Overview About RF And PA Requirements For 5G NR Franz Dielacher

WFB-1Overview about RF and PA Requirementsfor 5G NR and Challenges for HardwareImplementationFranz Dielacher, Yannis Papananos,Peter Singerl, Marc Tiebout,Daniele Dal Maistro, Christos ThomosInfineon Technologies Austria AGWFB-1

OUTLINE Introduction Trends in communications and key characteristics RF Frontend Architectures for 5G Massive MIMO Technology Full Digital Beamformer/mMIMO for sub 6GHz Hybrid beamforming for mm-Wave RF Technology considerations Chip design for 5G technology demonstration (some examples) Transmitter line-up and radio architecture PA technology trade-offs Case studies– Case 1: Doherty PA in GaN technology for sub 6GHz– Case 2: mm-Wave RF beamforming example– Case 3: Efficient 30 GHz Doherty PA design in SiGe SummaryWFB-12

7 Billion Devices 2014 500 Billion Devices 2022Courtesy Prof. Fitzek, TUDresdenWFB-133

In the headlines Verizon hits 1.45Gbps 4G LTE speeds in New York (With a little help from Nokia and Qualcomm) 5G mm-wave base station shipments: increased plans in the USA by AT&T and Verizon to pursuemobile 5G in the 24-39 GHz bands, not just fixed wireless. The China 5G ramp at 2.6 through 3.5 GHz was adjusted to account for new expectations from MIITin China: Each of the three operators in China are expected to deploy 500,000 base stations withintwo years of receiving the 5G spectrum license.Source:dilbert.comWFB-14

ITU IMT-2020 5G Vision and Research ChallengesPeak Data Rate[Gbps][20]User ExperienceData Rate[Mbps][100-1000]Area Connection Density[#/km²]WFB-1Data rates exceeding 10 GBpsLatency[ms]Network latency under1 millisecondCapacity expansion by a factorof 1,000Energy efficiency gains by afactor of 1,000 per transportedbitSource: ITU and Ericsson Mobility Report55

Business impacts to semiconductor industry Wireless major driver for semiconductor industry 5G expected to be the next major driver of wireless semiconductor market from2020 onwards Challenges– Massive MIMO sells much more signal paths – what happens to cost– Cost per antenna– Can’t be 100x more expensive– Moore’s law – will it come to end and what is the impact to 5G Cost per unit very critical parameter for business successWFB-166

Identification of Key 5G Characteristicsand relevance for RF Architectures Massive MIMO Efficient Small Data Transmission RAN Transmission Centimeter andMillimeter Waves Wireless Backhaul / Access Integration New Waveforms Shared Spectrum Access Advanced Inter-Node Coordination Simultaneous Transmission Reception Flexible Networks Flexible Mobility Context Aware Networking Information Centric Networking Moving Networks Multi-RAT Integration & Management D2D CommunicationsWFB-17

New Spectrum Scenarios Complementary use of alternative spectrum– Unlicensed spectrum, secondary spectrum usage, spectrum sharing, „LAA“ Usage of very high frequency bands (for 5G NR Phase 2)– Lots of spectrum available Extreme capacity and data rates– Small wave length Possibility for large array antenna solutionsFuture spectrum rangeCurrent spectrum range300 MHz3 GHz30 GHz300 GHzcellular bandsSource: Sven Mattisson, ISSCC-2016WFB-18

3GPP Time Line: LTE Adv. through 5G-NRLTE Advanced20MHz UEmax Ch. BWLTE Advanced Pro20MHz UEmax Ch. BW5G NR100MHz UEmax Ch. BW400MHz UEmax Ch. BW WFB-1SKYWORKS 2017 White paper: ‘5G in Perspective’9

OUTLINE Introduction Trends in communications and key characteristics RF Frontend Architectures for 5G Massive MIMO Technology Full Digital Beamformer/mMIMO for sub 6GHz Hybrid beamforming for mm-Wave RF Technology considerations Chip design for 5G technology demonstration (some examples) Transmitter line-up and radio architecture PA technology trade-offs Case studies– Case 1: Doherty PA in GaN technology for sub 6GHz– Case 2: mm-Wave RF beamforming example– Case 3: Efficient 30 GHz Doherty PA design in SiGe SummaryWFB-110

Array Antenna Systems / Massive MIMOMassive MIMO is multi-userMIMOImpact is very demanding:- 10 times increased capacity- 100 times reduced radiatedpower- Overall: improvement inradiated energy efficiency(bits/J) 1000 times, onthe uplink & the downlink.Source: Ove Edfors, ISSCC-2019, Forum-1WFB-111

Beamforming/mMIMO optionsDigital BeamformingHybrid BeamformingX2 DACs/ADCss1 DACs/ADCsX 1BeamformerUE1s1UE1XN/MsMUE2UEMsΜUEΜXΝ DACs/ADCsX1s2X 2X ΝsM DACs/ADCsBeamformersΜDPDs2Linear pre-coding/detections1X1 DACs/ADCsMixers1MixerPAs / LNAsX1Channel /User terminalsXN/MN MDigital beamforming/mMIMOis used at sub 6GHz frequency bands todayWFB-1Hybrid beamformingis used at mm-Wave frequency bands today12

Pros/cons investigation for beamforming optionsSimple BFDigital BFHybrid BFComplex Hybrid BFPower Eff. Area Eff. SingleMultipleMultipleMultiple Complexity Spectral Eff. Nr. of StreamsFlexibilityWFB-113

Full Digital Beamformer/mMIMO for sub 6GHzRF-Frontend for MassiveMIMO systemNear linear increase of area and power consumption with array size Traditional analog/RF approach becomes inefficient Efficient System in Package solution for: PA LNA Switch Control Alternatives like all-digital transceiver to be considered like: PWM-based digital RF, RF-DAC, .WFB-11414

Analog/RF Requirements (some considerations)Establishing translations between circuit and radiated performance is challenging For bands below 6 GHz the requirements arein the same ball-park of existing systems The Tx output power per antenna elementdepends on EIRP and array size On the receiver side due to required coverage andcell-edge bitrates, the performance requirementslike noise-figure are increasing. IBW requirement is increasing Beamforming and AAS/massive MIMOimplies new challenges due to e.g. antennacross talk etc.WFB-1Source: Sven Mattisson, ISSCC-2016, Forum 315

Hybrid beamforming for mm-WaveSimple Hybrid beam-forming Beam-steering is sub-optimal Analog combiners are an issueComplex Hybrid beam-forming Optimum performance – Equivalent todigital solution Analog combiners are an issueWFB-11616

Design criteria for beamforming chip WFB-1Number of channelsBandwidthPhase Shifter versus True Time DelayRx performance like NF, phase/time resolution, Gain, Linearity,power consumption, .Tx performance like of output power (P1dB, back-off,.), phase/timeresolution, gain, evm, linearity, power consumption, .Phase shift or true time delay immune to temperature variationPhase invariant programmable gainIntegrated test and calibration capabilities like LO-generation, signalinjection and detection, .1717

Absolute phaseAbsolute phasePhase Shifter versus True Time Delayαfrequencyfrequencydφ1φ(x,y) f(α,ω,x,y)WFB-1n.dφiφnTRX1818

OUTLINE Introduction Trends in communications and key characteristics RF Frontend Architectures for 5G Massive MIMO Technology Full Digital Beamformer/mMIMO for sub 6GHz Hybrid beamforming for mm-Wave RF Technology considerations Chip design for 5G technology demonstration (some examples) Transmitter line-up and radio architecture PA technology trade-offs Case studies– Case 1: Doherty PA in GaN technology for sub 6GHz– Case 2: mm-Wave RF beamforming example– Case 3: Efficient 30 GHz Doherty PA design in SiGe SummaryWFB-119

RF Transistor and RF-IC Technology ChartLDMOS TransistorsLDMOS ICsGaN-SiC & GaN-Si Transistorsand MMICSmart PowerTechnologySi-Ge/BiCMOScapable of mmW TRx@ 1W Tx &incl. RF-beamformingGaN (GaN-SiC, GaN-Si) 2GHz to 100GHzPower30V LDMOS 2.7GHz ?InP 6 to 100GHzGaAs up to 100GHzSPT for Bias&ControlRF CMOS 6GHz1GHzWFB-12GHz5GHzRF CMOS limitation onPerformance for mmWtransceiverto be studiedSi-Ge/BiCMOS up to 100GHzRF CMOS up to 100GHz10GHz30GHz60GHz100GHz2020

Packaging Technology CharteWLB (embedded Wafer LevelBallgrid-Array) PackagingUnder investigation:Packaging by ‘Chip-Embedding’eWLB withIntegrated AntennaCeramic PackageWFB-1Cu FlangePCB basedPackageFull RF-PowerModuleUnder investigation:Multilayer RF-laminate, plastic overmold2121

OUTLINE Introduction Trends in communications and key characteristics RF Frontend Architectures for 5G Massive MIMO Technology Full Digital Beamformer/mMIMO for sub 6GHz Hybrid beamforming for mm-Wave RF Technology considerations Chip design for 5G technology demonstration (some examples) Transmitter line-up and radio architecture PA technology trade-offs Case studies– Case 1: Doherty PA in GaN technology for sub 6GHz– Case 2: mm-Wave RF beamforming example– Case 3: Efficient 30 GHz Doherty PA design in SiGe SummaryWFB-122

Traditional Transmitter line-upDSP (Baseband)LPFDACDC power 5WIQ mixer 10W 80WIPGALO0oBPFDriverPABPF90 oLPFDACQPA eff. is the major bottleneckTX power 0.1WWFB-1 2W 40W ( 45%)23

Emerging RF architecturesDigitally assisted RF RF-sampled AD/DA converters andDoherty PA- GHz-range, high resolution ADCs and DACs Envelope Tracking RF path is broadband and reconfigurable SMPA– Requires wideband and efficient supplymodulator Digital (PWM and outphasing) transmitter SMPA with high-efficiency e.g. Class-E PA Reconfigurability– High resolution DTC for linearity (ACPR),– High bandwidth (IBW) requirementWFB-12424

GaAs, GaN, SiGe, or CMOS for the PA Choice of semiconductor process for the PA– Tricky balance of output power, linearity, and efficiency.– At 2,5-6 GHz, the PA process technology is likely to be GaN,– Above 20 GHz, the choice is more complex. mm-Wave ( 24 GHz) PA’s– Class-A, Class-AB in GaAs (used in many trials)– Huge heat load of several hundred Watt or higher.– Significant improvement needed for volume deployment Fineline GaN technology Doherty PA with RF predistortion Relaxed ACLR specificationWFB-12525

Comparison of CMOS, SiGe, GaAs and GaN formmWave PAsSource: GA TechWFB-126

A few assumptions for mm-Wave PA’s Peak-to-Average Ratio of the waveform of 10-12 dB. ACLR requirements likely to be set to about -30 dBc.– Ericsson’s input to the 3GPP RAN4 committee, ACLR requirements tighter than -35 dBc yield little benefit (and are probablynot achievable in practical systems). Multiple deployment scenarios - different transmit power levels– Urban Deployment– Dense Urban Deployment– Indoor Small Cells Output power per PA depends on EIRP and number of antenna elementsWFB-12727

Power Amplifier Technology Selection vs. Array SizeBackoff in the calculation below is 10dBWFB-128

OUTLINE Introduction Trends in communications and key characteristics RF Frontend Architectures for 5G Massive MIMO Technology Full Digital Beamformer/mMIMO for sub 6GHz Hybrid beamforming for mm-Wave RF Technology considerations Chip design for 5G technology demonstration (some examples) Transmitter line-up and radio architecture PA technology trade-offs Case studies– Case 1: Doherty PA in GaN technology for sub 6GHz– Case 2: mm-Wave RF beamforming example– Case 3: Efficient 30 GHz Doherty PA design in SiGe SummaryWFB-129

Case 1: Doherty PA for sub 6GHzMain (Carrier)Offset lineIc MInputZM Peak (Auxil.)4 4Z02Ip p 4Z01ZPZLR0 Multi-carrier cellular signalshave high PAPR( 8dB after CFR) PA architecture– Doherty is best in efficiency atdeep back-off– Linearization is a must in orderto meet both power andemissions requirements:predistortion is a must Discrete or MMIC integrationWFB-13030

4.5GHz Doherty MMIC research in GaN-MMICMMICArea:2,5x2,5mmMMICWFB-1Design targets:› Frequency: 4,5GHz› IBW: 400 MHz› PAE: 40%› Pout: 33dBm› PAPR: 9dB› Gain: 30dB› Supply: 28V3131

Case 2: mm-Wave RF beamforming example For current mm-wave amplifiers and beamformers:– PAE 4% (@10dB back-off)– Beamforming cost 250 mW/channel “Optimum” array configuration:– 200 elements (14 x 14)– 10 dBm average power per element @ 4% PAE 50 W– 250 mW/element beamforming 50 W– Total per antenna array 100 W just for the RF front-end WFB-13232

Beamer 28 RFIC Block Diagram Variable TTD supports widebandsignals Center frequency 28 32 GHz,instantaneous Bandwidth 800MHz Quadruple bidirectional channelsincluding Wilkinson splitter/combiner True time delay range 180ps and1ps resolution Integrated BITE for Test andcalibration (RF signal generation andmonitoring)WFB-133