1 dB
NumElementsX=16 [NumEl]
NumElementsY=16 [NumEl] CalcMode=Auto
ActiveLoading=None FinalAmp {RFAMP} DriverAmp {RFAMP} Theta=20 ° [beamTheta]
RxTx=Tx [ArrayState] G=20 dB10 G=15 dB10 Phi=0 ° [beamPhi]
NF=5 dB10 NF=3 dB10
LNA {RFAMP}
G=20 dB10
NF=2.5 dB10 ArrayAttn2 {ArrayAttn} ArraySplit1 {ArraySplit}
SW1 {SWITCH_Linear2}
IL=0.5 dB
L
InsertionLoss=1 dB
Window=Taylor
SideLobeLevel=-20 dB10
NumBars=2
InsertionLoss=1 dB
Mode=SubArray
NumRows=16 [NumEl]
NumCols=16 [NumEl] RX Receive Case
“N” input stimuli, 1 channel out:
Lim_1 {MOD_Limiter}
1 IL=0.2 dB10
Pmax=10 dBm 1
2 Rx
ArrayPort2 {ArrayPort}
ArrayAnt1 {ArrayAnt} RxTx=Rx [ArrayState]
SW2 {SWITCH_Linear2} ArrayPhase1 {ArrayPhase}
Configuration=Uniform Rectangular Array
Pre-calculated array
measurements
• Gain/Temp (receivers)
• EIRP (transmitters)
• Gain, Directivity
at (fmeas,qmeas)
Initial analysis of operating point, by node Consult device look-up table Estimate PAE for each path
for Power PAE %
S
Aggregate
PAE on each
path to the
Array level
Array of “N”
Amplifier
Amplifiers
[]
Control
Amplifier
Amplifier
[][]
I Phase O Amplifier [][] []
Amplifier
Shifter [][]
Amplifier
Amplifier
M28 {PhaseShifter_M@Data Flow Models} M22 {Amplifier_M@Data Flow Models} C7 {Combiner_M@Data Flow Models} UE3 {DigitalDemodE
PhaseShift=(64x1) [0; 124.966; -110.06; 14.905; … GainUnit=dB Mode=FullArray Start=800e-9 s [Start_
…
Gain=0 SymbolRate=250
ResultLe
ModTyp
MeasurementFilte
ReferenceFilte
Alph
Dataflow T&M
Algorithmic
(dynamic signals)
RF_LINK
Spectrasys
Phased Array
(static model)
[S] [X]
Farfield
Primary RF Tools Coupling [S]
Dynamic,
Beamformed
Platforms
RF
Architectural limitations
• Gain/phase quantization effects
• Squinting
• Timing skew
Interference
• Self-interference, from multi-uesrs
• External interferers & Jamming
• Imperfect element
patterns express
themselves at the
beam level
ArrayCoupling_M
ArrayAnt3 {ArrayAnt}
Configuration=Uniform Rectangular Array Loaded with 8x8 S-parameters
ActiveLoading=S-Parameter Dataset
DatasetName=8x8_array_touchstone.s64p
from ADS Momentum
RxTx=Tx
Phase Time
Shifters Delays
ArrayDelay1 {ArrayDelay}
ArrayPhase1 {ArrayPhase}
Phase Time
Shifters Delays
• “Oversteers” the angle for wider signal BW • Consistent boresight angle for all signal BWs
• Shifted sidelobes • Symmetric sidelobes
• As frequency increases, the beam narrows and sidelobes change watch for interference/susceptibility
• With chirped Radar systems, the beam “wiggles” during the pulse
5GHz
15GHz
© Keysight Technologies 2016 Page 44
Squinting: An alternative lower-cost architecture
Two-stage design: Coarse (delay) and Fine (phase) stages
STAGE 2 STAGE 1
Conventional Coarse
Gain/Phase Delay
2-stage design provides most of the benefit at lower cost, allowing a conventional DSP architecture
Dataflow
Control
Amplifier I Phase O
Shifter
Env
A2 {Amplifier_M@Data Flow Models} P1 {PhaseShifter_M@Data Flow Models} D10 {DelayEnv_M@Data Flow Models}
GainUnit=voltage PhaseShift=0 ° Quantization=YES
Quantization=YES InsertionLoss=0 dB10 NumBits=6
NumBits=6 Quantization=YES
StepSize=0.5 s
StepSize=0.5 dB10 NumBits=6
MaxGain=10 dB10 Sensitivity=1 rad
NoiseFigure=0 HilbertFilterLength=64
GCType=dBc1
dBc1out=-10 dBm
RF
ArrayAttn1 {ArrayAttn} ArrayPhase1 {ArrayPhase}
InsertionLoss=1 dB InsertionLoss=1 dB ArrayDelay1 {ArrayDelay}
InsertionLoss=1 dB
Window=Taylor CalcMode=Auto
CalcMode=Auto
SideLobeLevel=-30 dB10 Theta=0 ° Theta=0 °
NumBars=2 Phi=0 ° Phi=0 °
Quantization=YES Quantization=YES Quantization=YES
NumBits=6
NumBits=6 NumBits=6
StepSize=1 ns
StepSize=0.5 dB10 Error=None Error=None
Dataflow
Control
Amplifier I Phase O
Shifter
Env
A2 {Amplifier_M@Data Flow Models} P1 {PhaseShifter_M@Data Flow Models} D10 {DelayEnv_M@Data Flow Models}
GainUnit=voltage PhaseShift=0 ° Quantization=YES
Quantization=YES InsertionLoss=0 dB10 NumBits=6
NumBits=6 Quantization=YES
StepSize=0.5 s
StepSize=0.5 dB10 NumBits=6
MaxGain=10 dB10 Sensitivity=1 rad
NoiseFigure=0 HilbertFilterLength=64
GCType=dBc1
dBc1out=-10 dBm
RF
ArrayAttn1 {ArrayAttn} ArrayPhase1 {ArrayPhase}
InsertionLoss=1 dB InsertionLoss=1 dB ArrayDelay1 {ArrayDelay}
InsertionLoss=1 dB
Window=Taylor CalcMode=Auto
CalcMode=Auto
SideLobeLevel=-30 dB10 Theta=0 ° Theta=0 °
NumBars=2 Phi=0 ° Phi=0 °
Quantization=YES Quantization=YES Quantization=YES
NumBits=6
NumBits=6 NumBits=6
StepSize=1 ns
StepSize=0.5 dB10 Error=None Error=None
22
Phase Shift 14
Variances
directivityCutSweep (dB)
-2
-10
-18
-26
-34
Atten/Gain 6
directivityCutSweep (dB)
-2
Variances -10
-18
-26
-34
-42
-50
-90 -72 -54Directivity
-36 Distribution
-18 from
0 Element
18 Failures
36 54 72 90
30 Angle (deg)
directivityCutSweep
22
14
6
10dB SLL
directivityCutSweep (dB)
Element -10
-2
Failures -18
-26
-34
-42
-50
-90 -72 -54 -36 -18 0 18 36 54 72 90
Angle (deg)
directivityCutSweep
22
Phase Shift 14
Variances
directivityCutSweep (dB)
-2
-10
-18
-26
-34
22-50 -90
14
-72 -54
10 dB SLL
-36 -18 0
Angle (deg)
directivityCutSweep
18 36 54 72 90
Atten/Gain 6
directivityCutSweep (dB)
Variances -10
-2
variation
-18
-26
-34
-42
-50
-90 -72 -54Directivity
-36 Distribution
-18 from
0 Element
18 Failures
36 54 72 90
30 Angle (deg)
directivityCutSweep
22
14
directivityCutSweep (dB)
Element -10
-2
Failures -18
-26
-34
-42
-50
-90 -72 -54 -36 -18 0 18 36 54 72 90
Angle (deg)
directivityCutSweep
Rules of thumb
• 1mm = 5ps (coax)
• 18o @10GHz = 5ps
Uncalibrated Calibrated
8-channel T&M 8-channel T&M © Keysight Technologies 2016 Page 52
Phased Array Impairments – Nonlinear Interference from Multiple users
4 modulated 4 beamforming Beam measure: De-mod
baseband signals TX angles (f,q) ideal vs. distorted each signal
Signal 1
Signal 2 DBF
Signal 4
Distorted Beam
© Keysight Technologies 2016 Page 54
Phased Array Impairments – Nonlinear Interference scenarios
Subnetwork2 {DirectionalSource}
THETA=20 [IncidentTHETA]
PHI=0
RECEIVE
NumElements=4 [RXarrayEdge]
Fcenter=28 [FCarrier]
ARRAY
DESIRED 20
Subnetwork1 {QAM_source}
ModType=4
GainDB=0 [SigPower]
Env
CarrierFreq=28 [FCarrier] Full RX DE-MODD3 {DigitalDemodEnv@Data Flow Models}
SymbolRate=6 MHz
A1 {AddEnv_M@Data Flow Models}
20 ResultLength=200
ModType=256-QAM
MeasurementFilter=Root Raised Cosine
Subnetwork5 {DirectionalReceiver} ReferenceFilter=Raised Cosine
Dig Demod Alpha=0.5
THETA=20 [ReceiverTHETA]
PHI=0 SaveSymbolResults=NO
NumElements=4 [RXarrayEdge] ContinuousMode=YES
Subnetwork7 {DirectionalSource}
Fcenter=2.4
THETA=0 [InterfererTHETA]
Bits_Gain=6
PHI=0 S2 {SpectrumAnalyzerEnv@Data Flow Models}
Bits_Phase=6 Spectrum Analyzer
NumElements=4 [RXarrayEdge] Mode=ResBW
Subnetwork6 {JammerCW}
Foffset=6 [Foffset]
InterfererPower=-60 [InterfererPower]
Interferer
Amplitude
& Freq Offset
Interferer
position
2D Beam Cut
ARRAY
CONFIG
EVM
DIGITAL
DE-MOD
3D BEAM
SPECTRUM SIDELOBES
& NULLS
DEMO
Document R4-166116
“Consideration on ACLR for 5G NR”
Submitted by CMCC
ftp://ftp.3gpp.org/tsg_ran/WG4_Radio/
……../TSGR4_80/Docs/R4-166116.zip
– The Application of Taylor Weighting, Digital Phase Shifters, and Digital Attenuators to
Phased-Array Antennas, Billy Brock,
http://www.sandia.gov/RADAR/publications/index.html
www.keysight.com/find/eesof-how-to-videos
© Keysight Technologies 2016 Page 72