Sub 1 Ohm Broadband Impedance Matching Network Design Methodology for High Power Amplifiers W. McCalpin dBm Engineering, Inc., Boulder, CO February 19, 2020 dBm Engineering 5446 Conestoga Ct. Boulder, CO 80301 dBmEngineering.com High Power Amplifier Design Challenges A fixed-tuned broadband 50-Ohm test fixture is required with the following design constraints: Prototype RF Power Transistor: Single-Ended 250W Pulsed LDMOS part The design bandwidth required is 15% (1200 to 1400 MHz) as compared to ~ 3% for Wireless bands High Power Pulsed Load Pull measurements have produced sub-1 Ohm target Load / Source impedances (ZLoad=0.8 j1.0) to be held nearly constant over the bandwidth of interest February 19, 2020 Prototype 250W Pulsed Power LDMOS

Transistor February 19, 2020 Outline Motivation Load Pull Impedance Accuracy Proposed Simulation / Design Method Matching Network Topology Selection Simulation of Ideal Topology / Trajectory Conversion of Ideal Circuit Simulation to a Physically-based Circuit Simulation Fabrication of TRL Verification and DUT fixtures Comparison of Measured vs. Simulated Trajectory and ZLoad Summary February 19, 2020 Motivation Q: Why is it difficult to go from High Power Load Pull measurements to working hardware?

Q: Why do RF Power engineers still commonly rely heavily upon intuition, tuning and iteration? Is Load Pull wrong? The target impedances generated from doing Load Pull may not be accurate enough to truly represent what the DUT actually sees. and / or Is the simulation / design process wrong? Common simulation techniques may not produce hardware that faithfully recreates the target impedances (when the designer fabricates the physical circuit from the February 19, 2020 simulated circuit). Load Pull Tuner Impedance Accuracy (calibrated impedance vs. actual impedance during measurement) e.g. for a ZCal point, what is the Output Tuners Load reflection coefficient when ZMeas is normalized to the ZCal point ZCal ZMeas ZMeas

50 ZMeas ZCal ZMeas ZCal R.L.(dB ) 20 log 10 | | February 19, 2020 Load Pull Tuner Impedance Accuracy (cont.) Focus Microwaves - PMT Load Pull System * The Tuner by itself: 50 ZMeas Tuner * * * * * *

Tuning to the indicated points: Worst Case R.L.(dB) meas. = ~ - 51 dB Tuning repeatability: Worst Case R.L.(dB) meas. = ~ - 47 dB February 19, 2020 (Based on a TRL 7/16 Connectorized VNA calibration with Sii -60 dB) ZLoad Impedance Accuracy at the DUT plane The impedance presented to the DUT: * * * * * * * ZMeas

Load Pull Test Fixture 50 Tuner Tuning to the indicated points: Worst Case R.L.(dB) meas. = ~ - 31 dB Tuning repeatability: Worst Case R.L.(dB) meas. = ~ - 31 dB February 19, 2020 (Based on a TRL Impedance Transforming PCB VNA calibration with Sii -41 dB) Load Pull Tuner Impedance Accuracy (cont.) Given the proper discipline in calibration and setup, the accuracy of the Load Pull targets is very good Load Pull targets should be useful for design as well as device characterization February 19, 2020 Proposed Simulation / Design Method

Ideal impedance matching network topology / trajectory simulation (using circuit models and lumped elements) Identification of the circuit response to tuning variations Conversion of distributed elements to EM based multi-port S-parameter blocks Conversion of lumped element models to measured one-port and two-port S-parameter blocks Simulation refinement and design centering February 19, 2020 Proposed Simulation / Design Method (cont.)

Design and fabrication of a TRL verification fixture and the break-apart DUT impedance matching fixture Assembly of each half of the DUT fixture with component by component verification of the impedance matching trajectory using the TRL verification fixture Full DUT fixture assembly Top-level tuning and design centering Break-apart measurement of the final Load and Source impedances. February 19, 2020 Ideal Topology / Trajectory Simulation Output Matching Network Z0 = 4.48 Ohms 50 Ohms Load Matching Network: Two-section Low-Pass with ideal elements SRLC SRLC26 Term

Term16 Num=16 Z=50 Ohm MLIN TL43 MLIN TL42 SRLC SRLC25 SRLC SRLC33 MLIN TL41 SRLC SRLC32 SRLC SRLC27 MLIN TL40 SRLC SRLC28

MLIN TL39 MLIN TL38 SRLC SRLC30 MLIN TL37 SRLC SRLC31 SRLC SRLC29 February 19, 2020 Section 2 MLIN TL36 Section 1 R

R2 R=50 Ohm Conversion of Distributed Elements to EM based Simulation (using Momentum) Generates a 13 port .s13p file to import into ADS to combine February 19, 2020 with the measured .s1p files of the components Conversion of Lumped element circuit models to measured s1p and s2p files Measured using 50-Ohm TRL standards either wafer-probed (preferred) or connectorized (as shown) Using the application substrate and component mounting configuration to include all substrate parasitic effects Commercially available Model Libraries are ideal parameterized to substrate material, full range of values by product type, accurate and February 19, 2020 wideband to include harmonic frequencies, etc. Design and Fabrication of a Break-Apart DUT Impedance Matching Fixture ZLoad

Fabricated to be Break-apart using the substrate selected for the application February 19, 2020 Design and Fabrication of an Impedance Transforming TRL Verification Test Fixture Fabricated on the substrate to be used in the application with a line width equal to the lead width of the DUT (500 mils) February 19, 2020 Measurement of the DUT Break-apart fixture using the TRL Verification fixture ZLoad After TRL calibration using the Verification fixture, multiple measurements can be made as each lumped component is February 19, 2020 added ''C:\ADS\GHz\3_block_output_prj\data\PA_paper_MOM_Output_8.ds''..S(1,1) PA_paper_MOM_Output_4..S(1,1) PA_paper_MOM_Output_3..S(1,1)

''C:\ADS\GHz\3_block_output_prj\data\PA_paper_MOM_Output_2.ds''..S(1,1) PA_paper_MOM_Output_4..S(1,1) PA_paper_MOM_Output_5..S(1,1) PA_paper_MOM_Output_3..S(1,1) ''C:\ADS\GHz\3_block_output_prj\data\PA_paper_MOM_Output_6.ds''..S(1,1) ''C:\ADS\GHz\3_block_output_prj\data\PA_paper_MOM_Output_2.ds''..S(1,1) PA_paper_MOM_Output_4..S(1,1) PA_paper_MOM_Output_5..S(1,1) PA_paper_MOM_Output_3..S(1,1) S(10,10) ''C:\ADS\GHz\3_block_output_prj\data\PA_paper_MOM_Output_2.ds''..S(1,1) PA_paper_MOM_Output_13..S(11,11) PA_paper_MOM_Output_3..S(1,1) S(9,9) ''C:\ADS\GHz\3_block_output_prj\data\PA_paper_MOM_Output_2.ds''..S(1,1) PA_paper_MOM_Output_12..S(11,11) S(8,8) ''C:\ADS\GHz\3_block_output_prj\data\PA_paper_MOM_Output_2.ds''..S(1,1) PA_paper_MOM_Output_11..S(11,11) PA_paper_MOM_Output_13..S(11,11) S(7,7) PA_paper_MOM_Output_10..S(11,11) PA_paper_MOM_Output_12..S(11,11) PA_paper_MOM_Output_13..S(11,11) S(6,6) PA_paper_MOM_Output_13..S(13,13) PA_paper_MOM_Output_9..S(10,10) PA_paper_MOM_Output_11..S(11,11)

PA_paper_MOM_Output_12..S(11,11) PA_paper_MOM_Output_13..S(11,11) S(5,5) S(4,4) PA_paper_MOM_Output_13..S(11,11) PA_paper_MOM_Output_9..S(1,1) PA_paper_MOM_Output_10..S(11,11) PA_paper_MOM_Output_11..S(11,11) PA_paper_MOM_Output_12..S(11,11) PA_paper_MOM_Output_13..S(11,11) S(1,1) S(5,5) S(4,4) S(4,4) S(1,1) S(4,4) Simulated vs. Measured ZLoad Impedance Trajectory Term Term16 Num=16 Z=50 Ohm MLIN TL43 Chart Zo = 4.48 Ohms

Target Load Impedance = 0.8 - j1.0 freq freq freq freq (1.200GHz (1.200GHz (1.200GHz (1.200GHz toto to 1.400GHz) to 1.400GHz) 1.400GHz) 1.400GHz) freq (1.200GHz to 1.400GHz) freq freq(1.100GHz (1.100GHztoto1.500GHz) 1.500GHz) Load Matching Network: Two-section Low-Pass with ideal elements SRLC SRLC26

MLIN TL42 SR LC SR LC25 SR LC SR LC 33 MLIN TL41 SRLC SRLC32 Section 2 SRLC SRLC27 MLIN TL40 SRLC SRLC28 MLIN TL39 MLIN TL38

SRLC SRLC30 MLIN TL37 SRLC SRLC31 February 19, 2020 Section 1 MLIN TL36 SRLC SRLC29 R R2 R=50 Ohm Pulsed P1dB, Efficiency and Gain vs. Frequency 60.00 14 59.00

13 Gain at P1dB 58.00 57.00 56.00 11 Eff. at P1dB 10 55.00 54.00 12 9 P1dB Pout vs Freq Eff. at P1dB Gain at P1dB

53.00 8 7 52.00 6 51.00 5 50.00 1200 February 19, 2020 1250 1300 Frequency (MHz) 1350 4 1400

Gain at P1dB (dB) P1dB (dBm) and Drain Eff. at P1dB (%) Test conditions: Vdd=32V, Idq = 500 mA (pulsed), Pulse width = 6mS, Duty Cycle = 25% Summary Invest in your Methodology! High Power Load Pull is useful for design as well as characterization Choose a topology that will hit the required impedance targets and achieve the desired matching network trajectory Replace the ideal circuit elements with EM and measurement based elements as early as possible Verify each simulated step by measuring at every step Know the impacts of tuning locations and keep them small Center and finalize the tuning and measure the final impedances 2020 Improve the process February 19,

References [1] D. Williams and D. Walker, On-Wafer Measurement Accuracy, ARFTG Short Course on Measurements and Metrology for RF Telecommunications, November 2000 [2] http://www.boulder.nist.gov/micro February 19, 2020