Digital Pre-Distortion system for GaN power amplifier on base station Sang-Hyun Chun, Kyu-Jin Choi, Jin-Ho Kim and Jong-Heon Kim < Kwangwoon University > Seoul, Korea 2010. 6. 30 Smart Radio System Lab. Contents • Introduction • Principle of Digital Pre-distortion Linearization • Connected Solution • Linearization Test of DPD • Result (Linearization Performance) • Conclusion Smart Radio System Lab. Introduction • Requirements of PA for Wireless Communication - High quality of transmission (High Linearity) - Low power consumption (High Efficiency) - Require digital control on hardware platform to reset and reuse (Improving reliability, Advantageous maintenance) Requirements High Linearity (Improving quality of Transmission) High Efficiency (Reduce Power Consumption) Solution Using Linearization Technique (Feed Back, Feed Forward, Pre-Distortion.. etc)) Structure of High Efficiency PA (Doherty, EER, Envelope Tracking.. etc) Smart Radio System Lab. Introduction • Comparison Linearization Techniques Analog PD Feed back Feed forward Digital PD Bandwidth very wide Narrow wide moderate Linearity good good very good very good Complexity medium medium high high Power Efficiency High high low high • Strength of Digital Pre-distorter (Digital PD) - Better power efficiency than feed forward technique using widely - Feed forward technique is optimized only single mode PA, but DPD can be applied multimode by resetting the system (maintenance widely use) - The most suitable linearization technique for next generation linearized PA requiring the digitally control hardware platform Smart Radio System Lab. Principle of DPD linearization • Compensate non-linearity of PA - Self-characteristic of PA(Non-linearity) : Amplitude and Phase Distortion (Characteristic of output power is compressed due to increasing input power) - Non-linearity of PA : Decline efficiency of output spectrum - Pre-distorted Linearization : Input signal is distorted intentionally to cancel the non-linear output characteristic - Improve the linearity of PA by expending output according to increase input - Digital Pre-distorted linearizer is made up of base-band signal processing Smart Radio System Lab. Principle of DPD linearization Principle of DPD system • DPD linearization using adaptive algorithm - Modeling the PA property using Polynomial method - Compute the error comparison Desired and Output signal - Optimized linearization performance when DPD coefficient W(n) is minimum value - Actively, adapt environmental change and TR property by using the real-time updating coefficients Smart Radio System Lab. Connected Solution • Conventional measurement method of DPD system - Need a part of signal generation - Need a part of digital control - Need a converting device of digital / analog - Need Up/Down Converter • Measurement method of DPD system using Connected Solution - Using control PC ADS : Signal generation and Control the signal, DPD system MATLAB : Compute the DPD coefficients I/O Libraries Suite : Interlocking Control PC and Measurement Device - PSG 8267D : Signal generator, D/A Converter, Up Converter - PSA E4440A, VSA: Down Converter, A/D Converter Smart Radio System Lab. Connected Solution Schematic of DPD system using Connected Solution Smart Radio System Lab. Connected Solution(role of each part) 1. Signal Generator I(t) Q(t) DUT DAC LO PC(ADS) PSG8267D • Generating WiBro 1FA Signal and Linking PSG - Generating Signal using WiBro Downlink Block of ADS - Using ESG Sink Block of ADS to link PSG - Insert I/Q signal of base-band -> D/A Conversion -> Up-Conversion -> Generating RF test signal Smart Radio System Lab. Connected Solution (role of each part) 2. Spectrum Analyzer • Analysis output signal of PA - Output signal of PA converts base-band signal (Down Conversion) - Analog signal converts to Digital Signal for Data processing of control PC - Consider 3rd IMD and Bandwidth of WiBro 1FA signal(8.75MHz) to linearization : Require 30MHz bandwidth which is 3 times wider than bandwidth of signal source - Output signal of PA -> Convert base-band signal -> A/D Conversion -> Insert control PC Smart Radio System Lab. Connected Solution (role of each part) 3. Software • ADS (Advanced Design System) - Generating Input signal to insert the PA Link signal generator as I/Q signal of base-band Link control PC(ADS) with measured PA output signal data from spectrum analyzer Using VSA Sink Block Construct DPD system Link to MATLAB for computing DPD coefficients • MATLAB - Computing and Saving DPD coefficient • VSA (Vector Signal Analyzer) - Receive output signal from spectrum analyzer and save as data file to connect ADS • I/O Libraries Suite - IO Libraries Suite Connect measurement device and control PC Smart Radio System Lab. Linearization Test of DPD Test Method • Test Method 1. Run PSG Simulation (Initialization) : Signal generated by control PC(ADS) is inserted as input signal through signal generator 2. Run VSA Simulation : Link output signal of PA to control PC, Synchronize input and output signal 3. Generate Memory Polynomial Coefficients : Generating DPD coefficient using in/output signal of PA 4. Run PSG Simulation (with Pre-distortion) : Insert pre-distorted signal to the PA Test Process Smart Radio System Lab. Linearization Test of DPD 1. Run PSG Simulation (Initialization) Run PSG Simulation (Initialization) • Run PSG Simulation (Initialization) Spectrum and CCDF of input signal - Process of insert signal source to signal generator after generating test input signal of PA - To Link WiBro signal source block of ADS and PSG by using ESG sink block - Insert scale factor to Control between amplitude of signal source and PSG source - Test input signal converts I/Q signal to link PSG Smart Radio System Lab. Linearization Test of DPD 2. Run VSA Simulation • Run VSA Simulation - Process of converting the available data for PC. Captured output signal of PA from PSA is converted through VSA. - Synchronization after Evaluating delay time of generated In/Output data rows Run VSA Simulation - Increase samples of primary input data rows to evaluate accuracy delay time - Convert primarily number of data samples after compensation of evaluating delay time using Cross-correlation Synchronized In/Output Data Smart Radio System Lab. Linearization Test of DPD 3. Generate Memory Polynomial Coefficients Run VSA Simulation • Generate Memory Polynomial Coefficients - Synchronized In/Output data rows on ADS save and transfer to MATLAB data rows through the ADS to MATLAB sink Modeling the PA using synchronized In/Output data rows from MATLAB Compute and save the DPD coefficients to apply DPD algorithm which Consider the Memory Effect to improve linearity of PA Smart Radio System Lab. Linearization Test of DPD 4. Run PSG Simulation (with Pre-distortion) Run PSG Simulation • Run PSG Simulation - Generating Input signal using DPD coefficient from Generate Memory Polynomial Insert pre-distorted signal to signal generator using ADS Conform linearized output signal to insert pre-distorted to non-linear PA Smart Radio System Lab. Results GaN Power Amplifier • Doherty Power Amplifier using GaN power transistor - Fc : 2.345GHz - Max. output power 51.5dBm - Ave. output power 42 dBm (9.5dB Back-off) Structure of Doherty Power Amplifier Smart Radio System Lab. Results GaN Power Amplifier Output Spectrum of GaN PA Output Spectrum of PA with DPD(P:7, M:3) Results Before Linearization (dBr) After Linearization (dBr) Quantity of Improvement(dBr) Lower (@ -4.77 MHz) -35.47 -46.57 11.1 Upper (@ +4.77 MHz) -36.53 -49.9 13.37 Smart Radio System Lab. Conclusion • Construct Test Bed using Connected Solution - Complex test environment can be simple using the connected solution - Effective and time-saving test environment - Problem resolve easily and quickly through connecting MATLAB and ADS - Available for DPD test for various signal and different method • Digital Pre-distorted Linearization System - Develop DPD algorithm with Memory Polynomial for WiBro base station GaN PA using MATLAB - Construct DPD linearization system using ADS(Ptolemy) and Connected Solution Test Bed - Conform the linearization performance about 12dB result from test for WiBro base station GaN PA Smart Radio System Lab. Reference [1] Agilent Application Note 5989-8309EN [2] Agilent Application Note 5988-6044EN [3] Agilent Datasheet 5988-0697EN [4] Agilent Configuration Guide 5989-1326EN [5] Agilent Datasheet 5980-1284E [6] Agilent Configuration Guide 5989-2773EN [7] Agilent Technical Overview 5989-1679EN [8] TTAS.KO-06.0098, “2.3GHz Radio Conformance Test Specification for 2.3GHz band Portable Internet Service” Dec.2005 [9] J. Vuolevi, T. Rahkonen, J. Manninen, "Measurement technique for characterizing memory effects in RF power amplifiers," IEEE Trans. Microwave Theory and Tech, vol. MTT-49, no. 8, pp. 1383-1389, Aug. 2001. Smart Radio System Lab.