GaN HPA optimized for telecom - Linearity results & DPD assessment March 2017 christophe.auvinet@ums-gaas.com
GaN technology toward 5G 1. Toward 5G with GaN 2. AB class HPA optimization 3. Doherty linearity assessment 4. Conclusion Date / Ref doc 2
GaN technology toward 5G 1. Toward 5G with GaN 2. AB class HPA optimization 3. Doherty linearity assessment 4. Conclusion Date / Ref doc 3
Toward 5G with GaN Fixed-Wireless What is the need? Fixed-wireless access as the first phase of 5G deployments AT&T, Verizon, Nokia trials in the US mmwave spectrum: 27.5-28.35 / 37-38.6GHz / 38.6-40 / 57-71 GHz Qualcomm Snapdragon X50 5G modem (800 MHz / 28GHz 5Gb/s) Maxlinear (former Broadcom) BCM85100 (FDD / 60 GHz 10 Gbps) wwwave ICs suitable for 5G fixed-wireless access Date / Ref doc 4
Toward 5G with GaN TRx for Phase Array Front-end technology? Phase Array antenna and massive MIMO techniques Ericsson, Huwaei, NTT Docomo & Samsung trials User and Spatial multiplexing increasing bit rates Beamforming for propagation loss compensation Phase array antenna TRx device TRx Half-duplex architecture Key parameters: PS PA Efficiency Dimension, Integration SW PS LNA SW Cost Linearity Broad bandwidth GaN technology is offering High Power Density & High PAE Date / Ref doc 5
Toward 5G with GaN Design & Challenges Application requires linearity To pass spectrum requirement System figure of merit to translate into component requirement Linked to intrinsic linearity of the HPA Performances defined at back-off vs Psat (due to PDF) PAE vs Output Average Power TX chain to be compatible with Digital Pre-Distorsion system System validation required DPD to be implemented Generation of RF vector Signal Test Bench Synoptic DRV DPD Loop DUT Sampling, Demodulation & Analysis Date / Ref doc 6
Toward 5G with GaN Design & Challenges New approach for circuit design 2 Tones Load-pull on GH25 Tr GaN / 10W Run#1 Goals: Demonstrate high linearity on GH25 Provide HPA compatible with DPD Simu vs Meas. Challenges: Modeling accuracy Memory effects (traps, decoupling circuit, ) Unbalanced spectrum GaN / 10W Run#1 GaAs / CHA6552-QJG GaN / 10W Run#1 Date / Ref doc 7
GaN technology toward 5G 1. Toward 5G with GaN 2. AB class HPA optimization 3. Doherty linearity assessment 4. Conclusion Date / Ref doc 8
AB class HPA GH25 Example Main Features 5.5-7.5 GHz Psat > 39dBm Gain = 18dB C/I3 = 34dBc @10dB back-off MER wodpd = -27dB @30dBm MER withdpd < -50dB @30dBm DC bias: Vd = 25V / IdQ = 340mA UMS GaN 0.25um / QFN 6x5 / MSL3-20 -24-28 MER& ACPR (db) versus Output Power (dbm) Vd=25V, IdQ1=220mA / IdQ2=200mA QAM256 / CS=56MHz / RRC=0.2 MER & ACPR (db) -32-36 -40-44 -48-52 -56 MER w/o DPD MER with DPD ACPR w/o DPD ACPR with DPD >20dB Date / Ref doc -60 25 26 27 28 29 30 31 32 33 34 35 36 Average Output power (dbm) GaN HPA compatible with low consumption DPD 9
AB class HPA Way to design RUN1: Classical trade-off Pout/PAE PSAT > 3W/mm (@8dBcomp) PAEmax # 35% High Linear Gain # 22dB Smooth compression Wide band Run1 C/I3 improved & flattened Linear Gain & Frequency Band reduced RUN2: Linearity oriented Run2 Same output stage size Tighter AM/AM & AM/PM variation Optimum impedance & biasing for IM3 Back-off between stages increased Enhanced on chip decoupling 2nd harmonic load Date / Ref doc 10
AB class HPA Designs Comparison IMD3 vs Pout IMD3 vs Pout IMD3 (dbc c) -16-18 -20-22 -24-26 -28-30 -32-34 -36-38 -40-42 -44 At nominal biasing IMD3_Low Run#1@Nominal IMD3_High Run#1@Nominal IMD3_LOW Run#2@Nominal IMD3_High Run#2@Nominal -46 22 24 26 28 30 32 34 36 38 Output Power DCL (dbm) IMD3 (dbc c) Date / Ref doc 11-16 -18-20 -22-24 -26-28 -30-32 -34-36 -38-40 -42-44 At optimum biasing IMD3_Low Run#1@Optimum IMD3_High Run#1@Optimum IMD3_LOW Run#2@Optimum IMD3_High Run#2@Optimum -46 22 24 26 28 30 32 34 36 38 Output Power DCL (dbm) IMD3 improvement at 30dBm (# 10 B.O.) Linearity better around average power for a 256QAM signal
AB class HPA Enhanced decoupling ) IMD3 (dbc) -16-18 -20-22 -24-26 -28-30 -32-34 -36-38 -40-42 -44 IMD3 vs Pout Run#2 IMD3_LOW @10kHz IMD3_High @10kHz IMD3_Low @100kHz IMD3_High @100kHz IMD3_Low @1MHz IMD3_High @1MHz IMD3_Low @2MHz IMD3_High @2MHz IMD3_Low @5MHz IMD3_High @5MHz IMD3_Low @10MHz IMD3_High @10MHz IMD3_Low @20MHz IMD3_High @20MHz IMD3_Low @30MHz IMD3_High @30MHz IMD3_Low @40MHz F = 10 khz to 40MHz Un-balanced tones at 10KHz Fully balanced tones from 100kHz to 40MHz -46 22 24 26 28 30 32 34 36 38 Output Power DCL (dbm) Spectrum Run#2 P AVG =30dBm / I DQ nom QAM256 / CS=56MHz / Poly-DPD ACPR # 34dB Fully balanced spectrum each side of 56MHz signal bandwidth Date / Ref doc Enhanced decoupling from khz to MHz ensure balanced spectrum in modulation 12
AB class HPA Spectrum with & w/o DPD Run#1 F=7GHz / Pout AVG =30dBm / I DQ nom QAM256 / CS=56MHz / Poly-DPD Run#2 FRF=7GHz / Pout AVG =30dBm / I DQ nom QAM256 / CS=56MHz / Poly-DPD ACPR withdpd ACPR w/odpd # 5dB ACPR withdpd ACPR w/odpd # 20dB Intrinsic ACPR improvement and balanced correction on Run#2 Run#2 design easily linearizable with DPD Date / Ref doc 13
AB class HPA Constellation with DPD Run#1 F=7GHz / Pout AVG =30dBm / I DQ nom QAM256 / CS=56MHz / Poly-DPD MER = -33dB (with DPD) Run#2 FRF=7GHz / Pout AVG =30dBm / I DQ nom QAM256 / CS=56MHz Poly-DPD MER = -57dB (with DPD) Constellation with DPD Date / Ref doc 14 Constellation with DPD Less distortion, dynamic effects and noise on Run#2 Better Modulation Error Ratio
GaN technology toward 5G 1. Toward 5G with GaN 2. AB class HPA optimization 3. Doherty linearity assessment 4. Prospects Date / Ref doc 15
Doherty Suitable for high PAR Doherty is very suitable for application with PAR providing constant PAE at back-off AB/B class PAE & PDF vs P AVG Doherty PAE & PDF vs P AVG PDF PAE PDF PAE PAE Souhaitée PAE Classe B Pout dbm Pout dbm Pout(t) Pout(t) Date / Ref doc 16
Doherty Linearity is a challenge Doherty behaviour difficult to predict Modulation of the main amplifier drain load by the peak amplifier AM/AM & AM/PM sensitive to bias Compatibility with DPD to be assessed Main (AB/B class) Zc, λ/4 Example of AM/AM & AM/PM vs biasing 90 PAE Peak (C class) PAE Main PAE Doherty PAE Peak P OBO P SAT Pout Date / Ref doc 18
Doherty GH25 Q-MMIC Example Main Features 7-8 GHz Psat > 42dBm Gain > 18dB C/I3 > 20dBc @10dB back-off MER wodpd = -24dB @32dBm MER withdpd = -48dB @32dBm PAE > 24% @32dBm UMS GaN 0.25um / QFN 8x8-20 -24 40 35 7.1 GHz 7.5 GHz MER & ACPR (db) -28-32 -36-40 -44-48 -52 MER w/o DPD MER with DPD ACPR w/o DPD ACPR with DPD # 14dB PAE (%) 30 25 20 15 10 7.9 GHz PAE >24% -56 5-60 28 29 30 31 32 33 34 35 36 Average Output power (dbm) DPD able to linearize Doherty PA Date / Ref doc 18 0 20 22 24 26 28 30 32 34 36 38 40 42 Output power (dbm)
Doherty Spectrum with & w/o DPD Doherty with & w/o DPD F=7.3GHz / Pout AVG =32dBm / I DQ opt QAM256 / CS=28MHz / Poly-DPD Doherty with DPD F=7.3GHz / Pout AVG =32dBm / I DQ opt QAM256 / CS=28MHz / Poly-DPD MER = -48dB (with DPD) ACPR withdpd ACPR w/odpd # 15dB Significant ACPR improvement with DPD. Low dynamic effects. DPD capability demonstrated with Doherty PA Date / Ref doc 19
System Validation Modulation & DPD Synthesis w/o DPD DPD w/o memory DPD with memory Run#1 ACPR=-30dBc MER=-26dB ACPR=35dBc MER=-33dB ACPR=39dBc MER=-34dB Run#2 ACPR=-35dBc MER=-28dB ACPR=57dBc MER=-55dB ACPR=60dBc MER=-57dB Doherty ACPR=-28dBc ACPR=-52dBc ACPR=-56dBc MER=-24dB MER=-48dB MER=-50dB AB class Run#2 significant improvement vs Run#1 Doherty promising results Date / Ref doc 20
GaN technology toward 5G 1. Toward 5G with GaN 2. AB class HPA optimization 3. Doherty linearity assessment 4. Conclusion Date / Ref doc 21
Conclusion Linear HPA can be achieved with GaN Technology Low consumption DPD able to linearize GaN HPA (AB class and also Doherty PA) GaN a good candiate for future of telecom applications (PtP booster / 5G BTS TRx module) Date / Ref doc 22
THANK YOU Date / Ref doc 23