The Ohio State University, Columbus, OH University of ... · the RF PA for shaping table design, including bias and load effects. A Envelope Tracker Driver RF PA Envelope ... X-parameters

X‐parameter applications for characterizing and modeling power amplifiers for envelope

tracking applicationsHaedong Jang*, Andrew Zai†, Tibault Reveyrand‡,

Patrick Roblin*, Zoya Popovic†, and David E. Root§

*The Ohio State University, Columbus, OH†University of Colorado, Boulder, CO

‡XLIM-UMR CNRS, Limoges Cedex, France §Agilent Technologies, Santa Rosa, CA

more details in session TU3F-1 Tues 13:50-14:10 ref [21]

WSO: Transceiver and Technologies for Femto/Pico Cell Comm. Systems IMS2013, Seattle, June 2-7, 2013 2

Outline

• Introduction to ET • Introduction to X-parameters• X-parameter models for dynamic signals:

quasi-static approximation• Simulation-based X-parameter models

for ET applications• Measurement-based GaN PA X-parameter model • Future work• Conclusions


Envelope tracking

Fixed Bias PA

Wasted power

Modulated Bias

DCDC

inin

DC

in

DC

inout

IVIVG

PPG

PPP 11PAE

PDC

Pin Pout

Pdiss

PA


Envelope tracking

Shaping table depends on the properties of the RF PA, viewed as a three-terminal component.

X-parameters provides a procedural approach to characterize & model the RF PA for shaping table design, including bias and load effects.

A

EnvelopeTracker

RF PADriver

EnvelopeShaping

Delayalign

EnvelopeDetection

RFup-convert

I

Q

QI

1 23

I2 + Q2


Envelope tracking

• Design considerations with ET• High PAR input signals• Varying supply voltage• Varying load

• XP model, which is a frequency domain black box behavioral model, is studied

• Design and characterization assumption: • SM is ideal• Interconnection impedance is

minimum• PA is quasi-static to the supply

voltage variation• Load is approximately matched

(but more on this later)PA

SM

1

2

3

Γ


Envelope tracking

Practical approach is designing the shaping using static characterization at multiple supply voltages under quasi-static assumption of the PA

9dB constant gain

5 10 15 20 25 300

2

4

6

8

10

12

2V

6V

9dB

P1dB

Gai

n (d

B)

Pout (dBm)

Maximizing efficiency using saturation point (P1dB)

Drain voltage is swept from 2 V to 6 V


X-parameters

H a r m o n ic B a la n c eH B 2

E q u a t io n N a m e [3 ]= " Z lo a d "E q u a t io n N a m e [2 ]= " R F p o w e r "E q u a t io n N a m e [1 ]= " R F f r e q "U s e K r y lo v = n oO r d e r [1 ]= 5F r e q [1 ]= R F f r e q

H A R M O N I C B A L A N C E

Measure X-parameters-or-

Generate X-parameters from circuit-level designs

X-parameter Component :Simulate using X-

parameters

ADS, SystemVue & Genesys:Design using X-parameters

X-parameters are the scientifically correct extension of S-parameters to large-signal conditions.

– Measurement and simulation based, identifiable from a simple set of automated NVNA measurements or directly from ADS circuit-level designs

– Vector nonlinearities (Magnitude and phase of distortion)– Intrinsic DUT properties (calibrates out source impurities & multi-freq. mismatch)– Cascadable (correct behavior in mismatched environment)– Extremely accurate for high-frequency, distributed nonlinear devices– Includes bias dependence

X2PXNP1File=

21

Ref


X-parameter use model

X

X-Parameters

Reflected Transmitted

Incident

Reflected

Incident

1 2EDA Software

Design

Agilent Nonlinear Vector Network Analyzer

Measure

Model

( )B X A

Same use model as S-parameters

but much more powerful


Complex Spectra and Nonlinear Maps

,p kB,p kA

Port IndexHarmonic Index

2 kA1kA

1kB 2 kB

, 0 , ,

, , ,0

2Z ( )

2Z

p k p k p k

p k p k p k

V A B

I A B

1V 2I

time time

0,( ) Re jk t

p p kk

i t I e 0

,( ) Re jk tp p k

k

v t V e

NVNA

, , 1,1 1,2 1,3 2,1 2,2 2,3, , , ..., , , , ...p k p kB F A A A A A A


Time-invariance and spectral linearization

,,

( ) *1

( )11

,1

( ), 1

,1( ,| |) ), ( ,( )

ef gef gh hef

S f hgh

g h

T f hg

F fe

g hf hB X DC A X DP C A DC A P AP A X

X-parameters allow us to simplify the general B(A) relations:Trade efficiency, practicality, for generality & accuracyPowerful, correct, and practical; Native Freq. Domain Model

1 1 2 2( )) (i i iB S DC A S DC A The simplest X-parameters are just linear S-parameters

new mismatch termsload & harmonic mismatchmatched response

DC dependence automatically included. NVNA controls DC supplies and synchronizes measurements with applied RF signal


X-parameters of GSM PA

“X-parameters provide a nonlinear electronic interactive datasheet”

Mismatch versus phase at GSM_OUT

0 0.2 0.4Real

-1.7

-1.6

-1.5

-1.4

-1.3

-1.2

-1.1

-1

Imag

measured

“Hot S22”

X-params

Skyworks amp

Horn et al., EuMC 2009


Mini-Circuits LAVI-22VH+ (TB-433) Double-balanced Mixer3-ports

New measurement: IF Phase!

X datasheet__ model

X-parameters of mixers: 3 rf ports


Simulation of dynamic signals

B2,k

A1,1

1,1( )2, 2, 1,1, j AF

k kB X DC A e

( )2,

FkX

Now assume the signal is modulated in time: 2,1 2,1( )A A t

1,1 ( )( )2, 2, 1,1( ) , ( ) j A tF

k kB t X DC A t e

Quasi-static approximation: evaluate static X-parameter function at each time instant


Simulating Dynamic signals (2)

t

tA1,1(t)

B2,1(t)B2,1

A1,1

1,1

2,1

( )( )2,1 1,1

( )

, ( ) j A tF

B t

X DC A t e

Envelope Domain:same mechanism generatesspectral regrowth from digitally modulated signals(e.g. estimates of ACPR, EVM)LIMITATIONS (Quasi-Static approximation):Symmetric intermods, independent of modulation rateNo BW dependence; No MemoryValid in the slow modulation rate limit (narrow band)

freq

freq

( )2,1

FX

1,1 ( )( )2, 2, 1,1( ) ( ), ( ) j A tF

k kB t X DC t A t e


Envelope simulationEnvelope simulation

Simulation-based approach

• Instantaneous• AM/AM• AM/PM• PAE

• Instantaneous• AM/AM• AM/PM• PAE

Fix ETCkt

PATR

Fix ETXP

• Input signal• LTE• 16 QAM• 5 MHz

• Input signal• LTE• 16 QAM• 5 MHz

• XP models are extracted in two cases to show the sources of dynamic response contribution

– entire circuit– bare transistor

• Fixed bias performance is compared to ET operation for the reference circuit model and the two X-parameter models

six cases where there is a dynamic signal


Case 1(whole PA)

Case 2(bare transistor)

Simulation-based extraction


DC

Generating the X-parameters (1)

Fixed gate Vdd sweep3-6 V

Fixed load

Input power sweep-10~37 dBm

Entire PA

W2305 X-parameter generator


Generating the X-parameters (2)

Fixed Vgs = 1.75 VInput power sweep-20~27 dBm

Vdd sweep1.5-6 V

Load sweepMag:0.69 – 0.89Phase:154.6°-160.6°

Transistor

Suggested: Sweep load over a large region in the Smith Chart.Sometimes active LP needed to go beyond it


LTE UL16 QAM

Envelope Tracking Implementation

MeasEqn

1

Ref3

2

ENVELOPE

VarEqn

DAC

VarEqn

VarEqn

Drain_Bias

Vload

Vs_high

Vinput Vload

X3P

VCVS

VAR

VAR

SDD2P

DataAccessComponent

R

SDD3P

VAR

VtDataset

V_DC

MeasEqn

C

C

Envelope

Term

I_Probe

I_Probe

I_Probe

IQ_DemodTuned

R

XNP1

SRC4

VAR2

VAR3

SDD2P1

DAC1

R2

SDD3P1

DEMOD1

VAR1

SRC3

SRC1

C2

C1

Env1

Term2

Iload

Is_high

I_input

Precompute_data_to_dataset

R1

File="xparam_sample_PA.xnp"

R2=0 OhmR1=50 OhmG=1

Detected_Pin_dBm=10*log(mag(_v1)/100+1e-10)+30

Delta_Vdrain=Vdrain-3.25Vdrain=file{DAC1, "voltage"}

Cport[1]=C[1]=I[2,0]=-(Delta_Vdrain)/50.0I[1,0]=0

iVal1=Detected_Pin_dBmiVar1="Pin_dBm"ExtrapMode=Constant ExtrapolationFile="Vdrain_vs_Pin_dBm_iso_gain_Andy.mdf"

R=50 Ohm

Cport[1]=C[1]=I[3,0]=0I[2,0]=-(_v1**2+_v3**2)/50I[1,0]=0

Rout=50 OhmFnom=RFfreq

RFfreq=850 MHzGain=4.2

Step=32.5520833 nsecStop=50 usecOrder[1]=9Freq[1]=RFfreq

Expression="Waveform"Dataset="LTE_UL_TxSpectrum.ds"

Vdc=3.25

C=1.0 uF

C=1.0 uF

Freq=RFfreq

R=50 Ohm

Z=50Num=2

XnP Componentfor the sample PA

ADS ET examples can be downloaded from [1]: http://edocs.soco.agilent.com/display/eesofkcads/Applying+envelope+tracking+to+Improve+Efficiency

Ideal envelope tracking

ShapingTable from X-parmodel


Fixed ET

0 10 20 30-5

0

5

AM

-AM

(dB

)

0 10 20 30-40

-20

0

20

40

Pout (dBm)

AM

-PM

( )

0 10 20 30-5

0

5

0 10 20 30-50

0

50

Pout (dBm)

Simulation-based validation

Circuit

XP (Transistor)

XP (PA)


• Instantaneous PAE was calculated using LTE signals and drawn over the histogram of the signals.

• Both XP and circuit model predict PAE improvement over fixed bias as expected.• It is interesting to notice that, under ET, the instantaneous PAE predicted by the

circuit model shows wider spreading than fixed bias.

Simulation-based validation

PAEavg (%)

XP(ET) 43.3

Ckt(ET) 42.4

Ckt(Fix) 31.7


Measurement-based X-parameters

• GaN HEMT 8 W Class-F-1 (Triquint TGF2023-02)• Drain voltage sweep : 12 ~ 30 V • Load-pull performed with VTD SWAP-X402 (now Agilent)


X-par extraction for 8 W GaN PA

PNA-XN5242A

DUT

F = 2.14 GHzGain = 15~21 dBPpeak = 38 dBmA

C

-10 dBm -20 dBm

-20 dBm-10 dBm

+38 dBm

+22 dBm (ET)

4142

0 ~ +25 dBm (DT)9 dBm (ET)

Couplers CF = -16dB

83020AGain >= 30dB Pmax = +30 dBm (1 W)P@1dB = 28 dBmF = 2 – 26.5 GHz

30 dB

20 dB(1W)

0 dB

13 dB(1W)

30 dB

Source 1-20 ~ 5 dBm (DT)

J11

Source 2-11 dBm (ET)-2 dBm (ET)

0 dB

20 dB(1W)

6 dB (5W)DC-18 GHz28 dBm (ET)

25 dBm (DT)

J8

Source Out

6 dB (0.5W)DC-26.5 GHz

10 dB (1W)

R1

R3

VGS = -3.8 VVDS = 12~30V


Validation of measurement-based X-parameter model

Tuner Load Set 1

X-parameters at 50 ohmLoad-pull with tunerNVNA 50 ohm

Vdd increasing

Load-sensitivity of X-par model



Tuner Load Set 2


Vdd increasing



Tuner Load Set 3


Vdd increasing



Tuner Load Set 4


Vdd increasing



Tuner Load Set 5


Vdd increasing


Load-dependence of shaping table

Loads variation Shaping table


ET measurement setup

• The WCDMA test signal was fed to the RF signal generator through arbitrary signal generator

• The shaped signal based on the XP model was fed the supply modulator through another synchronized arbitrary signal generator

• The supply modulator from [4], which has 70 MHz bandwidth and 32 Vmax, was used for the test

• WCDMA test model 1• 3.84 MHz• 10.3 dB PAR• 30.72 MHz sample rate• 1 ms repetition• 28.8 dBm Pout.avg


Results for GaN PA

ET Fixed (26 V) Fixed (28 V) Fixed (30 V)

10 15 20 255

15

25

Gai

n (d

B)

10 15 20 25-30

0

30

Pin (dBm)

AM

-PM

( )


Discussion

• Static XP model based shaping table design shows significant PAE improvement under ET operation

• ET operation showed more gain compression and wider AM-PM spreading than under constant bias operation– lack of linearity improvement might be attributed to increased

memory effects under dynamic biasing different from the ideal assumption in the simulation

– Quasi-static assumption is likely valid when the supply modulator is ideal

– Supply modulator output impedance and inter-connection impedance between the modulator and the PA may not be ideal

– Slew-rate and bandwidth of the tracking amplifier are high and are not likely the problem


Suggestions for future work

• Treat RFPA as three-terminal “incommensurate mixer” X-parameter model

1,[1,0 ] 3,[0 ,1]

1,[1,0 ] 3,[ 0 ,1]

( )2,[ , ] 2,[ , ] 1,[1,0] 3,[0,1]

( ) ' '2,[ , ]; ,[ ', '] 1,[1,0] 3,[0,1] ,[ ', ']

', ',

( )2,[ , ]; ,[ ', '] 1,[1,0] 3,[0,

( ) , ( ) ( ) ( )

( ) , ( ) ( ) ( )

( ) ,

F n mn m n m

S n n m mn m p n m p n m

n m p

Tn m p n m

B X A t A t P t P t

X A t A t A P t P t

X A t A

,[ ', '] 1,[1,0 ] 3,[ 0 ,1]

* ' '1]

', ',( ) ( ) ( )

p n m

n n m m

n m pt A P t P t

See talk WE3D-4 14:50-15:10 ref. [22]


Suggestions for future work (2)

• Application of Dynamic X-parameters to ET applications: beyond quasi-static approximation

• Characterize the modulator and take it into better account in the design


Conclusions

• The envelope simulation and measurement results show good quantitative agreement for the static nonlinearity of the PA versus power and drain voltage, and also as a function of load

• The load-sensitivity of the lookup table predicted by the XP model was independently validated by time-domain loadpull measurement


Acknowledgment

• This work was supported by a Grassroots grant from Agilent Technologies and in part by ONR under the DARPA MPC Program N00014-11-1-0931.

• The authors thank A. Howard, J. Horn, R. Biernacki, M. Marcu, P. Cain, A. Cognata, J. Xu, and A. Cidronali for support and valuable discussions.


References

• [1] A. Howard, “Simulating envelope tracking with Agilent ADS – a “Proof of Concept” example, “ Agilent EesofKnowledge Center, http://edocs.soco.agilent.com/display/eesofkcads/Applying+envelope+tracking+to+Improve+Efficiency

• [2] G. Wimpenny, “Understand and characterize envelope-tracking power amplifiers,” EETimes Design line, http://eetimes.com/design/microwave-rf-design/4233749/Understand-and-characterize-envelope-tracking-power-amplifiers?pageNumber=0

• [3] S. Baker, “Envelope tracking for efficient RF transmitters,” IEEE International Microwave Symposium, Montreal, June 2012

• [4] J. Hoversten, S. Schafer, M. Roberg, M. Norris, D. Maksimovic, Z. Popovic, “Codesign of PA, supply, and signal processing for linear supply-modulated RF transmitters,“ IEEE Trans. MTTS, vol. 60, no. 6, Jun. 2012

• [5] L. Sankey, Z. Popovic, “Adaptive tuning for handheld transmitters,” IEEE International Microwave Symposium Digest, pp. 225-228, June 2009

• [6] M. Roberg, J. Hoversten, and Z. Popovic, “GaN HEMP PA with over 84% power added efficiency,” Electronics letters, 11th, Vol. 46, No. 23, Nov. 2010

• [7] M. Hassan, L. E. Larson, V. W. Leung, D. E. Kimball, and P. M. Asbeck, “A wideband CMOS/GaAs HBT envelope tracking power amplifier for 4G LTE mobile terminal applications,” IEEE trans. MTTS, Vol. 60 No. 5. May 2012

• [8] D. Kim, D. Kang, J. Choi, J. Kim, Y. Cho, and B. Kim, “Optimization for envelope shaped operation of envelope tracking power amplifier,” IEEE trans. MTTS. Vol. 59, No. 7, Jul. 2011

• [9] J. Horn, D. E. Root, and G. Simpson, “GaN device modeling with X-parameters,” IEEE CSICS Oct. 2010• [10] P.M. Asbeck, H. Kobayashi, M. Iwamoto, G. Hanington, S. Nam, L.E. Larson, “Augmented behavioral

characterization for modeling the nonlinear response of power amplifiers,” IEEE MTT-S Intl. Microw. Symp. Dig. 2002


References

• [11] J. Xu, J. Horn, M. Iwamoto, and D. E. Root, “Large-signal FET model with multiple time scale dynamics from nonlinear vector network analyzer data,” IEEE MTTS International Microwave Symposium Digest, May 2010.

• [12] J. Verspecht and D. E. Root, “Poly-harmonic distortion modeling,” IEEE MTTS Microwave Magazine, June 2006• [13] J. Verspecht, J. Horn, L. Betts, D. Gunyan, R. Pollard, C. Gillease, and D. E. Root, “Extension of X-parameters to

include long-term dynamic memory effects,” IEEE MTTS International Microwave Symposium Digest, pp741-744, June 2009

• [14] J. Verspecht, J. Horn and D. E. Root, “A simplified extension of X-parameters to describe memory effects for wideband modulated signals,” IEEE ARFTG 75th, May 2010

• [15] A. Soury and E. Ngoya, “Implementation of X-parameter models in harmonic-balance simulators,” IEEE MTTS International Microwave Symposium, June 2012

• [16] A. Soury and E. Ngoya, “Handling long-term memory effects in X-parameter model,” IEEE MTTS International Microwave Symposium, June 2012

• [17] E. Ngoya, C. Quindroit and J. M. Nebus, “On the continuous-time model for nonlinear-memory modeling of RF power amplifiers,” IEEE trans. MTTS Vol. 57. No. 12, Dec. 2009

• [18] D. K. Su, and W. J. McFarland, “An IC for linearizing RF power amplifiers using envelope elimination and restoration,” IEEE JSSC, Vol. 33, No. 12, Dec. 1998

• [19] http://www.agilent.com/find/x-parameters• [20] http://www.agilent.com/find/NVNA• [21] H. Jang, A. Zai, T. Reveyrand, P. Roblin, Z. Popovic, D. E. Root, “Simulation and Measurement-based X-parameter

Models for Power Amplifiers with Envelope Tracking,” IEEE International Microwave Symposium, June, 2013• [22] G. Casini, A. Cidronali, G. Manes, “Investigation of X-parameters Modeling for Accurate Envelope Tracking Power

Amplifier System Simulations,” IEEE International Microwave Symposium, June, 2013

Documents

The Ohio State University, Columbus, OH University of ... · the RF PA for shaping table design, including bias and load effects. A Envelope Tracker Driver RF PA Envelope ... X-parameters