Unlocking Wideband 5G & mmwave Insights to 110 GHz

Transcription

1 Unlocking Wideband 5G & mmwave Insights to 110 GHz Daren McClearnon 5G Solution Marketing Webcast, November 2016

2 Agenda Challenges at Millimeter Wave Evolving landscape: 5G, industry, and regulatory Technology: Components; signals & noise; stability Requirements: flexibility, performance Approaches Achieving wideband, low-evm at 28GHz; a co-existence study Adapting to 5 GHz bandwidths and charting new territory to 110GHz Page 2

3 Spectrum Activity: New Air Interface Approaches for All Bands 3GHz 10GHz 40GHz 90GHz <6GHz Updated CP-OFDM (like LTE). With Flexible numerology, TTI, and Frame Orthogonal waveform, but Non-orthogonal MA scheme FDD and (flexible) TDD Unconstrained by LTE compatibility new radio new spectrum Accommodate higher-order MIMO >6GHz <40GHz Compatible with <6GHz Mostly (flexible) TDD OFDM-based with scaled numerology from <6GHz Emphasis on universal frame structure for all bands (same frame structure as new <6GHz multiple-access numerology) >40GHz May require single-carrier waveform TDD or PTP/PTMP only (71-86GHz exception?) Most likely used for fixedwireless applications (not embb) Duplex FDD TDD Flexible Duplex Full Duplex Multiple Access Scheme OFDMA SCMA NOMA MUSA Waveform Type Single-carrier Multi-carrier: CP-OFDM FBMC UFMC/UF- OFDM GFDM W-OFDM Modulation Format OQPSK QAM New constellation mapping Orthogonal Frequency Division Multiplexing(OFDM) Filter Bank Multicarrier(FBMC) Universal Filtered Multicarrier(UFMC) Universal filtered OFDM (UF-OFDM) Windowed OFDM (W-OFDM) General Frequency Division Multiplexing (GFDM) Orthogonal Frequency Division Multiple Access (OFDMA) Non-orthogonal Multiple Access (NOMA) Sparse Code Multiple Access (SCMA) Multi-User Shared Access (MUSA) Page 3

4 Millimeter-wave 5G Frequency bands Frequency Ofcom FCC (Jun 2016) ITU (Oct 2015) For study 28GHz (25-27GHz) BW 425MHzX GHz 37GHz BW 400MHzX GHz 39GHz BW 200MHzX GHz FFS GHz GHz GHz GHz GHz 57-66GHz GHz 64-71GHz 66-71GHz (extend ISM unlicensed) 66-76GHz 71-76GHz 81-86GHz Most Likely Uses of Spectrum as of Sept 2016 Significant investment in EMBB Mobile, Multiple Access <40GHz due to cost and simplicity. 28GHz: Korea, Japan, and USA GHz USA and perhaps more likely for Europe 24-27GHz : Europe 45GHz: Focus for aj in China 57-86GHz Bands more likely for high-speed point-topoint and extensions of ISM-based WiFi Examples of Public Activity (Updated Summer 2016) FCC Announced rules on mmwave proposals 14 July 2016 Ericsson will provide 28GHz system for SKT (Korea) and 15GHz system for CMCC Intel banking on 28GHz in US and implementing accordingly AT&T, Verizon, T-Mobile filed for experimental licenses (3.5, 3.7, 15, 28, 37, 39GHz) Most large players demonstrating high-rate capabilities from 15-90GHz Page 4

5 Summary of Regulatory Landscape 28 GHz centimeter-wave band and also GHz What performance (eg - EVM) can be expected for pre-5g signals? What compliance/interoperability should be considered vs. existing satellite/lmds bands? FCC Ruling in July 2016* opened up to 14 GHz of spectrum (57-71GHz) How will the worldwide market adapt to this new allocation, and apply it? How can existing testbeds address such large bandwidths, in new bands? What performance and interoperability is needed? Spectral Emissions policies are being considered above 86 GHz (eg to 110 GHz in Japan) What sensitivity and performance is required to perform these tests? How does one debug broadband compliance issues using individual banded solutions? Page 5

6 Why Go To mmwave? Challenges of bringing electronic products to market 5G Industry 5G Beamforming Massive Growth in Mobile Data Demand Massive Growth in Number of Connected Devices Resolve Subscribers in Dense Environments Compact, Mobile Platforms Exploding Diversity of Wireless Applications Decrease Interference Page

7 Summary: Numerous Design & Measurement Challenges at mmwave Small Dimensions and Complex Test Setups Challenging Ultra Wideband mmwave Measurements Small Signal Strength & Wideband Integrated Noise Page 7

8 Industry Challenge New Spectrum Wide bandwidths for high data throughput New issues 5G Industry Drivers Move to mmwave for wide bandwidths Massive Growth in Mobile Data Demand 100 GHz emerging 5G research Up to 5 GHz BW for 5G signals GHz Design and measurement challenges Challenging ultra wideband mmwave measurements Phase noise, IQ & freq response errors worse at mmw Wide bandwidth means more noise, more spurs, worse EVM Page 8

9 Industry Challenge Small cells provide great service in a crowd Increased emissions compliance 5G Industry Drivers Massive Growth in Mobile Data Demand mmwave signals attenuate quickly with distance mmwave small cells enable dense deployment Design and measurement challenges Emerging GHz small cell pointto-point backhaul Emissions measurement in GHz passive band Small signal strength Very small signals in more noise as frequency/bandwidth increases Can t use preamplifier for measurements like spectrum emission mask (SEM) Page 9

10 Industry Challenge Smaller wavelengths, smaller components new issues in calibration, stability High frequency leads to smaller components 5G Industry Drivers Massive Growth in Number of Connected Devices Compact, Mobile Platforms Design and measurement challenges Small dimensions & complex test setups Smaller, fragile cables, adaptors; calibration, stability Ultra wideband frequency spans, more spurs and noise More difficult to maintain the same EVM at mmw Page 10

11 Agenda Challenges at Millimeter Wave Evolving landscape: 5G, industry, and regulatory Technology: Components; signals & noise; stability Requirements: flexibility, performance Approaches Achieving wideband, low-evm at 28GHz; a co-existence study Adapting to 5 GHz bandwidths and charting new territory to 110GHz Page 11

12 Typical Testbed to Evaluate 28 GHz and 39 GHz bands DSOV334A 33 GHz Oscilloscope M8190A AWG with SystemVue Installed on Embedded Controller E8267D PSG N9040B 50 GHz UXA (1 GHz BW) Page 12

13 N-bit ADC/DAC range Wideband EVM Considerations Easier to remove IQ Gain Imbalance IQ Skew Flatness vs. frequency Slow drift vs. time/temperature Harder to remove Effective AWG/Digitizer number of bits, vs. high crest-factor signals Higher ktb noise floor from wider BW ~LOG 10 (BW). Losses & noise are inescapable Phase noise Gain/Phase changes vs. amplitude Realistic expectations The EVM may be S/N-limited, even before you even stimulate the DUT Greater care to maintain the signal level in the sweet spot of AWG/Digitizers Page 13

14 Question: where does basic Equalization make the system flat? Laptop PC LAN M8190A AWG Direct Out CH1 1? Ref Clk In Direct Out CH2 10 MHz Out 2? 3? E8267D with wideband IQ inputs 28 GHz 0 dbm Lossy 28GHz cable DUT Ref In Lossy, 28GHz cable 50 GHz UXA 4? Differential I/Q Signals Page 14

15 Question: where does basic Equalization make the system flat? Laptop PC Answer: 4 At the analyzer. So there is one more potential improvement. LAN Ref Clk In 10 MHz Out Ref In 4 M8190A AWG Direct Out CH1 Direct Out CH2 E8267D with wideband IQ inputs 28 GHz 0 dbm Lossy 28GHz cable DUT Lossy, 28GHz cable 50 GHz UXA Differential I/Q Signals Page 15

16 Calibrating your system for best EVM at 28GHz Build confidence into your 5G system 5G Signal Generation 5G Signal Analysis Wideband Calibration Page 16

17 Flexible Waveform Generation: Software + AWG+ PSG Software Signal Optimizer Page 17

18 Keysight Signal Optimizer to calibrate 28 GHz Test Configuration Laptop PC with K3101A Signal Optimizer 1. Apply Comb reference to DUT output plane Flatten your RX cable, downconverter and analyzer LAN Ref Clk In 10 MHz Out Calibrated reference Ref In M8190A AWG Direct Out CH1 Direct Out CH2 E8267D with wideband IQ inputs U9391 comb generator 50 GHz UXA Differential I/Q Signals Page 18

19 Keysight Signal Optimizer to calibrate 28 GHz Test Configuration Laptop PC with K3101A Signal Optimizer 2. Connect Source (direct thru without DUT) to Analyzer Flatten your TX source + upconverter + cable. LAN Ref Clk In 10 MHz Out Ref In M8190A AWG Direct Out CH1 Direct Out CH2 Differential I/Q Signals E8267D with wideband IQ inputs 28 GHz 0 dbm Modulated RF/ uwave out THRU RX calibration plane 50 GHz UXA Page 19

20 Keysight Signal Optimizer to calibrate 28 GHz Test Configuration Laptop PC with K3101A Signal Optimizer 3. Now your signal AT THE DUT input is clean, with low EVM. Connect your DUT, and measure it! LAN Ref Clk In 10 MHz Out Ref In M8190A AWG Direct Out CH1 Direct Out CH2 Differential I/Q Signals E8267D with wideband IQ inputs 28 GHz 0 dbm Modulated RF/ uwave out TX calibration plane DUT RX calibration plane 50 GHz UXA Page 20

21 Signal Optimizer Results M8190A AWG + E8267D PSG, and 50 GHz UXA with 1 GHz option (H1G) Signal type 16QAM, 1 GHz BW 28 GHz, -10 dbm single carrier modulation Calibrated result 0.76% RMS EVM Page 21

22 Signal Optimizer Results M8190A AWG + E8267D PSG, and 50 GHz UXA with 1 GHz option (H1G) Signal type Custom OFDM 700 MHz BW 28 GHz, -10 dbm Calibrated result 0.93% RMS EVM Page 22

23 Comparison: 28 GHz with 700 MHz Using VSA Equalization only Signal type (5) 100MHz LTE carriers 700MHz BW 28GHz Tilt down vs. freq VSA EQ-only result 1.21% RMS EVM Page 23

24 Comparison: 28 GHz with 700 MHz Using Signal Optimizer calibration, then VSA Equalization to track drift Signal type (5) 100MHz LTE carriers 700MHz BW 28GHz Flatter vs. freq Calibrated & Equalized result 0.92% RMS EVM Page 24

25 Why is clean EVM at the DUT important? Additional considerations Why Calibrate A Low-EVM test system allows greater resolution for low added- EVM devices. (How low is enough?) Not all errors are corrected by equalizers. (some static IQ modulator impairments, vs. frequency) Absolute signal quality to the DUT is critical for nonlinear devices (eg - DPD) Cable loss is not flat across the bandwidth at 28GHz: simple USB power meter not enough Other notes: E8267D PSG updated firmware to optimize the power level/agc of the source for wideband signals, for better EVM Calibrations can still drift at millimeter-wave frequencies with temperature variations, mechanical disturbance, and change of carrier frequency and/or bandwidth Page 25

26 Wideband Satellite Waveform 5G & Satellite 28 GHz Evaluate Potential Interference Scenarios Wideband Custom OFDM Waveform Wideband APSK Satellite Waveform 5G Candidate Waveform Source: Page 26

27 5G & Satellite 28 GHz Scenario 1- Good Coexistence Between Satellite and Candidate 5G Custom OFDM Satellite Custom OFDM Demodulation Page 27

28 5G & Satellite 28 GHz Scenario 1- Poor Coexistence Between Satellite and Candidate 5G Custom OFDM Satellite Custom OFDM Demodulation Page 28

29 5G & Satellite 28 GHz Scenario 1- Poor Coexistence Between Satellite and Candidate 5G Satellite signal is interfering with candidate 5G signal subcarriers EVM vs. Subcarrier Page 29

30 Agenda Challenges at Millimeter Wave Evolving landscape: 5G, industry, and regulatory Technology: Components; signals & noise; stability Requirements: flexibility, performance Approaches Achieving wideband, low-evm at 28GHz; a co-existence study Adapting to 5 GHz bandwidths and charting new territory to 110GHz Page 30

31 Case Study 2 Charting New Territory to 110 GHz A 5GHz wide E-band PA at GHz, and beyond 1mm SA Port 1mm cable 6x HARMONIC UPCONVERTER WAVEGUIDE AMPLIFIER WG-1mm ADAPTER Page 31

32 Question about 1mm connectors: Which is Female vs. Male? Page 32

33 Question about 1mm connectors: Which is Female vs. Male? It s difficult to measure performance that you can t even see. Page 33

34 110 GHz configuration used for this case study Wideband E-band signal source, with new 110 GHz signal analyzer N9041B 110 GHz UXA signal analyzer M8195A 65 GSa/s AWG FOR DEMOD wideband 5GHz IF Output N5183B MXG Signal generator as clean LO Ch1 Data Out Direct-to-IF fc=5 GHz 0 dbm f=14.75ghz -4 dbm (88.50GHz high-side LO) IF Input LO In E-band upconverter WG DUT 1mm fc=83.5 GHz -20 dbm DSOS840A S-series 8 GHz Infiniium oscilloscope Page 34

35 Advantages of wideband baseband platforms Capable beyond 5 GHz bandwidths, for whole-band and multi-carrier signals M8195A AWG for Direct-to-IF signals Typical signal IF=5 GHz Up to 20 GHz of analog bandwidth No IQ errors, very flat Best noise & SFDR into mm-head N9029ACST-U12 Harmonic upconverter for the 60-90GHz band x2 instead of x6, lower noise, spurs Very wideband Compact size DSOS840A with VSA software for Direct-from-IF analysis Digitizes low-ghz IF directly Performs clean DDC to baseband Very flat for the wide bandwidth Page 35

36 Unfiltered DUT output, GHz as a test case Question: Which spectral responses are real? Page 36

37 Unfiltered DUT output, GHz as a test case Typically a GHz BPF filter would be used IMAGE (N-1) Lower DESIRED SIGNAL (N) Lower Sideband 2GHz 83.5GHz IMAGE (N) Upper Page 37

38 Spectral Emissions Masks Worldwide policy activity in V, E, and W bands such as ETSI EN ETSI EN V2.2.1 ( ) (pg 26), regarding 71-76GHz and 81-86GHz bands Page 38

39 Spectral Emissions Masks New policy activity in V, E, and W bands New mm-wave bands are opening up demand for new policies > 86 GHz Already discussion of GHz in each region SEM requires lower noise floor and SFDR for compliance and R&D debug Do you have the sensitivity for Spectral Emissions? Page 39

40 Displayed Avg Noise Level (DANL), 3 Hz 110 GHz Full span, ResBW=1MHz Ultra low noise architecture enables new applications dbm (@1 Hz BW) 105 GHz Page 40

41 Displayed Avg Noise Level (DANL), 105GHz 1 MHz span, 1kHz ResBW Page 41

42 Now let s re-consider our E-band TX source and amplifier DUT Sensitivity and spur-free dynamic range are critical Page 42

43 QAM16 Modulation performance at 83.5 GHz, with PA 500 MSym/s, 650 MHz BW 1.6 GSym/s, 2 GHz BW Page 43

44 QAM64 Higher-order Modulation performance at 83.5 GHz 1.6 GSym/s, 2 GHz BW 4.0 GSym/s, 4.8 GHz BW Page 44

45 Disable baseband source AWG (turn off modulation) (n)*14.75ghz feedthrough harmonics from the Mixer LO Image PRIMARY 88.5GHz Image Page 45

46 Looking for additional spurs, subharmonics, oscillations Increase the sensitivity of the system Any more hidden spurs? Subharmonics? Turn on AVERAGING Reduce RES BW Reduce ATTENUATION, if possible Page 46

47 Looking for other EMI and other compliance issues Turn off RF output on the GHz Mixer LO still some noise Keep in mind: still sweeping from 60 GHz to 110 GHz Page 47

48 Enable Noise Marker accounts for ResBW, then renormalizes to 1Hz Modulation = OFF LO is OFF. There is no signal. Where is the extra residual noise coming from? Page 48

49 Mixer and PA Bias OFF. Noise floor reduces by another 6dB Question: is it from the Mixer collecting noise? or the PA Noise Figure? In-band NF reduces the system dynamic range Solutions? Filtering? Loading? Image rejection? Packaging? Page 49

50 Conclusions about mm-wave Testbeds Flexible testbeds are required to adapt to rapidly evolving needs New 5G waveform generation & analysis Emerging frequency bands and spectral policies for each region, also considering co-existence Wider modulation bandwidths and better flatness Higher sensitivity with lower noise, spurs The N9041B UXA dramatically improves measurements up to 110 GHz and 5GHz bandwidths, to help you revolutionize millimeter-wave Page 50

51 Thank You!!! Questions and Answers Find more information: Page 51