X-Parameters with Active and Hybrid Active Load Pull

X-Parameters with Active and Hybrid Active Load Pull Gary Simpson, CTO Maury Microwave EuMW 2012 www.maurymw.com 1

General Load Pull Overview 2

Outline 1. Introduction to Maury Microwave 2. Basics and Benefits of Load Pull 3. Pulsed Measurements 4. Harmonic Load Pull 5. Active Tuning 6. Load Pull with X-Parameters 7. Ultra-Fast Noise Parameters 8. Summary 3

Maury Microwave Corporation Headquarters 2900 Inland Empire Blvd, Ontario California, USA Load Pull Overview 4

1957 Incorporated in California Experts in Precision Machining for Microwave Components 1980 s Aligned with HP/Agilent to provide ALL 8510 Cal Kits 1987 World s First Commercial Automated Load Pull and Noise Parameter System 2004 Launched LSNA (Large Signal Network Analyzer) 2008 Load Pull with X-Params 2008 Ultra-Fast Noise Params* *Patent Pending 5

Certification, Conformance, Accountability AS9100 & ISO 9001 CE Conformance ANSI/NCSL Z540-1 MIL-STD-45662A Traceability to NIST (National Institute of Standards & Technology) 6

Space Flight Applications One Piece Flange and Body No Braze Joints No Dielectric Versions For High Radiation Requirements Optimized Tuning 7

Measurement Science Knowledge At Maury we continue to value credibility, honesty, integrity Daily Interaction with Industry Experts in Leading Companies Noise Parameter Testing Power Load Pull/Source Pull Nonlinear/Large Signal S-Parameters X-Parameters 8

Experts In Full System Integration Solutions Experienced Technical Team Waveguide & Coaxial Turnkey & Customized Solutions 9

Maury Product Overview Components Device Characterization 10

Market Leader Device Characterization Emphasis in Automated Impedance Control USB Controlled Tuners Software - Automated Measurements Load Pull & Source Pull Large-Signal / Nonlinear Measurements Noise Parameter Measurements Agilent Global Channel Partner* 11

What is a Tuner? A device to control impedance Probe Reflection Control Method: Magnitude move probe up / down Phase move probe horizontally Probe End View Mismatch probe in a 50-Ohm slab line USB Interface Side View 12

Automated Mechanical Tuners 0.25 110 GHz Coaxial and Waveguide Repeatability > 40 or 50 db USB Control ATS software or DLL Reliable and proven technology More than 20 years in Tuner Technology 14

What Is Load Pull? Measurement vs. Impedance 16

What is Load Pull Stimulus and Measurement TUNER DUT TUNER source Load 17

Why Load Pull? Characterize Non-Linear Devices Small Signal S-Parameters Not Sufficient Must Measure with Actual Operating Conditions Most Power Devices are Far from 50 Ohms Stability / Ruggedness Test Linear or Non-Linear Device Test at High Reflection, All Phases Characterize Noise Parameters Measure vs. Source Impedance Load Match Optional - to Reduce Uncertainty 18

Benefits of Load Pull See impedance matching trade-offs Allows design to specs Faster time to market, Eliminate cut-and-try Develop High-Efficiency Power Amps Test Stability under any mismatch S-parameters not accurate for large-signal Osc. Test Ruggedness Improve Reliability Avoid amplifier degradation or failure 19

Basic Load Pull Setup Power Measurements Measure Pout, Gt, Bias, Eff, ACPr vs. Gamma (load or source) 20

Load Pull with AM/PM Power Measurements Measure Gamma_in, AM/PM, Gp, True PAE, plus basic parameters vs. Gamma 21

Load Pull with Vector Receiver Measure a, b Waves Derive All Other Parameters 22

Load Pull Stimulus Measure vs. Gamma (load or source) Measure vs. Power Measure vs. Bias Measure vs. Frequency Sweep Plan: Measure vs. All Variables Result is Multi-Dimensional Data Covering Complete Region of Operation 23

Typical Measured Parameters Pout, Gain, Bias, Efficiency Linearity ACPr, EVM, AM/PM, Intermodulation... Spurious Signals, in, Waveforms... ATS Software Version 5 76 Built-in Parameters 25 User Defined Parameters 24

Load Pull Contour Display Displays Constant Parameter Contours 25

On-Wafer Load Pull 26

Pulsed IV measurements Short pulse : Quasi-isothermal conditions Low duty cycle : Constant mean temperature Quiescent bias point : Thermal conditions fixed Several quiescent bias point 28

Pulsed S-Parameters 29

Pulsed IV & S parameter measurements Synchronization between Pulse IV and Measurements 30

Commercial compact FET models Mostly used models for GaN HEMTs FET models Number of parameters Electrothermal effect Trapping Effects Original Device Context Curtice3 [1] 59 No No GaAs FET CFET [2] 53 Yes No HEMT EEHEMT1 [3] 71 No No HEMT Angelov [4] 80 Yes No HEMT/MESFET AMCAD HEMT1 [5] 65 Yes Yes GaN HEMT AMCAD GaN HEMT1 is the only model here with a complete extraction flow based on pulsed IV/RF measurements 31

Pulsed Bias/RF Load Pull Power Measurements Power Meter Spectrum Analyzer Pulsed System Power Sensor Signal Generator WA Attenuators Coupler TUNER TUNER Coupler Power Sensor 32

Harmonic Load Pull 34

Harmonic Load Pull Multiplexer Method Cascaded Tuner Method Active Tuning Method 35

Harmonic Load Pull Multiplexer Method RF Source Tuner F1 DUT T R I P L E X E R Tuner F3 Tuner F2 Tuner F1 Load Load Power Meter Separate Fundamental / Harmonics with Filters 36

Harmonic Load Pull Cascaded Tuner Method RF Source Tuner F1 DUT Tuner F1,2,3 Tuner F1,2,3 Tuner F1,2,3 Power Meter Select combination of tuner states to simultaneously set F1, F2, F3 37

Active Tuning DUT b2 a2 Synthesize Gamma by Injecting Signal that Acts Like Reflection Reflection 39

Hybrid Active Tuning DUT TUNER b2 a2 Synthesize Gamma by Adding Injected Signal to Passive Reflection High Power Amp NOT Required Reflection 40

Hybrid Setup Passive + Active Tuning Passive Tuner Reflects Most F1 Power Eliminates Very High Power Amp 41

Hybrid tuning setup A DUT Passive Tuner PA 42

Hybrid tuning setup B DUT Passive Tuner F1 F2 PA 43

Hybrid tuning setup C PA DUT Passive Tuner 44

Active tuning with the PNA-X PNA-X a1 b1 a2 b2 F2 F1 DUT PA F3 45

Active tuning with X-Parameters PNA-X Drive Signal a1 b1 a2 b2 Port 1 Port 3 Extraction Signal PA DUT PA F1 46

Hybrid tuning with the PNA-X PNA-X a1 b1 a2 b2 F2 F1 Passive Tuner DUT Passive Tuner PA F3 47

Easy Upgrade With Vector Receiver Load Pull Add One Source per Harmonic With PNA-X, Use Built-in Source 48

Active Load Pull Benefits: Simple (with Vector Receiver) Can Achieve Gamma = 1 or greater Overcome Fixture Losses Higher Gamma at Fundamental Independent Harmonic Tuning Accuracy based on single VNA cal Economical Add-on to Existing ATS 49

New Paradigm for PA Design Instant Large Signal Model Major Breakthrough Dream of Microwave Engineers for years Solution: Load Pull + NVNA + ADS 51

Load Pull with LSNA Time Domain - 2004 52

Time Domain Data Absolute Mag and Phase of a and b Waves At Fundamental and Harmonics a1 b1 DUT a2 b2 Calibrated Reference Planes 53

NVNA - 2008 Nonlinear Vector Network Analyzer Superset of LSNA Large Signal Network Analyzer Measures Time Domain Measures Time Domain and X-Parameters X-Parameters are Unique to Agilent NVNA X-Parameters act as Large Signal Model 54

S-Parameters VNA measures amplitudes and phases of Linear signals. 55

X-Parameters NVNA measures amplitudes and phases of non-linear signals. 56

X-Parameter Summary Non-Linear Generalization of S-Parameters S-Parameters are special case where distortion = 0 Data covers one Large Signal Operating Point Acts like Large Signal, Non-Linear Model In Region near Large Signal Operating Point 57

Multiple X-Parameter Measurements Load Pull measurements combine to cover region of interest. 50 Ohm Region 58

Solution: Load Pull with X-Parameters Instant Large Signal Model 59

Instant Large Signal Model X-Parameter Component Industry Breakthrough 60

Instant Large Signal Model Maury Load Pull Agilent NVNA Maury SW in PNA-X Run a Sweep Plan Save X-Parameter File Simulate in ADS 61

Load Pull + NVNA Setup PNA-X NVNA + Maury Software Limited Power Maury Tuner DUT Maury Tuner Bias System 62

High Power Measurement Setup Input Amp PNA-X Output Amp DUT Use Front Panel Links to Receivers 63

High Power Calibration Setup Input Amp PNA-X Cal Stds Output Amp 64

High Power Load Pull with X-Parameters 20 Watt Setup Simple Setup Use Front Panel Links Main thing Power Budget Protect NVNA Protect Amps 65

Traditional Large Signal Model Use Small Signal and DC measurements Fit Model Parameters to Data Time-Consuming Accuracy Limited in Some Regions of Operation Technology Dependent Model is Extrapolated to Large Signal Load Pull is Accuracy Reference Load Pull is Measured at Actual Large Signal 66

X-Parameter Large Signal Model Based on Large Signal Measurements of X-Parameter Data Technology Independent Hides Design Details of Device Load Pull Sweep Plan for PA Design New Paradigm for Modeling and Design 67

Comparison of Models Traditional Circuit-based Model Time consuming Extrapolate to Large Signal Can relate to Physics Can be Scalable X-Parameters Behavioral Model Instant Model Large Signal Directly New Paradigm for PA Modeling and Design 68

Results Packaged FET Load Pull Overview Extremely Accurate Agreement Pout PAE Blue Simulated, Red - Measured 69

MeasuredVoltage SimulatedVoltage SimulatedCurrent MeasuredCurrent MeasuredVoltage SimulatedVoltage Load Pull Overview Results - Packaged FET Extremely Accurate Measurements X-param Simulation WJ FP2189 1W HFET P out Contour (dbm) Measured and Simulated Voltage and Current Waveforms 16 0.5 14 0.4 12 10 0.3 8 0.2 6 0.1 4 SimulatedCurrent MeasuredCurrent 2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 time, nsec Measured and Simulated Voltage and Current Waveforms 20 0.30 0.30 Measured and Simulated Dynamic Load Line 15 10 5 0 0.25 0.20 0.15 0.10 SimulatedCurrent MeasuredCurrent 0.25 0.20 0.15 0.10-5 0.0 0.2 0.4 0.6 0.8 1.0 time, nsec 0.05 0.05-2 0 2 4 6 8 10 12 14 16 18 MeasuredVoltage SimulatedVoltage 70 Agilent Restricted 70 August 21, 2009

Harmonic Tuning Simulation with X-Parameters Fundamental Load Pull with X-Parameters Simulate Harmonic Load Pull in ADS Benefits: Reduce Setup Complexity Reduce Measurement Time Reduce Data Storage 71

Experiment to Test Simulation with X-Parameters Setup 1 F1, F2, F3 Load Pull Passive Tuner DUT Cree 10W GaN Passive Tuner Passive Tuner Passive Tuner Setup 2 F1 Load Pull Passive Tuner DUT Cree 10W GaN Passive Tuner With X-Parameters 72

Measured_Iout XParam_Iout Measured_Vout XParam_Vout gamma[1] gamma[2] gamma[3] XParam_PAE Measured_PAE Load Pull Overview Blue Measured 3 Harmonic Load Pull Red Simulated from F1 Load Pull Harmonic Load Simulation 2.0 1.5 1.0 0.5 0.0-0.5 Load conditions (0.000 to 0.000) Dynamic Load Line -10 0 10 20 30 40 50 60 70 XParam_Vout Measured_Vout Not in X-Parameters Measurement Excellent Agreement 70 60 50 40 30 20 10 0-10 70 60 50 40 30 20 10 PAE vs. Available Input Power 4 6 8 10 12 14 16 18 20 Pin_avail Voltage and Current Waveforms at output 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 time, nsec 2.0 1.5 1.0 0.5 0.0-0.5 Measured_Iout XParam_Iout 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 time, nsec 73

Software Integration ATS Software Version 5 Installs Inside PNA-X Tightly Coupled with PNA-X / NVNA Full X-Parameter Support 74

Selecting X-Parameters On/Off Turn off X-Parameters for Faster Initial Tuning Turn on X-Parameters for Sweep Plan to make Large Signal Model 75

Sweep Plan Up to 7 Variables Fully Automated Measurement Sweep Plan Setup Measure over Full Range of Device Operation 76

Hybrid Active Tuning Setup Passive F1 Tuner Active Harmonic Simple User Controls Tuning Example Configuration 77

Passive / Active / Hybrid Tuning Same User Controls 78

Noise Parameters Consist of Four Scalar Values Most Common Set: Fmin Minimum noise figure opt Optimum magnitude opt Optimum phase rn - Equivalent noise resistance s - opt 2 F= Fmin + 4rn 1+ opt 2 (1 - s 2 ) 80

Benefits of Noise Parameters See noise figure for any source impedance Combine with s-parameters for complete Low Noise Amplifier (LNA) design Accurate prediction of LNA performance Maury Ultra-Fast Method (200X+ Faster)* Faster time to market Simpler More Accurate than traditional approach Test in production = competitive advantage *Patent Pending 81

Noise Parameter Measurement Basic Setup 82

Noise Parameter Measurement Sequence 1. System Cal 2. Receiver Cal 3. DUT Measurement 83

Noise Parameter Measurement Traditional Method 1. System Cal Characterize Tuners Over Entire Chart One Frequency at a Time 2. Receiver Cal and Measurement One Frequency at a Time Allows Ideal Impedance Pattern 84

Noise Parameter Measurement Traditional Method Time Consuming Can Have Drift Issues Use System Cal for Long Time To Save Time Calibrate Parts Separately Based on 1969 Paper Used by everyone for Almost 40 years 85

Noise Parameter Measurement New Method* Main Idea: Characterize One Set Of Tuner States Sweep Frequency at Each State Take Advantage of Fast Sweep of Modern Instruments *Patent Pending 86

Noise Parameter Measurement New Method PNA-X with Noise Option Noise Source Maury Tuner DUT 87

Noise Parameter Measurement New Method 88

Noise Parameter Results Old Method 73 Frequencies 30 Hours 89

Noise Parameter Results Old Method Assoc Gain Гopt Fmin rn Same Data, 0.5 GHz Steps 90

Noise Parameter Results New Method* 73 Frequencies 8 Minutes, 224x Faster *Patent Pending 91

On-Wafer, 0.8-18 GHz Fmin= 0.4 db Measured Data, No Smoothing Applied 92

Noise Parameter Measurement 50 GHz PNA-X with Noise Option Noise Source Tuner DUT Noise Receiver Module 93

Noise Receiver Module 50 GHz Operation 1. Thru Path for S-Parameters 2. Direct Noise Path (Below 26.5 GHz) 3. Down-Convert Path (26.5 to 50 GHz) USB Control 94

On-Wafer, 8-45 GHz Fmin= 0.4 0.7 db Measured Data, No Smoothing Applied 95

Maury Ultra-Fast Noise Parameters Industry Breakthrough (224x Faster) Simpler Setup and Measurement Better Accuracy Works Well to Very Low Noise 96

Summary Load Pull Applications Nonlinear Devices Stability / Ruggedness Harmonic Tuning Active Tuning Pulsed IV Load Pull with X-Parameters New Paradigm for Modeling and Design Instant Large Signal Model Major Industry Breakthrough Ultra-Fast Noise Parameters 200 Times Faster Much Simpler More Accurate Major Industry Breakthrough 97