ICC2013 Co-located Workshop

Jinbiao Xu, author

Sr. Applications Engineer

Agilent Technologies

Daren McClearnon, speaker

System-Level EDA Prod Mktng Mgr.,

Agilent Technologies

ICC2013 Co-located Workshop

© 2013 Agilent Technologies

“A Novel Wideband Digital Pre-Distortion (DPD)

Modeling Platform using SystemVue”

2

1. Introduction and Problem Statement

2. Digital Pre-Distortion (DPD) Concepts

3. DPD verification with Agilent Hardware

4. DPD simulation with Agilent EDA Tools

5. Crest Factor Reduction (CFR)

6. Summary

Agenda

“Wideband DPD Modeling for 4G””© 2013 Agilent Technologies

• Modern communication systems:• Signals have high peak-to-average power ratios (PAPR).• Must operate with high power-added efficiency (PAE).

• High PAPR is a consequence of high spectral efficiency• Multiple-Carrier Signals (MC GSM, MC WCDMA)• CDMA (WCDMA, CDMA2000)• OFDM (LTE, WiMAX)

• High PAE is achieved when the RF power amplifier (PA) is driven towards saturation

• Operation near saturation inherently results in higher signal distortion

Digital Pre-Distortion (DPD): Problem Statement

3


DPD Problem Statement

4

How to handle signals with high PAPR, while driving the PA to operate with high PAE, while also having low signal distortion?

Conflicting requirementsConflicting requirements

Higher

DC-RF

Efficiency

Higher

Peak

Power

Higher

Spectral

Efficiency

IncreaseDrive levels

Causes high Causes high distortion

levels

“Back off” the drive

levels


DPD Solution Approach

5

Solution: Preconditioning the signal (CFR) and correcting for the hardware (DPD) will both be discussed in this presentation

Higher

Spectral

Efficiency

Higher throughput levels for

subscribers

Higher

DC-RF

Efficiency

Higher

Peak

Power

IncreaseDrive levels

Causes high Causes high distortion

levels

CFRCFR

DPDDPD


6






6. Summary

Agenda


7

Psat

Pin

LINEAR GAIN

INPUT

POWER

OUTPUT

POWER

Pdesired

Pactual

Pin needed

to achieve

Pdesired

PA, WITH GAIN

COMPRESSION

Digital Pre-distortion principles – compressing PA


8

Digital Pre-distortion principles – pre-expansion

Maximum

correctable

power

Psat

LINEAR GAIN

INPUT

POWER

OUTPUT

POWER

LINEAR REGION

DPDREGION

PA, WITH GAINCOMPRESSION

DPD GAIN

EXPANSION

+


9

Maximum

correctable

power

Psat

INPUT

POWER

OUTPUT

POWER

LINEAR REGION

DPDREGION

LINEARIZED

DPD + PA

Digital Pre-distortion principles – linearized result

PA, WITH GAINCOMPRESSION

DPD GAINEXPANSION

+ =


10

Linear Operation with time-varying envelope

Psat

LINEAR GAIN

INPUT

POWER

OUTPUT

POWER

Peak-to-Avg Power Ratio (PAPR)


11

Nonlinear Operation – peaks are compressed

Psat

LINEAR GAIN

INPUT

POWER

OUTPUT

POWER

(compressed peaks)

CCDF (LTE)

Peak-to-Avg Power Ratio (PAPR)


12

DPD Pre-Expansion – peaks are exaggerated

Psat

LINEAR GAIN

INPUT

POWER

OUTPUT

POWER

Possible Improvements

• Compensate for artificially higher avg. signal power

• Crest Factor Reduction (CFR)

(expanded peaks)


13

DPD Net Result: Linear gain of complex-valued RF

carrier envelope over a specific range of power levels

LINEAR

INPUT

POWER

OUTPUT

POWER

DPD pre-expanded peaks

INPUT

POWER

PA compresses peaksLINEAR

Baseband Digital Pre-Distortion

RF Power Amplification


What does a DPD look like? (Volterra Model)

14

∑=

=K

k

k nznz1

)()( ∑ ∑ ∏= = =

−=

Q

m

Q

m

k

l

lkkk

k

mnymmhnz0 0 1

1

1

)(),,()( LL

∑∑∑= ==

+−−+−+=

Q

m

Q

m

Q

m

mnymnymmhmnymhhnz0 0

21212

0

1110

1 21

)()(),()()()( K

Volterra series pre-distorter can be described by

where

Which is a 2-dimensional summation of power series & past time envelope responses

A full Volterra produces a huge computational load. People usually simplify it into

• Wiener model • Hammerstein model • Wiener-Hammerstein model• Memory polynomial model


15

DPD principles – Memory Polynomial Model

If only diagonal terms are kept, Volterra reduces to “Memory polynomial” model.

Agilent uses the “Indirect Learning” algorithm to extract MP coefficients.

You can now add your own model, extraction algorithm, and even your own GUI.

L. Ding, G. T. Zhou, D. R. Morgan, Z. Ma, J. S. Kenney, J. Kim, and C. R. Giardina, “Memory

polynomial predistorter based on the indirect learning architecture,” in Proc. of GLOBECOM, Taipei, Taiwan, 2002, vol. 1, pp. 967–971.

∑∑= =

−−−=

K

k

Q

q

k

kq qnyqnyanz1 0

1)()()(

Where• K is Nonlinearity order • Q is Memory length


16

Agenda






6. Summary


Generalized Wireless Transmitter Path

BB PHY

CFR DPDUp

convertPA

EnvTracking

DAC

Downconvert

ADCAdapt

Duplexer

• Which blocks are included with your final product?• What IP do you have access to? Or, are able to imitate? Able to modify?• What final system specifications do you need to test against?

17

?

??

?

??


Vector Signal

Analyzer

MXA, PXA, Modular

Agilent Measurement-based DPD Modeling Platform

BB TX PHY

CFRDPD

model

Upconvert

PADAC

Downconvert

ADCGenerate

Coefficients

Vector Signal

Generator

AWGESG, MXG, PSG

W1716

DPD

Step-by-Step GUI

W1918

LTE-A

IP Library

BB RXPHY

Throughput

BER/FER

ACPR

EVM

89600 VSA

Optional Reference RX

W1461 SystemVue

Also:

3G, WLAN60GHz,

DVB, OFDM

18


Measurement-Based DPD Modeling Flow

Method 1 – Measure both PA Input and Output signals

Get baseband complex waveforms of

PA input and output

Extract DPD Model (includes delay estimation

and adjustment)

Apply DPD Model, and Get DPD+PA Response

Verify DPD Performance

Create DPD Stimulus

RF Input RF Output

11

55

44

33

22

Adjust current to

control switch

Switch

MXA / PXA

MXG

Power Splitter

DC Power Analyzer

DUT

THRU

19


20

Measurement-Based DPD Modeling Flow

Method 1 – Measure both PA Input and Output signals

• DPD flow consists of 5 steps in SystemVue • Convergence improves with more iterations • 2-3 iterations are typical for real PAs


PA input and output


and adjustment)

Apply DPD Model, and Get DPD+PA Response


Create DPD Stimulus

RF Input RF Output

11

55

44

33

22


Measurement-Based DPD Modeling Simplification:

Calculated PA Input, Measured PA Output

Single connection allows automation, iterationsEliminates one measurement, physically fasterIdentical extraction algorithms, verification process

MXA / PXAMXG

DUT

21



PA input and output


and adjustment)

Apply DPD Model, andGet DPD+PA Response


Create DPD Stimulus

Simulated

BB Input

Meas RF

Output

11

55

44

33

22

22

Measurement-Based DPD Modeling Simplification:

Calculated PA Input, Measured PA Output

• Uses the Ideal BB stimulus waveform vs. measured PA output waveform to extract the DPD model.

• Advantages:- Single connection- PA remains “ON”- Easier to automate- Faster speed

• Is typical of industry practice today

• Linearizes the entire system, not just the PA

• Provides very acceptable accuracy for quick Evaluation and MFG Test applications.

• Assumptions:- Source flatness- Source linearity- No additional source

signal conditioning


PA input and output


and adjustment)

Apply DPD Model, andGet DPD+PA Response


Create DPD Stimulus

Simulated

BB Input

Meas RF

Output

11

55

44

33

22


Comparing Methods: BB Input vs. Measured RF

ACLR -2BW

Lower

-1BW

Lower

+1BW

Upper

+2BW

Upper

Raw PA output

54.06 35.33 35.68 53.58

DPD+PA w/ BB input

55.05 50.15 52.28 54.59

DPD+PA w/ PA input

55.80 51.23 54.32 55.41

LTE-Advanced DL (20 MHz) 6-Carrier GSM

ACLR of DL 20 MHz System

Measured RF input

BB input DPD+PA

RESULTS

23


SystemVue DPD Modeling Flow for LTE/LTE-A

The download power and length of the waveform can also be set.

24

Step 1. Create DPD stimulus waveform• Set LTE parameters such as BW, Resource Block allocation and others• Choose between built-in LTE or LTE-Advanced waveform generation


THRU : Connect the MXG/AWG directly to the PXA/M9392A and click the “Capture Waveform” button. This is the true RF PA input.

DUT: Connect the MXG to the PA, connect the PA to the PXA/M9392A, and click the “Capture Waveform” button. The captured signal is the output of the PA DUT.

The measured I/Q files are stored and used in following steps.

25

Step 2. Capture PA response• SystemVue downloads directly to the MXG or M9330A AWG (source), and

capture data back from PXA or M9392A (analyzer).• Equipment parameters such as number of signal, trace assignment, and file

name can be set.



DPD AM-to-AM Characteristic

26


Step 3. DPD Model Extraction• DPD model parameters such as number

of training samples, memory order, and nonlinear order can be set.

PA AM-to-AM Characteristic


27


Step 4. Capture DPD+PA Response• The signal is pre-distorted by the DPD model

and re-downloaded into the MXG or AWG.

DPD+PA (measured RF output)

PA input (original RF input)

Set the RF powerDPD+PA AM-to-AM Characteristic



Step 5. Verify DPD+PA response• LTE performance for the DPD model used with the PA hardware is verified.

Spectrum, EVM andACLR are calculatedand plotted automatically

28


29

Accommodating Proprietary IP

• Use your own extractor IP instead of Agilent’s • Continue to enjoy an integrated environment• Allows remote & distributed DPD teamwork • Greater user control of algorithm details, IP

security, performance, delivery date, quality, etc

Custom DPD Model Extraction(.m math language)

Custom Digital Pre-distorter(.m math language)


DPD of LTE-Advanced DL with Doherty PA (200W)Spectrum, ACLR and EVM results (10 MHz DL System)

ACLR -2BW

Lower

-1BW

Lower

+1BW

Upper

+2BW

Upper

BB input (sim)

58.67 49.63 49.17 58.01

Raw PA output

49.90 28.69 28.35 47.31

DPD+PA

output

48.88 45.10 45.16 48.83

ACLR (dB)

EVM

EVM (%) EVM (dB)

Simulation BB input 5.33 -24.46

Raw PA output 10.13 -19.89

DPD+PA output 5.52 -25.16

CFR was applied to this LTE-Advanced DL signal, with a

maximum EVM target of 10% for 16-QAM.

Raw PA output

PA+DPD, after 1 iteration to extract DPD coefficients

30

RF Vector Source：N5182B MXGRF Vector Analyzer: N9030A PXA


DPD of LTE-Advanced DL with Doherty PA (200W)Spectrum, ACLR and EVM results (20MHz DL System)

ACLR -2BW

Lower

-1BW

Lower

+1BW

Upper

+2BW

Upper

BB input (sim)

64.73 55.09 57.10 64.92

Raw PA output

51.01 30.69 30.04 49.50

DPD+PA

output

50.31 45.16 45.56 51.40

ACLR (dB)

EVM

EVM (%) EVM (dB)

BB input signal (sim) 6.10 -24.28



CFR was applied to this LTE-Advanced DL signal with a maximum EVM

target of 10%,8% and 6% for QPSK, 16-QAM and 64-QAM, respectively.

Raw PA output


31



LTE-A Results with 200W LDMOS Doherty PARaw PA Output (DL 20MHz System)

32


LTE-A Results with 200W LDMOS Doherty PADPD+PA Output (DL 20MHz System)

33


ACLR (dB)

CC0 EVM (QPSK)

CC1 EVM (16-QAM)

LTE-A Results with 200W LDMOS Doherty PAResults with (2x10MHz) Carrier Aggregation of 2 separate CC’s

ACLR -2BW

Lower

-1BW

Lower

+1BW

Upper

+2BW

Upper

BB input (sim)

63.11 56.75 56.70 62.72

Raw PA output

50.58 30.80 30.22 49.06

DPD+PA

output

51.74 45.75 45.73 51.18

EVM (%) EVM (dB)

Baseband signal (sim) 0.21 -53.43



EVM (%) EVM (dB)

Baseband signal (sim) 0.20 -54.11



Raw PA output


34



Multi-Standard Radio (MSR) into Doherty PA (200W)

2 CarriersGSM

2 Carriers WCDMA

2 Carriers LTE

2 Carriers EDGE

Raw PA output

PA+DPD

35


Example 2: Efficiency vs. Distortion (20W Doherty)2x50W devices, Fc=2.14 GHz, LTE=20MHz, PAPR=7.3dB

Vgs1: +3.05V Vgs2=1.8V Idqs: 434mA Vds: +28V

Transistor: MRF6S21050 FreeScale 50W x 2


36

Example 2: PA output, no DPD (Raw output) Pout=+42dBm; ACPR = 31dBc; PAE=38%

MeasurementAttenuation

= 62.5dB


37

Example 2: PA output, with DPD(Full power) Pout=+42dBm; ACPR = 31dBc; PAE=39%


38

Example 2 Summary: No DPD vs. DPD

Doherty PA 50W x2, Freq 2.14GHz

LTE 20MHz,

PAPR 7.3dB with CFR

PA Output

Without DPD

PA Output

With DPD

Improve-

ment

Output Power +42dBm (15.8W) +42dBm(15.8W) 0

Gain 13.5 dB 13.5 dB 0

ACPR Lower ACPR: -30.0 dBc

Upper ACPR: -31.3 dBc

Lower ACPR: -45.2dBc


15 dB

EVM QPSK: 10.6%

16QAM: 10.8%

64QAM: 9.9%

QPSK: 6.6%

16QAM: 6.8%

64QAM: 6.7%

3.8%

Efficiency 38% 39% +1%

Memory Effect 1.4 dB 0.04 dB 1.3 dB


39

Example 2: PA output, no DPD(Backed off 10dB) Pout=+33dBm; ACPR > 45dBc; PAE=12%


40

Example 2 Summary: 10dB Back-off vs. DPD

Doherty PA 50W x2, Freq 2.14GHz

LTE 20MHz,

PAPR 7.3dB with CFR

PA Output

Backed off 10dB to meet

ACLR=45dB spec

PA Output

With DPD

Specificati

on

Improved

Output Power +33dBm (2W) +42dBm (15.8W) +9dB

Gain 16 dB 13.5 dB -2.5dB

ACPR Lower ACPR: -45.6 dBc


Lower ACPR: -45.2 dBc


0 dB

EVM QPSK: 6.7%

16QAM: 6.9%

64QAM: 6.7%

QPSK: 6.6%

16QAM: 6.8%

64QAM: 6.7%

0%

Efficiency 12% 39% +27%

Memory Effect 0.3 dB 0.04 dB 0.3 dB


41

Example 3: LTE-Advanced, 2CC 40MHzContiguous Carrier Aggregation, with CFR

19-22dB

RF Source: N5182B MXG or M9381A VSGRF Vector Analyzer：N9030A PXA


42

Example 3: LTE-A 2CC 40MHzPA Input (EVM=6.1%, ACPR=60.1dB, PAPR=6.90dB)


43

Example 3: LTE-A 2CC 40MHzPA Output (EVM=8.3%, ACPR=34.0dB, PAPR=5.82dB)


44

Example 3: LTE-A 2CC 40MHzDPD+PA Output (Iter #1, EVM=6.1%, ACPR=52.8dB, PAPR=6.90dB)


45

Example 3: LTE-A 2CC 40MHzDPD+PA Output (Iter #2, EVM=6.1%, ACPR=55.8dB, PAPR=6.84dB)


46

ACPR (dB) EVM (%)

Lower Upper CC0 CC1

PA Input -60.46 -60.13 6.13% 6.10%

PA Output -33.57 -34.01 8.26% 8.13%

DPD+PA

Output Iter 1 -52.60 -52.85 6.09% 6.04%

DPD+PA

Output Iter 2 -55.27 -55.80 6.15% 6.07%

Example 3 Summary: LTE-A 2CC 40MHz

~21.8dB final improvement


47

Example 4: 802.11ac 5GHz WLAN 80MHz with CFR (PAPR=7dB)

11-14dB

RF Source：N8241A AWG + E8257D PSGVector Analyzer：M9392A PXI VSA


48

ACPR (dB) EVM

Lower Upper EVM % EVM dB

PA input -46.45 -46.18 6.71% -23.47 dB

PA output -34.91 -34.79 8.68% -21.23 dB

DPD+PA

output

-46.29 -45.39 6.97% -23.13 dB

Example 4 Summary: 802.11ac 5GHz WLAN 80MHz with CFR (PAPR=7dB)

~11dB final improvement


49

802.11ac 80MHz with CFR (PAPR=7dB) PA Input


50

802.11ac 80MHz with CFR (PAPR=7dB) PA Output


51

802.11ac 80MHz with CFR (PAPR=7dB) DPD+PA Output


52

53

Agenda






6. Summary


54

DPD with Agilent EEsof EDA simulatorsPredictive PA modeling and linearization

Benefits of using RF Simulation for DPD

• Predict the final DPD result, while Analog PA can still be changed • De-risk module or wafer iteration, to save time and money• Explore vendors, waveforms, statistical spreads, analog variables• Validate system-level specifications with preliminary RF & BB

Trade offs:

• Accuracy. Dynamic “circuit envelope” behavior depends on– the simulation engine (and any behavioral modeling)– the device-level transistor models, for traps, self-heating, mismatch

• Speed. – Real HW measurements >> faster than Simulations

Conclusion: it is still worth doing


Simulation vs. Measurement DPD Extraction

55

• ADS & GoldenGate Circuits as simulated RF DUTs- Complex loading, memory FX, dynamic behaviors

• NVNA X-parameter measurement model,- Great for smaller solid-state devices

SIMULATION-BASED DPD

(predictive)

X-parametersX-parameters

CO-SIM, MODELS

CO-SIM, MODELS

MODEL

ADS

GG

RF DUTN5241,2 PNA-X

External Trigger

Attenuator

N5182 MXG, or E8257D PSG

as external modulatorM9330A AWG if > 100 MHz

89600

VSA

89600

VSA

M9392A PXI VSA (>140MHz)or N9030A PXA (<140 MHz)

I,Q RF

RF DUT

MEASUREMENT-BASED DPD



BB PHY

CFR DPDUp

convertPA

EnvTracking

DAC

Downconvert

ADCAdapt

Duplexer

• Which blocks are included with your final product?• What IP do you have access to? Or, are able to imitate? Able to modify?• What final system specifications do you need to test against?

56

?

??

?

??


Agilent Simulation-based DPD Modeling Platform

BB TX PHY

CFRDPD

model

Upconvert

PADAC

Downconvert

ADCGenerate

Coefficients

W1716

DPD

Step-by-Step GUI

W1918

LTE-A

IP Library

BB RXPHY

Throughput

BER/FER

ACPR

EVM

89600 VSA

Optional Reference RX

W1461 SystemVue

Also:

3G, WLAN60GHz,

DVB, OFDM

Agilent ADS

Agilent GoldenGate

RF circuit-level EDA software

57


Simulation-based, predictive DPDSystemVue co-simulation with circuit-level PA in ADS

EnvOutSelector

O1

OutFreq=FCarrier

GainRF

G4

Gain=1

SVCosimSinkTimed

S3

SVReceiverID="ADSToSystemVue"BlockSize="BlockSize"

in put

SVCosimSourceTimed

S4

SVSenderID="SystemVueToADS"

TimeStep="Tstep"

Fc="FCarrier"BlockSize="BlockSize"

o utp ut

DF

DF

Vdc

FETLS12_MWTD_FETMGA_545P8_FET2

datafile=myfile

Fingers=14Width=1020 um

FETLS12_MWTD_FETMGA_545P8_FET1

datafile=myfile

Fingers=1Width=32 um

VARVAR3

myfile=strcat(mypath,"/adsptolemy/lib/data/bdaa.txt")mypath=echo("$HPEESOF_DIR")

EqnVar

I_ProbeI_in1

I_Probe

Ig

I_ProbeIout

I_Probe

I_d

MSUB

MSub1

TanD=0.03T=2 mil

Hu=3.9e+034 milCond=1.0E+50

Mur=1Er=4.6

H=10.0 mil

MSub

VARVAR2

Cout=10Rout=50

Rbias=655N=2.75

Fin=5.5Pwrin=13

Eqn

Var

MLOCTL3

W=16.0 milSubst="MSub1"

MLOC

TL2

W=16.0 mil

Subst="MSub1"

Port

P1Num=1

Port

P2Num=2

MLIN

TL6

L=100.0 milW=16 mil

Subst="MSub1"

MLIN

TL5

L=100.0 milW=16 mil

Subst="MSub1"

MLINTL4

L=100.0 mil

W=16 milSubst="MSub1"

MLIN

TL1

L=100.0 milW=16 mil

Subst="MSub1"

CC31

C=3.3 pF

C

C25C=2.2 pF

LL35

R=L=.13 nH

MTEETee2

W3=16 milW2=16 mil

W1=16 milSubst="MSub1"

MTEETee1

W3=16 milW2=16 mil

W1=16 milSubst="MSub1"

RR2

R=42 OhmCC32

C=1.5 pF

LL14

R=L=3.9 nH

RR1R=655 Ohm

ADS circuit-level PA(circuit envelope simulation)

ADS Ptolemy(circuit-system co-simulation)

SystemVueSTIMULUS

SystemVueRESPONSE

CO-SIM CO-SIM

ADS reads data

from SystemVueADS sends data

to SystemVue

ADS circuit-level

PA，needs Circuit Envelop

to co-simulate

with SystemVue.

58



Re

Im

R1

Ga in=0.978 [PowerAl ignm ent]

G1

Pe riodic =YES

Fi le= ' Step3_DPD_Coe ffi c ients _Rea l_ Iter2.tx t [DPD_Co effi c ien ts_Real_Fi leName]

Fi le= 'Step1_BBDa ta_ Im ag_Iter1 .tx t [Step1_BBData_Im a g_Fi leName]

R8

Fc

CxEnv

E2

T

Sam p leRa te=69 .33e+6Hz [Sam p lingRate]

S4

Segm en tTim e=50µs

Sta rt=0s

Re

Im

C1

DPD_Pr eDist ort erDPD_I nput

DPD_Coef

DPD_O ut put

Num OfInpu tSam ples=20000 [Num OfInpu tSam ples]

Nonl inea rOrde r=9 [Non linea rOrde r]

M em oryOrder=5 [M em ory Orde r]

D1

Fc

EnvCx

Fc =2 .505GHz [FCa rrie r]

C2

ADS Cos im

OutputID=' ADSToSys tem Vue

InputID=' Sy s tem VueToADS

OutputBlock Siz e=1000 [Bloc kSize]

InputBloc kSiz e=1000 [Bloc kSize]

Ou tpu tFc=2.505e+9 [FCarrier]

A1

Da taFileNam e=' Step4_DPD_PAOutputdata_ Im ag_ Ite r2 .tx t [Step4_DPD_PAOu tpu t_ Im ag_Fi leNam e ]

Sam pleStop=99998 [StopSam p le- 1]

Sam p leSta rt=0 [Sta rtSam ple ]

StartStopOption=Sam ples

123

ExtractCapture PA input vs. output waveforms for DPD extraction

VerifySee linearized result, including DPD

Re

Im

C7 CxToRect@Data Flow Models

Fc

CxEnv

E1 EnvToCx@Data Flow Models

Fc

EnvCx

Fc=2.505e+9Hz [FCarrier]C1 CxToEnv@Data Flow Models

T

SampleRate=69.33e+6Hz [SamplingRate]S1 SetSampleRate@Data Flow Models

Re

Im

R3 RectToCx@Data Flow Models

R1 ReadFile@Data Flow Models

Periodic=YES

File='Step1_BBData_Imag_Iter2.txt [Step1_BBData_Imag_FileName ]


123

DataFileName='Step2_PAOutputdata_Real_Iter2.txt [filename_PAout_I]SampleStop=99999 [NumOfCapturedSamples-1]

SampleStart=0

StartStopOption=Samples

PAInputData_Real2 Sink@Data Flow Models

123

DataFileName='Step2_PAOutputdata_Imag_Iter2.txt [filename_PAout_Q]

SampleStop=99999 [NumOfCapturedSamples-1]SampleStart=0


PAInputData_Imag2 Sink@Data Flow Models

ADS Cos im

OutputID='ADSToSystemVue

InputID='SystemVueToADSOutputBlockSize=1000 [B lockSize]

InputBlockSize=1000 [BlockSize]

OutputFc=2.505e+9 [FCarrier]A1 ADSCosimBlockEnv@Data Flow Models

SystemVue

DPDSystemVue

PA

ADS

PA

ADS

The UI to connect with ADS in SystemVue, corresponding to the schematic (Ptolemy co-sim with circuit-level design) in ADS.

59


60


40dB improvement after 2 iterations

6-Carrier GSMCarrier Spacing: 4MHz

Sampling Rate：256 * 270.8333kHz =69.3333 MHz

PA input Spectrum (Green)PA output Spectrum(Blue)PA+DPD Spectrum (Red, first iteration)PA+DPD Spectrum (Orange, Second iteration)


Envelope Tracking (ET): Using ADS “Circuit

Envelope” to improve true modulated PAE

61

For more information about this application see blog article:http://www.rf-design-tips.com/envelope-tracking-simulation/


Re

Im

R6

Re

Im

R1

Gain=0 .821 [PowerAl ignm ent]

G1

Period ic =YES

Fi le='Step3_DPD_Coe ffic ients _Rea l_Iter2.tx t [DPD_Co effic ien ts _Rea l_Fi leName]

R4

Period ic =YES

Fi le= 'Step3_DPD_Coeffic ien ts _Im ag_Iter2.tx t [DPD_Co effic ien ts _Im ag_Fi leName]

R5

Fi l e= 'Step1_BBDa ta_ Im ag_Iter1.tx t [Step1_BBDa ta_ Im ag_Fi leName]

R8

Fi l e= 'Step1_BBDa ta_Real _Iter1 .tx t [Step1_BBData_Real_Fi leName]

R7

Fc

CxEnv

E2

Fc

EnvCx

Fc =2.505GHz [FCa rrie r]

C2

Fast Cir cuit Envelope

Fi l e= 'SIM_PA_2p505GHz _m 7dBm _lev e l3_ampaccu...

F2

T

Sam pl eRate=34.67e+6Hz [Samp l ingRate]

S4

Spect rum Anal yzer

Segm en tTim e=50µs

Start=0s

M ode=Tim eGate

AfterDPD

Re

Im

C1

Fc

EnvCx

Fc =2.505GHz [FCa rrie r]

C3

Spect r um Anal yzer

SegmentTim e=50µs

Sta rt=0s

M ode=Tim eGa te

Be fo reDPD

T

Sam pleRa te=34.67e+6Hz [Sam p l ingRate]

S1

DPD_Pr eDist or t erDPD_I nput

DPD_Coef

DPD_O ut put

NumOfInpu tSam pl es =30000 [Num OfInputSamp les]

Nonl i nearOrder=9 [Nonl inearOrder]

M em ory Order=5 [M em ory Orde r]

D1

123

DataFi leName=' Step4_DPD_PAOutpu tda ta_Im ag_ Ite r2.tx t [Step4_DPD_PAOu tput_ Im ag_Fi leName]

Sam pl eStop=99998 [StopSam pl e- 1]

Sam p leSta rt=0 [Sta rtSam p le]

StartStopOp tion=Samp les

DPD_PAOu tputData_ Im ag

123

DataFi leName=' Step4_DPD_PAOutpu tda ta_Rea l_Iter2.tx t [Step4_DPD_PAOu tpu t_Real_Fi leName]

Sam pl eStop=99998 [StopSam pl e- 1]

Sam p leSta rt=0 [Sta rtSam p le]

StartStopOp tion=Samp les

DPD_PAOutputDa ta_Real

Re

Im


Fc

CxEnv


Fc

EnvCx

Fc=2.505e+9Hz [FCarrier]

C1 CxToEnv@Data Flow Models

T

SampleRate=34.67e+6Hz [SamplingRate]

S1 SetSampleRate@Data Flow Models

Re

Im


Periodic=YES

File='Step1_BBData_Real_Iter2.txt [Step1_BBData_Real_FileName]


Periodic=YES

File='Step1_BBData_Imag_Iter2.txt [Step1_BBData_Imag_FileName ]


123

DataFileName='Step2_PAOutputdata_Real_Iter2.txt [filename_PAout_I]

SampleStop=99999 [NumOfCapturedSamples-1]

SampleStart=0



123



SampleStart=0



Fast Cir cuit Envelope

File='SIM_PA_2p505GHz_m7dBm_level3_ampaccu...

F2 FastCircuitEnvelope@Data Flow Models

Re

Im

C2 CxToRect@Data Flow Models 123

DataFileName='Step2_PAInputdata_Real_Iter2.txt [filename_PAin_I]


SampleStart=0



123

DataFileName='Step2_PAInputdata_Imag_Iter2.txt [filename_PAin_Q]


SampleStart=0



Spect rum Anal yzer

SegmentTime=50µs

Start=0s

Mode=TimeGate

PAOut_spec

Spect r um Anal yzer

SegmentTime=50µs

Start=0s

Mode=TimeGate

PAIn_spec

PA

Simulation-based, predictive DPDSystemVue with native FCE model, extracted from GoldenGate

CMOS Handset PAFast Circuit Envelope (FCE) model extracted from GoldenGate (direct co-sim is also possible, but slower)

DPD PA

ExtractCapture PA input vs. output waveforms for DPD extraction


62


Simulation-based, predictive DPDSystemVue with native FCE model, extracted from GoldenGate

30dB improvementafter 2 iterations

6-Carrier GSMCarrier Spacing: 600kHz

Sampling Rate：128 * 270.8333kHz=34.6667 MHz

PA input Spectrum (Green)PA output Spectrum(Blue)PA+DPD Spectrum (Red, first iteration)PA+DPD Spectrum (Orange, Second iteration)

63


Simulation-based, predictive DPDSystemVue with analog X-parameter model (100W PA)

Analog X-parameter device is placed into a Spectrasys subnetwork (RF simulation domain)

Source1=200 MHz at -10 dBm

MultiSource_3 MultiSourceZO=50Ω

Port_2 *OUT

*GND

VDC=22V

SG1 VDC

X1

2

3VDC

File='.\100W_3Hz_3H_3Bias_PHD.mdf

XP_1 XPARAMS

FILE

64


Re

Im


Fc

CxEnv


Fc

EnvCx

Fc=200e+6Hz [FCarrier]

C1 CxToEnv@Data Flow Models

T

SampleRate=34.67e+6Hz [SamplingRate]

S1 SetSampleRate@Data Flow Models

Re

Im


Periodic=YES

File='Step1_BBData_Real_Iter1.txt [Step1_BBData_Real_FileName]R1 ReadFile@Data Flow Models

Periodic=YESFile='Step1_BBData_Imag_Iter1.txt [Step1_BBData_Imag_FileName ]


123

DataFileName='Step2_PAOutputdata_Real_Iter1.txt [filename_PAout_I]

SampleStop=99999 [NumOfCapturedSamples-1]SampleStart=0



123



SampleStart=0StartStopOption=Samples


RF_Link

CalcPhaseNoise=NO

EnableNoise=NOFreqSweepSetup=Automatic

Schematic='Xparam_device

Data1 RF_Link@Data Flow Models

Spect rum Anal yzer

SegmentTime=50µs

Start=0sMode=TimeGate

PAOut_spec

Spect rum Anal yzer

SegmentTime=50µs

Start=0sMode=TimeGate

PAIn_spec

Re

Im


123

DataFileName='Step2_PAInputdata_Real_Iter1.txt [filename_PAin_I]


SampleStart=0StartStopOption=Samples


123

DataFileName='Step2_PAInputdata_Imag_Iter1.txt [filename_PAin_Q]SampleStop=99999 [NumOfCapturedSamples-1]

SampleStart=0

StartStopOption=SamplesPAInputData_Imag1 Sink@Data Flow Models


Re

Im

R6

Ga in=1.14 [PowerAl ignm ent]

G1

Periodic =YES

Fi le='Step3_DPD_Coe ffic ients _Rea l_ Ite r1.tx t [DPD_Coe ffic ients _Rea l_Fi leName]

R4

Periodic =YES

Fi le= 'Step3_DPD_Coeffic ien ts _Im ag_ Ite r1.tx t [DPD_Coe ffic ients _Imag_Fi leName]

R5

Fc

CxEnv

E2

Fc

EnvCx

Fc =0 .2GHz [FCa rri er]

C2

T

Sam pleRa te=34.67e+6Hz [Sam p l ingRate]

S4

Spect rum Anal yzer

Segm en tTi me=50µs

Sta rt=0s

M ode=Ti meGa te

Afte rDPD

Re

Im

C1

Fc

EnvCx

Fc =0 .2GHz [FCa rri er]

C3

Spect rum Anal yzer

Segm entTim e=50µs

Start=0s

Mode=Tim eGate

Be foreDPD

T

Sam pleRa te=34.67e+6Hz [Sam pl ingRate]

S1

DPD_Pr eDist or t erD P D _ I n put

D P D _ C o ef

D P D _ O u tput

Num OfInpu tSam ples =20000 [Num OfInputSam p les]

Non l inea rOrde r=11 [Nonl inearOrder]

M em ory Orde r=3 [Mem ory Orde r]

D1

RF_Link

Cal c Phas eNo is e=NO

EnableNo is e=NO

FreqSweepSetup=Autom atic

Sc hem atic = 'Xpa ram _device

Data1

Re

Im

R1

Fi le='Step1_BBData_Rea l_ Ite r1.tx t [Step1_BBDa ta_Rea l_Fi leName]

R7

Fi le='Step1_BBData_Imag_ Ite r1 .tx t [Step1_BBData_Imag_Fi leName]

R8

123

DataFi l eNam e='Step4_DPD_PAOu tputdata_ Im ag_Iter1 .tx t [Step4_DPD_PAOutpu t_Imag_Fi leName]

Samp leStop=99998 [StopSamp le- 1]

Sam pleStart=0 [StartSam ple]

Sta rtStopOpti on=Sam pl es

DPD_PAOutpu tDa ta_Im ag

123

DataFi l eNam e='Step4_DPD_PAOu tputdata_Real_ Ite r1 .tx t [Step4_DPD_PAOutput_Rea l_Fi leName]

Samp leStop=99998 [StopSamp le- 1]

Sam pleStart=0 [StartSam ple]

Sta rtStopOpti on=Sam pl es

DPD_PAOu tpu tData_Rea l

X-param

PAExtractCapture PA input vs. output waveforms for DPD extraction


RF_LinkBrings RF networks (incl. X-parameter devices) up to the dataflow simulation

DPD

X-param

PA

65



~40dB improvement(w/o memory effects)

PA input Spectrum (Green)PA output Spectrum(Blue)PA+DPD Spectrum (Red

6-Carrier GSMCarrier Spacing: 600kHz

Sampling Rate：128 * 270.8333kHz=34.6667 MHz

FILE

66


Page 67

DPD Modeling Simplification: Automation UI

MXA / PXA

MXG DUT

FastCircuitEnvelope

ADS Cosim

Both DPD extractions share the same UI:

• Measurement-based • Simulation-based

Measurement-based

GG co-sim (or FCE model)

ADS co-sim


Verification of simulation-based DPD Sweep power, re-extract DPD at each point, watch EVM, ACP

Input waveform: • IEEE 802.11ac, 5 GHz WLAN • No CFR (PAPR is 8.7dB)• Bandwidth = 80MHz system• 4x Oversampling rate=320 MHz

Device Under Test:• WLAN “FCE” model extracted from

Agilent GoldenGate RFIC simulator

EVM with DPD

EVM w/o DPD

ACLRwith DPD

Lower/UpperACLR

w/o DPD

Output < 0 dBm

DPD offerslittle benefit

0 < Output < +16.5 dBm

DPD offerssignificant benefit

EVM vs. Output Power ACP vs. Output Power

68


Verification of simulation-based DPD Sweep power, constant DPD coefficients, watch EVM, ACP

Useful Rangefor this set

of DPD coefficients

PA may be lesscorrectable

Different(or fewer)

DPD coefficients

needed

EVM with DPD

EVM w/o DPD

Lower/ UpperACLR

with DPD

Lower/UpperACLR

w/o DPD

ACP satisfies aspectral

compliancemask

High DC-RFefficiencybut poor

ACP

ACP may actually be worse

out of range:turn DPD off.

Question: “Do I need Adaptive DPD?”

69


Verification of simulation-based DPD Sweep power, re-extract at each point, see final Pout vs. Pin

Signal with PAPR = 8.7dBmust be backed-off, lower average power

Power OutputWith DPD

Signal with PAPR = 7.5dBcan be driven to higher average power

Linear Gain = 25.5dB

Using Crest Factor Reduction (CFR) to reduce the peaks, the average signal level can be increased farther to the right, resulting in higher DC-RF Efficiency, and longer distance coverage

70


Memory Polynomial vs. Volterra DPD models802.11ac 80MHz, FCE PA Model Co-sim

71

ACPR

Lower Upper

EVM

(dB)

Original input

-56.19 -57.20 -47.16

PA Output (No DPD)

-36.66 -38.43 -29.88

DPD+PAIter1

-50.28 -49.95 -42.20

DPD+PA Iter2

-53.39 -52.18 -44.41

ACPR

Lower Upper

EVM

(dB)

Original input -56.19 -57.20 -47.16

PA Output (No DPD)

-36.68 -38.45 -29.90

DPD+PA Iter1 -51.60 -49.79 -42.90

DPD+PA Iter2 -54.05 -54.29 -46.06

DPD+PA Iter3 -54.71 -55.26 -46.40

Memory Polynomial (21 coefficients) Volterra Series (24 coefficients)


Verification after DPD model extraction Verifying Memory Order and Nonlinear Order in Memory Polynomial

EVM and ACP

are stable

when memory

order>=3.

Memory effect

almost

removed

when memory

order >=3.

EVM and ACP

are stable

when

nonlinear

order>=7.

EVM vs. Memory Order

(@Nonlinear Order=7)

EVM vs. Nonlinear Order

(@Memory Order=3)

ACPR vs. Memory Order

(@Nonlinear Order=7)

ACPR vs. Nonlinear Order

(@Memory Order=3)

72


Verification after DPD model extraction A closer look at ACPR vs. Nonlinear Order (“how many terms do I need?”)

-39dB

-56dB

Order=3 Order=11

Nonlinear=9

(@Memory=3)

73


Verification after DPD model extraction A closer look at ACPR vs. Memory Order (“how many terms do I need?”)

-43dB

-54dB

Memoryless Order=5

Memory=3

(@Nonlinear=7)

74


Look-Up Table (LUT) DPD Algorithm Lower power, suitable for high volume (femtocell, handset PAs)

(3) Predistorters/Extractors now available in SystemVue DPD module:• Look-up Tables for Gain, Phase (does not mitigate memory FX)• Memory Polynomial • Volterra


75

76

Agenda






6. Summary


• Spectrally efficient wideband RF signals may have PAPR >13dB.• CFR preconditions the signal Reduce signal peaks without

significant signal distortion• CFR allows the PA to operate more efficiently – it is not a linearization

technique• CFR supplements DPD and improves DPD effectiveness• Without CFR and DPD, a basestation or handset PA must operate at

significant back-off from saturated power to maintain linearity. The back-off reduces efficiency

Benefits of CFR

1. PAs can operate closer to saturation, for improved efficiency (PAE).2. Output signal still complies with spectral mask and EVM

specifications

Crest Factor Reduction (CFR) Concepts

77


78

Crest Factor Reduction (CFR) Concepts

Instantaneous PAPR

(Peak-to-Avg Power Ratio)

Psat

INPUT

POWER

OUTPUT

POWER

Pavg

Psat

INPUT

POWER

OUTPUT

POWER

Pavg

WITHOUT CFRPAPR ~13dB

Raw LTE-Advanced

WITH CFRPAPR ~7dB

Run at +6dB higher avg power

Benefit: Effectively largerRFPA with same HW BOM


CFR for LTE-Advanced Downlink OFDMA

Controls EVM and band limits in the frequency domain.• Constrains constellation errors, to avoid bit errors.• Constrains the degradation on individual sub-carriers.

Allows QPSK sub-carriers to be degraded more than 64 QAM sub-carriers.Does not degrade reference signals, P-SS and S-SS.Subcarriers of out-of band are set to NULL.

79


CFR for LTE-Advanced Downlink OFDMA

• No side modifications for receiver• No out-of band spectral distortion (no spectral mask measurement pass/fail issue• EVM always meets specification•Good PAR reductions•No impact of timing and frequency and channel estimation of DL

80

B u s =NO

D ata Ty pe=C o mp lex

Ma pp ing D a ta D A TA P ORT

B u s =NO

D ata Ty pe =In teg er

Q m D A TA P ORT

FFT

Freq S e que nc e=0 -po s -n eg

D ire c ti on=Inv erse

S i z e=81 92 [D FTS ize]

FFTS iz e=8 19 2 [D FTS i ze]

i fft2

0+0*j

V alu e=0 [0 +0 *j]

z e ros

G a in=1

G 3

G ai n=1

G 2

A

B loc k S iz e s =1;6 00;699 1;6 00 [[1 ,H a l f_ U s edC arriers ,D FT_z e ros ,H al f_ U s ed Carriers]]

A3

A

B loc k S iz e s =600 ;60 0 [ [H a l f_U s e dC arriers , H al f_U s ed C a rriers]]

A2

FFT

Fre qS eq uen c e =0-pos -neg

D irec tio n=Forw ard

S iz e=8 19 2 [D FTS i ze]

FFTS iz e =81 92 [D FTS ize]

fft

FFT

Freq S e que nc e=0 -po s -n eg

D ire c ti on=Inv erse

S i z e=81 92 [D FTS ize]

FFTS iz e=8 19 2 [D FTS i ze]

i fft1

MOD256

Mo du lo=256

M 1

0+0*j

V alu e=0 [0 +0*j]

DC

DPD_Ra diusClip

input out put

C lip pin gTh res ho ld=16.5e-6 [C l ip pin gThre s hold]

D P D _R ad iusClip

B u s =NO

D ata Ty pe =In teg er

S C _ S tatu s D A TAPORT

DPD

LTE_A

CFR_PostProc

out put

Qm

SC_St at us

r ef

input

E V M _Thre s h old _64 Q A M =0.06 [E V M _Th res ho ld_64QAM]

E V M_ Thres h old _1 6Q A M =0 .1 [E V M_ Th res hol d_16QAM]

E V M _Th res hol d_Q P S K =0 .12 [E V M _Thre s hold_QPSK]

O v e rs a mp l ing O p tio n=R atio 4 [O v e rs a mp l ing O p tion]

B a ndw id th=B W 2 0 M H z [B an dwidth]

D 2 D P D _L TE _A _C FR _P os tP ro c @ DPD Models

G ain =8 192 [D FTS ize]

G 1

B u s =NO

D ata Ty pe=C o mp lex

ou tpu t D A TA P ORT


CFR of LTE-Advanced 20MHz Downlink

QPSK modulation, CFR algorithm set to Max EVM = 10%

81

PAPR=9dB w/o CFRPAPR=6.8dB w CFRPAPR=9dB w/o CFRPAPR=6.8dB w CFR

Spectrums with and w/o CFR

are same!

Spectrums with and w/o CFR

are same!


CFR of LTE-Advanced 20MHz DownlinkAlgorithm EVM targets: QPSK < 10%, 16QAM < 8%, 64QAM < 6%

82

PAPR=8.9dB w/o CFRPAPR=7.2dB with CFRPAPR=8.9dB w/o CFRPAPR=7.2dB with CFR

ObservedEVMs w/CFR


CFR of LTE-Advanced with Carrier Aggregation

83

• CFR performed separately on each Component Carrier (up to 20MHz BW)

• Component Carriers are then aggregated (summed)

CFR Approach 1 CFR Approach 2

• CC’s are carrier-aggregated (up to 100MHz BW), then CFR’d together

• Then each component carrier is re-filtered individually to remove out-of-band energy,and re-summed


CFR of LTE-Advanced with Carrier AggregationApproach 1, 2x20MHz contiguous CA

84

1. Both CC0 and CC1 adopt 16-QAM and QPSK, respectively.2. CC1 magnitude threshold of polar clipping is a little larger than CC0 because

QPSK modulation can tolerate larger EVM limit, according to EVM specification.

1 1 0 1 0

DataPattern=PN9B2

1 1 0 1 0

DataPattern=PN9B3

1 1 0 1 0

DataPattern=PN9B1

1 1 0 1 0

DataPattern=PN9B4

T

SampleRate=122.9e+6Hz [SamplingRate]S1

T

SampleRate=122.9e+6Hz [SamplingRate]S2

Gain=1GainUnit=voltage

G1

Gain=1GainUnit=voltage

G2

Fc

EnvCx

Fc=2.14e+9Hz [FCarrier1]

Fc

EnvCx

Fc=2.16e+9Hz [FCarrier2]C2

UE1_Dat a

HARQ _Bit s

f r m_TD

f r m_FD

UE1_M odSymbols

U E1_ChannelBit s

LTE_A

DL

Src

CFR

NumFrames=1ClippingThreshold=13.05e-6 [ClippingThreshold2]

CFREnable=YESCRS_NumAntPorts=CRS_Tx1

NumTxAnts=Tx1UEs_RevMode=0;0;0;0;0;0 [[0,0,0,0,0,0]]

Cyc licPrefix=NormalOversamplingOption=Ratio 4 [OversamplingOption]

Bandwidth=BW 20 MHz [Bandwidth]FrameMode=FDD

ShowSystemParameters=YESLTE_DL_Src_CFR1

UE1_Dat a

HARQ _Bit s

f r m _TD

f r m _FD

UE1_ModSym bols

U E1_ChannelBit s

LTE_A

DL

Src

CFR

NumFrames=1ClippingThreshold=11.75e-6 [ClippingThreshold1]

CFREnable=YESUEs_RevMode=0;0;0;0;0;0 [[0,0,0,0,0,0]]

CyclicPrefix=NormalOversamplingOption=Ratio 4 [OversamplingOption]

Bandwidth=BW 20 MHz [Bandwidth]FrameMode=FDD

ShowSystemParameters=YESLTE_DL_Src_CFR2

FcChange

Env

CCDF

Stop=50msStart=0s

CA_CCDF

Spect rum Anal yzer

SegmentTime=50µsStart=0s

Mode=TimeGateCA_Spectrum

Component Carrier 0 (CC0)


CCDF

CC1_CCDF

CCDF

CC0_CCDF1

Parameter CFREnable=YES


CFR of LTE-Advanced with Carrier AggregationApproach 1: 2x20MHz contiguous CA

85

EVM of PDSCH 16-QAM is 8.54% in CC0 and EVM of PDSCH QPSK is 11.11% in CC1.EVM values of P-SS, S-SS and RS <0.65%

CC0 PAPR =7.2 dB

CC1 PAPR = 6.7dB

2x20MHz 2CC with CFR #1

PAPR = 8.2dB



86

1. Both CC0 and CC1 adopt 16-QAM and QPSK, respectively.2. Aggregate CC0 and CC1 first, then do polar clipping on the

40MHz bandwidth composite CA signal.3. Each Component Carrier is filtered separately (20MHz each)4. Combine the filtered CC0 and CC1 into one CA signal again.

Parameter

CFREnable=NO

Fc

ChangeEnv

T

Sam pleRate=12 2.9 e+6Hz [SamplingRate]

S1

T

Sam pleRate=12 2.9 e+6Hz [SamplingRate]S2

Fc

EnvCx

Fc =2.16e+9Hz [FCarrier2]C2

Fc

EnvCx

Fc =2.14e+9 Hz [FCarrier1]

M ax im um Order=2057StopRipple=80

Stop Band wid th=19e6HzPas s Ripp le=0.1

Pas s Ba ndwidth=18e6Hz

FCenter=2.16e+9Hz [FCarrier2]F4

Fc

Change

Ban dwidth=0Hz

Outp utFc =2.16e+9Hz [FCarrier2]E4

Fc

Change

Bandwidth=0Hz

OutputFc =2.15e +9Hz [FCarrier]E5

Env

OutputFc=MaxA3

Fc

Change

Ban dwidth=0HzOutp utFc =2.14e+9Hz [FCarrier1]

E3

Fc

EnvCx

Fc =2.1 5e+9Hz [FCarrier]C4

T

Sam ple Rate=122 .9e +6Hz [SamplingRate]S4

Fc

CxEnv

E2

DPD_RadiusClip

in p ut o u t put

Cl ippingTh res hold=0.14250DPD_Ra diu s Clip_1

CCDF

Sto p=50ms

Start=0sCl ipp ing _CCDF

S p e c t r u m A n alyzer

Se gm entTime=50µs

Start=0sM ode=TimeGate

CA_ Spec trum _Clipping

M ax im um Order=2057StopRipple=80

Stop Band wid th=19e6HzPas s Ripp le=0.1

Pas s Ba ndwidth=18e6HzFCenter=2.14e+9Hz [FCarrier1]

F1

CCDF

Stop=50msStart=0s

CA_ CCDF

S p e c t r u m A n alyzer

Segm entTime=50µsStart=0s

M od e=TimeGateCA_ Spectrum

1 1 0 1 0

DataPatte rn=PN9

B2

1 1 0 1 0

DataPatte rn=PN9

B3

1 1 0 1 0

Data Pattern=PN9B1

1 1 0 1 0

Data Pattern=PN9B4

U E 1 _ D ata

H A R Q _ Bits

f r m_TD

f r m_FD

U E 1 _ M o d S ym bols

U E 1 _ C h a n n e lBits

LTE_A

DL

Src

CFR

Cl ip pin gThres ho ld=11.15e-6 [ClippingThreshold1]CFREnable=NO

Cy c li c Prefix=NormalOv ers am pl in gOptio n=Ratio 4 [Ov ersamplingOption]

Band wid th=BW 20 M Hz [Bandwidth]Fram eM ode=FDD

Sh owSy s te mPara meters=YESL TE_DL _Src _CFR2

U E 1 _ D ata

H A R Q _ Bits

f r m _TD

f r m _FD

U E 1 _ M o d S ym bols

U E 1 _ C h a n n e lBits

LTE_A

DL

Src

CFR

Cl ippingThres h old =12.5e-6 [Cl ippingThreshold2]CFREna ble=NO

Cy c l ic Pre fi x=NormalOv e rs a mp lingOp tion=Ratio 4 [Ov ersamplingOption]

Bandwidth=BW 2 0 M Hz [Bandwidth]Fra me Mo de=FDD

ShowSy s tem Param eters=YESLTE_DL_ Src _CFR1



Polar clipping

Filtering per each carrier

Combine carriers as CA signal



87

EVM of PDSCH 16-QAM is 7.80% in CC0 and EVM of PDSCH QPSK is 7.82% in CC1.All EVM values of P-SS, S-SS and RS are about 7%

2x20MHz 2CC w/o CFR

PAPR = 9 dB

2x20MHz 2CC with CFR #2

PAPR = 7.4dB


Sequential Asymmetric Superposition (SAS)New CFR algorithm block in SystemVue 2013.01

Innovative elements• Asymmetric window • Adaptive window length• New scaling function • Applies to common wireless time-

domain waveforms

Results• Easy implementation• Lower spectral regrowth • Better PAPR performance• Target PAPR value is the threshold_dB value

SAS CFR

dataIn dataOut

lengthFrame=1000max_N=31

max_iter=10Beta=12.0

threshold_dB=6.0D1 DPD_SAS_CFR@DPD Models


88

SAS Crest Factor Reduction exampleLTE-A 2CC Carrier Aggregation

CFR for 3GPP LTE-A DL Carrier Aggregation Waveform



T

Sam pleRate=1 22.9e +6Hz [Sam pl ingRate]S1

Gain=1

GainUni t=v oltageG1

Env

Fc

EnvCx

Fc =2.14e+9 Hz [FCarrier1]

CCDF

Stop=100ms

Start=0s

CC0_CCDF

1 1 0 1 0

DataPattern =PN9

B4

1 1 0 1 0

DataPattern =PN9

B1

1 1 0 1 0

DataPattern=PN9B2

1 1 0 1 0

DataPattern=PN9

B3

T

Sam pleRate=1 22.9e +6Hz [Sam pl ingRate]

S2

Gain=1

GainUni t=v ol tage

G2

Fc

EnvCx

Fc =2.1 6e+9Hz [FCarrier2]

C2

CCDF

Stop=1 00ms

Start=0sCC1_CCDF

U E 1 _ D ata

H A R Q _ B its

f r m _TD

f r m _FD

U E 1 _ M o d S y m bols

U E 1 _ C h a n n e lB its

LTE_A

DL

Src

CFR

Av oidPhas e Dis c o ntinuity=NO

ClippingThres hold=13.15e-6

CFREn able=NO

Num Fram es=1

UEs _n_ RNTI=(1x 6) [1,2,3 ,4,5,6]

UEs _Spe c ific RS=(1x 6) [0,0,0,0,0,0]UEs _Tran s M od e=(1x 6) [0,0,0,0,0,0]

Cy c l i c Prefi x =Normal

Ov ers am plingOption=Ratio 4

Ban dwidth=BW 20 M Hz [Ban dwidth]

Fram eM ode=FDDShowSy s tem Param eters=YES

LTE_DL_Src _CFR2

U E 1 _ D ata

H A R Q _ B its

f r m _TD

f r m _FD

U E 1 _ M o d S y m bols

U E 1 _ C h a n n e lB its

LTE_A

DL

Src

CFR

Av oidPhas e Dis c o ntinuity=NOClippingThres hold=13.15e-6

CFREn able=NO

Num Fram es=1

UEs _n_ RNTI=(1x 6) [1,2,3 ,4,5,6]

UEs _Spe c ific RS=(1x 6) [0,0,0,0,0,0]

UEs _Tran s M od e=(1x 6) [0,0,0,0,0,0]Cy c l i c Prefi x =Normal

Ov ers am plingOption=Ratio 4

Ban dwidth=BW 20 M Hz [Ban dwidth]

Fram eM ode=FDD

ShowSy s tem Param eters=YES

LTE_DL_Src _CFR1

Fc

CxEnv

E2

DPD_ SAS_CFR

length Fram e=3000

m ax _N=51

m ax _i ter=10Beta=12.0

thres hold_dB=6.0

D1

Fc

EnvCx

Fc =2 .15e+9Hz [FCarrier]

C4

CCDF

Stop=50ms

Start=0s

CFR_CA_CCDF

Spect rum Anal yzer

Seg m entTim e=50µs

Start=0s

M ode=Tim eGate

CA_ Spec trum _CFR

SAS CFR block


89

SAS Crest Factor Reduction exampleLTE-A 2CC Carrier Aggregation


90

SAS Crest Factor Reduction exampleMulti-Standard Radio (MSR)

GSM

WCDMA

LTE

GSM

CFR algorithm suitable for Multi-format, Multi-band, and Instrument waveforms


91

92

Agenda






6. Summary


Problem statement

Modern communication systems try to meet conflicting requirements:• Signals with high PAPR, that then require inefficient back-off • RF PAs with high PAE, that then cause time & freq distortions

Solution approaches

• Digital Pre-Distortion (DPD) and Crest Factor Reduction (CFR) algorithms together help overcome conflicting requirements.

• SystemVue offers a practical DPD modeling flow • Connects to/from open, enterprise modeling & EDA tools• Control your own IP, or leverage Agilent’s IP to model any HW or

Algorithms you don’t have access to• Re-use commonly available test equipment• Create virtual systems using simulators, test equip, scripting, UI

Summary

93



BB PHY

CFR DPDUp

convertPA

EnvTracking

DAC

Downconvert

ADCAdapt

Duplexer

• Model any blocks not included with your final product, and get on with your project• Imitate/Model key missing pieces of IP and hardware, and maintain control• Verify against realistic system specifications, which may be controlled externally

94


95

Questions & Answers


Agilent LTE/LTE-Advanced textbookDiscusses many of the concepts shown today

96


NEW FOR 2013SystemVue applications featured in this book• LTE-Advanced (3GPP Releases 8-12)• MIMO channel/Over-the-Air testing• Multi-Standard Radio (MSR)• DPD, Crest Factor Reduction• Integration of Design & Test

http://www.agilent.com/find/LTEbook

ISBN 978-1-1199-6257-1 (Wiley & Sons)

“Novel Wideband DPD Modeling Platform for 4G”

97

THANK YOU

W1716 Digital Pre-DistortionWeb - www.agilent.com/find/eesof-systemvue-dpd-builderApp Note - http://cp.literature.agilent.com/litweb/pdf/5990-6534EN.pdfApp Note - http://cp.literature.agilent.com/litweb/pdf/5990-7818EN.pdfApp Note - http://cp.literature.agilent.com/litweb/pdf/5990-8883EN.pdf

SystemVuewww.agilent.com/find/eesof-systemvue www.agilent.com/find/eesof-systemvue-videoswww.agilent.com/find/eesof-systemvue-evaluation

Or, contact your regional Agilent resourcewww.agilent.com/find/eesof-contact


Documents

ICC2013 Co-located Workshop