Large-Signal Network Analysis Tools and Techniques Page 1 Large-Signal Network Analysis Technology to help the R&D Customer.

Large-Signal Network Analysis Tools and Techniques Page 1

Large-Signal Network Analysis

Technologyto help the R&D Customer


Agenda

Introduction Large-Signal Network Analysis The Large-Signal Network Analyzer Calibration The core of the LSNA Technology Examples A typical LSNA measurement session Next steps in LSNA Technology Wrap-up


Design Challenge

“Customers are demanding more capabilities/performance from their devices.”

Designers are looking for better methods of characterizing their components

Demands translate to greater design complexities

More complex modulation schemes

Higher power efficiency requirements

Improved linearity

90

0 PhaseSplitter

I/Q Modulator

LO

I

Q

Rx/TxModule

MCPA

PA Module

Matched Transistors

Mixer

Transistors

ProcessEngineer

PADesigner

ModelingDesigner

ICDesigner

SystemDesigner


Why can’t I predict device behavior

To be successful in this environment, it is essential to fully characterize and understand device behavior

Need more realistic test conditions Devices that operate in large-signal environments

can’t be characterized with linear tools

Existing tools are insufficient Network analyzers only characterize small-signals (linear) behavior accurately Signal analyzers evaluate properties of signals interacting with the test device,

they do not analyze the interactions of analyzer with the test device


AM-PM

Amplifier Measurements

Device UnderTest

Loadpull

ACPR

GainPower in and out

Power Added Efficiency

Phase flatness


Build two MCPAs, one passes the other does not Do you know what to fix?

ACPR and other measurement data only represent symptoms of the problem No insight is provided as to the cause of the problem

ACPR of an MCPA

FAIL

PASS


Existing Measurements and Limitations

Spectral re-growth, IMD, ACPR

Characterizes signals caused by nonlinear behavior of components - in the frequency domain

EVM

Compares deviation of modulated signal from ideal - in the time domain

Limitations

Characterizes signals resulting from interaction DUT - measurement system, device performance is not isolated

Results will change when environment changes Different sources and analyzers can produce different results Characterizing just the DUT requires perfectly matched conditions

DUT Freq. (GHz)DUTZ1 Z2


Existing Measurements and Limitations con’t

AM-AM and AM-PM

Characterizes changes in output power and phase with changes in input power Starts defining the transfer function of the nonlinear behavior

Limitations

DUT performance is still not isolated from the rest of the system

Results will change with changes in the environment

Results also depend on type of test signal regardless of matched conditions

DUT Freq. (GHz)DUTZ1 Z2



Load Pull

Traditional: Characterizes applied impedances and powers at fundamental frequency Measures incident, reflected and transmitted power as a function of S and L

Harmonic: Characterizes applied impedances and powers at fundamental and harmonics

Provides more complete information than traditional load pull. Harmonic termination has large impact on performance

Limitations

Information is still missing, the DUT is not completely characterized

Does not allow to apply PA design theory (waveform engineering)

Measurements do not uniquely define a particular test state

May identify multiple local minimums as opposed to a optimal (global) minimum

DUTSourceTuner(S )

LoadTuner(s)

(L )

x x x x

VNA, SAor Pwr Mtr.

VNA, SAor Pwr Mtr.



Modulated S-parameters

Attempt to use known concepts in new situations

Hot S22

Characterizes the interaction of the DUT with the load under large - signal drive Depends on the chosen configuration

Limitations

Modulated S-parameters do not have a scientific basis Superposition principles do not apply for nonlinear behavior Results will vary with the test conditions when device is nonlinear

Hot S22 is still missing critical information for complete nonlinear characterization

The missing data mayor may not impact measurement results


Insufficient Modeling Tools

Ideal:

Measurements correlate with simulationsIn a linear environment, S-Parameters are an excellent example

The real world for non-linear characterization:

Insufficient models Incomplete informationPoor correlation between measurements and simulations

Model Simulate Build Meas S-PACPR

Power=

Model Simulate Build Meas


Results

Cut-and-try engineering (designers “imagineer” fixes) Design verification consumes 2/3rds of development time Time-to-market delays Unpredictable design processes Time consuming tuning and measurement requirements


How can Agilent help?

Large - Signal Network Analysis is a breakthrough new technology that provides unprecedented insight into transistor, component and system behavior using the same concepts across this complete spectrum

Through a small dedicated team Agilent is ready to work closely with early-adopter customers in different markets to create successes in their R&D environment through this technology


Agenda



Large - Signal Network Analyzer (LSNA) Technology Goals

complete characterization of a device, component and system under large - signal periodic stimulus at its ports. LSNA technology is presently limited to devices that maintain periodicity in their response

deriving nonlinear component characteristics which are invariant for the used equiment and test signals

Foundation: Large-signal Network Analysis


Small-Signal Network Analysis

Small-Signal Linear Behavior Test signal : simple, typically a sine wave Superposition principle to analyze behavior in realistic conditions

Network Transistor, RFIC, Basestation Amplifier, Communication system

Analysis Complete component characterization : S - parameters

(within measurement bandwidth)



Large-Signal Refers to potential nonlinear behavior Nonlinear behavior -> Superposition is not valid Requirement: Put a DUT in realistic large-signal operating conditions

Network Transistor, RFIC, Basestation Amplifier, Communication system

Analysis Characterize completely and accurately the DUT behavior for a given type of stimulus Analyze the network behavior using these measurements


Large-Signal Network Analysis: Overview

TransistorRFICSystem

RealisticStimulus

RealisticStimulus

Measurement System

1v

1i2v

2i1a

1b

2a

2b

0)|,,,,(

0)|,,,,(

2121

2121

ftiivvg

ftiivvf Physical Quantity Sets

Travelling Waves (A, B)

Voltage/Current (V, I)

Representation Domain Frequency (f)

Time (t)

Freq - time (envelope)

Analysis


Practical Limitations of LSNA for Large-Signal Network Analysis

Large-Signal Network analysis will be performed using periodic stimuli

one - tone and harmonics periodic modulation and harmonics

The devices under test maintain periodicity in their response


Continuos Wave Signal

Freq. (GHz)

1 2 3 4DC

DUT

Z1

Z2

Freq. (GHz)

1 2 3 4DC

Freq. (GHz)

1 2 3 4DC

Freq. (GHz)

1 2 3 4DC

Freq. (GHz)

1 2 3 4DC

All voltages and currents or waves are represented by a fundamental and harmonics (including DC)

Complex Fourier coefficients Xh of waveforms

X0

X1

X2

X3

X4


Amplitude and Phase Modulation of Continuos Wave Signal

Freq. (GHz)

1 2 3 4DC

DUTZ1 Z2

Freq. (GHz)

1 2 3 4DC

Freq. (GHz)

1 2 3 4DC

Freq. (GHz)

1 2 3 4DC

Freq. (GHz)

1 2 3 4DC

Complex Fourier coefficients Xh(t) of waveforms

Phasor

Amplitude

Phase

Fast change (GHz)

Slow change (MHz)

Modulation

X0(t)

X1(t)

X2(t)

X3(t)

X4(t)

time

time time

time time


Periodic Modulated Signals

Freq. (GHz)

1 2 3 4DC

DUT

Z1

Z2

Freq. (GHz)

1 2 3DC

Complex Fourier coefficients Xhm of waveforms

Freq. (GHz)

1 2 3DC

Freq. (GHz)

1 2 3DC

Freq. (GHz)

1 2 3DC

Phasor

Amplitude

Phase

PeriodicModulation

X0i

X1i

X2i X3i


Waves (A, B) versus Current/Voltage (V, I)

50cZTypically

2

IZVA c

2

IZVB c

BAV

cZ

BAI

“From device to system level”


Small-Signal Network Analysis: S-parameters

2221212

2121111

aSaSb

aSaSb


Measurement System

1a

1b

Analysis

sexperimentdifferent ngrepresenti with

,,, 2211

i

baba iiii

2a

2b

50

Experiment 1


Measurement System

1a

1b2a

2b

50

Experiment 2




RealisticStimulus

RealisticStimulus

Measurement System

1v

1i2v

2i1a

1b

2a

2b

0)|,,,,(

0)|,,,,(

2121

2121

ftiivvg

ftiivvf Analysis

sexperimentdifferent ngrepresenti with

,,, 2211

i

baba iiii

Different Experiments


Agenda



Vector Network Analyzer Measurement

50 Ohm

Acquisition

Calibration

Linear Theory

S-parameters

Stimulus

Response

ReferencePlanes

H

MEAS FORMATSCALE

REF

DISPLAY AVG CAL

MKRFCTN

MKR

CH 1 CH 2

MENU

START STOP

CENTER SPAN

SYSTEM LOCALUSER

PRESET

COPYSAVE

RECALL SEQ

7 8 9

4 5 6

1 2 3

0 . -

@

n

M

k

m

x1ENTRY

OFF

ACTIVE CHANNEL

RESPONSE

STIMULUS

ENTRY

INSTRUMENT STATE R CHANNEL

OUTR L T S

HP-IB STATUS

IN

PROBE POWER FUSED

PORT 1 PORT 2

TRANS FWDREFL FWD

TRANS REVREFL REV

+26 dBm RF 30 VDC MAX PORTS 1&2 AVOID STATIC DISHCARGE

8753DNETWORK ANALYZER

30 KHz-3GHz


Complete SpectrumWaveforms

Harmonics and Periodic Modulation

50 Ohmor

tuner

Acquisition

Calibration

Stimulus

Response

ReferencePlanes

ModulationSource

Large-Signal Network Analyzer


Filter

Filter

Filter

FilterDUT

TestSet

Data

-Acq

uisit

ion

LO

Source

2nd Source

PC

Sampling Converter

Cal Kit

Power Std

Phase Std

LSNA System Block Diagram

Calibration Standards

Converts carrier, harmonics

and modulationto IF bandwidth

Separates incident andreflected waves intofour meas. channels

Or Tuner

On waferConnectorized

•RF bandwidth: 600 Mhz - 20 GHz•max RF power: 10 Watt•Modulation bandwidth•Needs periodic modulation

E1430 - based4 MHz IF

10 MHz IF


Harmonic Sampling - Signal Class: CW

Freq. (GHz)1 2 3

50 fLO 100 fLO 150 fLO

Freq. (MHz)1 2 3

RF

IF

fLO=19.98 MHz = (1GHz-1MHz)/50

LP IF Bandwidth: 4 MHz

Cutt Off IF


Harmonic Sampling - Signal Class: Periodic Modulation LP

IF Bandwidth: 4 MHz

1 2 3

50 fLO 100 fLO 150 fLO

Freq. (MHz)1 2 3

RF

IF

fLO=19.98 MHz = (1GHz-1MHz)/50


Harmonic Sampling - Signal Class: Periodic Broadband Modulation

Freq. (GHz)1 2 3

150 fLO

Freq. (MHz)

RF

IF

LP

BW

BW

Adapted sampling process

BW of Periodic Broadband Modulation = 2* BW IF data acquisition

BW8 MHz


Agenda



50 Ohmor

tuner

Acquisition

Calibration

Stimulus

Response

ReferencePlanes

ModulationSource

LSNA Calibration

Measured waves

Actual waves at DUT

7 relative error termssame as a VNAAbsolute magnitude

and phase error term

ma1mb1

ma2mb2

DUTa1DUTa2

DUTb1

DUTb2

m

m

m

m

DUT

DUT

DUT

DUT

b

a

b

a

K

b

a

b

a

2

2

1

1

22

22

11

1

2

2

1

1

00

00

00

001

F0=1GHz

freq

1GH

z

2GH

z

3GH

z


Relative Calibration: Load-Open-Short

50 Ohm

Acquisition

50 Ohm

LoadOpenShort

50 Ohm

Acquisition

50 Ohm Thru

{f0, 2 f0, …, n f0}

Calibration for fundamental and Harmonics

22

22

11

1

00

00

00

001

K

{f0, 2 f0, …, n f0}

F0=1GHz


Power Calibration

50 Ohm

Acquisition

Power Meter

{f0, 2 f0, …, n f0}

22

22

11

1

00

00

00

001

K

Amplitude

{f0, 2 f0, …, n f0}

F0=1GHz

freq

2GH

z

3GH

z

1GH

z


Phase Calibration

50 Ohm

Acquisition

Reference Impulse Generator

f0

...f0

50 Ohm

22

22

11

1

00

00

00

001

K

Phase

{f0, 2 f0, …, n f0}freq

2GH

z

3GH

z

1GH

z

F0=1GHz


Measurement Traceability

Relative Cal Phase Cal Power Cal

National Standards(NIST)

Agilent Nose-to-Nose Standard


Agenda



The heart of the Large-Signal Network Analysis

This hardware is the core that will be used to work with the customer in providing LSNA technology

Combines capabilities of a vector network analyzer, sampling scope and ESG-VSA.

Provides complete waveform analysis capabilities CW/Multi-tones with harmonics 0.6 to 20 GHz frequency coverage 8MHz usable IF BW 10 W power handling capability


Agenda



Examples

Transistor reliability Transistor model verification (ICCAP / ADS) Transistor model tuning PA design using waveform engineering System level characterization Scattering functions Memory effect Dynamic bias


Gate - Drain Breakdown Current

Time (ns)

º transistor provided by David Root, Agilent Technologies - MWTC

º TELEMIC / KUL


Forward Gate Conductance

Time (ns)

º transistor provided by David Root, Agilent Technologies - MWTC

º TELEMIC / KUL


Examples



Use of LSNA measurements in ICCAP model verification, optimisation (and extraction)

ICCAP specific input

ADS netlist. Used, a.o., to impose themeasured impedance to the output ofthe transistor in simulation

sweep of Power Vgs Vds Freq


Transistor De-embedding

0 0.5 1 1.5 2

- 3

- 2

- 1

0

1

2 beforeafter

de-embedding

Time/period

Gat

e cu

rrent

/ m

A

Equivalent circuit of the RF test-structure, including the DUT and layout parasitics


Input capacitance behaviourVgs,dc=0.9 VVds,dc=0.3 V Vds,dc=1.8 V

Input loci turn clockwise, conform i=C*dv/dt


Dynamic loadline & transfer characteristicVgs,dc=0.3 VVds,dc=0.9 V


LSNA identifies modeling problem : extrapolation example SiGe HBT

100 200 300 400 500 600 700 8000 900

-0.002

-0.001

0.000

0.001

-0.003

0.002

time, psec

i1st

si1

mts

_de

100 200 300 400 500 600 700 8000 900

0.6

0.7

0.8

0.9

1.0

1.1

0.5

1.2

time, psec

v1

sts

v1

mts

_de

100 200 300 400 500 600 700 8000 900

1.3

1.4

1.5

1.6

1.2

1.7

time, psec

v2

sts

v2

mts

_de

100 200 300 400 500 600 700 8000 900

0.000

0.002

0.004

0.006

-0.002

0.008

time, psec

i2st

si2

mts

_de

SiGe HBT (model parameters extracted using DC measurements up to 1V) Vbe= 0.9 V; Vce=1.5 V; Pin= - 6 dBm; f0= 2.4 GHz

simul.meas.


LSNA identifies modeling problem : extrapolation example SiGe HBT

Measurement Simulation

SiGe HBT - DC characteristics

0.2 0.4 0.6 0.8 1.0 1.2 1.40.0 1.6

-0.010

-0.005

0.000

0.005

0.010

0.015

0.020

-0.015

0.025

VbDC

DC

meas1

..Ic

e

0.2 0.4 0.6 0.8 1.0 1.2 1.40.0 1.6

-0.010

-0.005

0.000

0.005

0.010

0.015

0.020

-0.015

0.025

VbDC

i2.i

Alcatel Microelectronics and the Alcatel SELStuttgart Research Center teams are acknowledged

for providing these data.


Examples



MODEL TO BE OPTIMIZED

generators apply LSNA measured waveforms

“Chalmers Model”

“Power swept measurements under mismatched conditions”

GaAs pseudomorphic HEMTgate l=0.2 um w=100 um

Parameter Boundaries

Empirical Model Tuning

º Dominique Schreurs, IMEC & KUL-TELEMIC


During OPTIMIZATION

Time domain waveforms Frequency domain

gate drain

voltage

current

gate drain

Voltage - Current State Space

Using the Built-in Optimizer


Verification of the Optimized Model

Time domain waveforms Frequency domain

gate drain

voltage

current

gate drain

Voltage - Current State SpaceAFTER OPTIMIZATION


Examples



Waveform Engineering Block Diagram

DUTTestSet Da

ta-A

cqui

sitio

n

Source

PC

Sampling Converter

Filter

Filter

Filter

Filter

LO

f0

f0

2f0

3f0IRCOM Setup


Example - Measured Waveforms

MesFET Class Ff0=1.8 GHzIds0=7 mAVds0= 6 V

Z(f0)=130+j73 Z(2f0)=1-j2.8 Z(3f0)=20-j97

PAE=84%

PAE50%

WaveformEngineering

º IRCOM / Limoges


Example - Performance ImprovementDerived Information from the V/I waveforms (swept input power at different terminations)

Z(f0)=123+j72 Z(2f0)=50 Z(3f0)=50

Z(f0)=123+j72 Z(2f0)=2 - j 4.0 Z(3f0)=50

Z(f0)=123+j72 Z(2f0)=2 - j 4.0 Z(3f0)=21-96

PAE74%

PAE74%

PAE84%º IRCOM / Limoges


Examples



RFIC Amplifier Characterization using periodic modulation

ModulationSource

a1

E1

a1

E1

A1 shows spectral regrowth

• Spectral regrowth on b1 combined with measurement system mismatch• Nonlinear pulling on source

5 dB

f0 = 1.9 GHz Evaluation Board


Transmission Characteristics

A1

Carrier Modulation

Carrier Modulation

B2

Carrier Modulation

3rd harmonicModulation

Harmonic DistortionCompression


Reflection Characteristics

A1

Carrier Modulation

Carrier Modulation

B1

3rd harmonicModulation

Harmonic Distortion

Carrier Modulation

2nd harmonicModulation

Expansion


Examples



Scattering Functions provide device understanding and enable CAE couplingTuners and active injection at harmonics

@ fundamental frequency

@ higher harmonics


Nonlinear behaviour and Scattering Functions

Nj

iijphijijphijphphphph aHaGaHaGaFb

,...,22,1

**2121212111

Functions of 11a

Index of: Port & harmonicNote: a’s and b’s are phase normalized quantities !!

As shown before: for small-signal levels (linear) this reduces to (fundamental at port 2)nnnnnn aSaSb 2122112121

(and independent bias settings)


Scattering functionsvariation versus input power

- 20 - 15 - 10 - 5 5

- 50

- 40

- 30

- 20

- 10

10

20

- 20 - 15 - 10 - 5 5

- 100

- 50

50

100

21F

2121G

2121H

21F

2121G

2121H


Generated reflection coefficients at port 2 at f0

Generated ’s

’s for verification meas.

21(a)


Time domain waveformsmeasured and simulated b-waves

200 400 600 800 1000

- 0.2

- 0.1

0.1

0.2

200 400 600 800 1000

- 6

- 4

- 2

2

4

6 tb1

tb2


Application of CDMA-like signal

- 25 - 20 - 15 - 10 - 5 5

5

10

15

20

25

Output power versus input power; CW HredL, CDMA HlinesL


20 40 60 80 100 120 140

- 60

- 40

- 20

21bFrequency domain

fc=2.45 GHz, f 50 kHz, modulation BW 1.45 MHz

red=measured, blue=model


Examples



Time domain

,,,,, '22

'112 tatatataftb m

250 300 350 400 450 500

2

3

4

5

6

tb m2Memory effects !


Memory effectsDUT behaviour under 2-Tone excitation

-8 -6 -4 -2 0 2 4

14

16

18

20

22

24

26

-8 -6 -4 -2 0 2 4

14

16

18

20

22

24

26

Modulation frequency = 20 kHz Modulation frequency = 620 kHz


Examples



What is “Dynamic Bias Behaviour”?

Freq. (GHz)

1 2DCFreq. (GHz)

1DC

Input Voltage Output Current

1V 2I

Dynamic Bias Behaviour

Frequency Domain: Generation of Low Frequency Intermodulation Products

Time Domain:“Beating” of the Bias


Dynamic Bias: Measurement Principle

TUNER

RF Data Acquisition

Dynamic Bias Data Acquisition

CurrentProbe

Bias 1Supply

CurrentProbe

Bias 2Supply

Computer


0 0.2 0.4 0.6 0.8 1 1.2

0

20

40

60

0 0.2 0.4 0.6 0.8 1 1.2-2

-1.5

-1

-0.5

0

0.5

RFIC Example in Time Domain

00.20.40.60.811.2-2-1.5-1-0.500.5

Output Current Waveform (without Dynamic Bias)

(mA)

(V)

Normalized Time

Normalized Time

Input Voltage Waveform

“MultiLine TRL”


0 0.2 0.4 0.6 0.8 1 1.2

25

30

35

40

45

Adding Measured Dynamic Bias

00.20.40.60.811.2-2-1.5-1-0.500.5

Output Current Waveform (including Dynamic Bias)

(mA)

(mA)

Normalized Time

Normalized Time

Dynamic Bias Current Waveform


Agenda



LSNA possible next steps driven by customer needs Extending modulation BW (3G) Increase power capability Extending frequency range (50 GHz and beyond …) Offer pulsed measurements to isolate the thermal effects Complete dynamic bias testing capabilities to characterize the effects

of modulation on bias Add impedance tuning measurements to determine the impact of

differing impedance conditions Use of LSNA technology in high speed digital applications


Example: Extending Power Capability

Acquisition

Calibration

Stimulus

ReferencePlanes

ModulationSource

Pre-matchingProper calibration elementsOn - board DC biasTuners

?

3rd party

Adapt test - setProper absolute calibrationMeasurement science

Agilent NMDG


Agenda



Wrap-up Large-Signal Network Analysis Technology is breakthrough technology to

characterize nonlinear behavior from transistor to system The technlogy is targeted toward research and design experts. It requires a

strong background in RF or Microwave theory to be successful. Agilent NMDG is assigned to make the technology a success with early-

adopter key customers More information at : “http://wirelesscentral.tm.agilent.com/wirelesscentral/cgi-

bin/epsg.cgi” If you think the LSNA technology can help you, please contact

[email protected]

Large-Signal Network Analysis Tools and Techniques Page 1 Large-Signal Network Analysis Technology to help the R&D Customer.

Documents

type of test signal

largesignal environments

device performance

device behaviorto

acprcharacterizes signals

properties of signals

large impact

particular test statemay