mm-wave Transceiver Challenges for the 5G and 60GHz Standards Prof. Emanuel Cohen Technion

Transcription

1 mm-wave Transceiver Challenges for the 5G and 60GHz Standards Prof. Emanuel Cohen Technion November 11, 11,

2 mm-wave advantage Why is mm-wave interesting now? Available Spectrum 7 GHz of virtually untouched spectrum from GHz, 10 GHz of BW in 70 and 80 GHz, 5 GHz BW in 90 GHz, New product opportunity Very high-rate, several Gbps, uncompressed data transmissions not possible with other wireless standards Recent advances in SiGe and CMOS High directivity Spatial reuse 2

3 mm-wave Applications Imaging Communication (60G, 5G) Radar Automotive 77-81GHz (and 24GHz) Backhaul LOS CAN2_AP1 CAN3_AP3 CAN3_AP2 CAN3_NAP CAN3_AP1 CAN1_NAP CAN1_AP1 CAN1_AP2 3

4 60G applications - Intro to WiGig Main Use Cases Main use case requiring phased array system with symmetrical configuration 4 4

5 5G mm-wave Vision 5

6 The mm wave Challenge Free space losses and high data require 30dB more link budget compared to 5 GHz systems directive antenna TX power that is dropping as we go higher in frequency and to more 2 advanced Si process P V / f antenna array max Need to cope with blocking and moving targets - scanning capabilities LOS 10m, 60GHz: 88dB WiFi 60GHz Diff Tx power ~17dBm 10dBm -7dBm Freq 5-6 GHz 60GHz -20dB BW ~16MHz 1.7GHz -20dB Constellation 64 QAM 16 QAM ~+6dB Distance ~30m ~10m (1/3) -10dB Antenna gain 0dBi ~15dBi Rx & Tx +30dB TX RX array _ gain array _ gain 20 log10 log ( N) ( N) 30dB gain TX+RX >30 elements 6

7 Glass Up, Glass Down, Lid open cases Antenna Challenges 7

8 5G direction Modern LTE Femto Cell vs. mmwave Small Cell HPBW: HPBW: 2-3 Average data-rate 50 Mbps/cell MU Average data-rate 50 Gbps MU 8

9 Array Architecture Massive MIMO with 2D array 9

10 Future increase in the PS requirements Relative Amplitude [db] deg 45deg ideal Angle scan [Deg] PS loss from 2 bit to 4 bit 0.8dB SLL -9dB for 2 bits PS and -14dB for 4 bits PtN -10dB for 2bit PS and -23dB for 4 bit Beam nulling is very sensitive to phase and amplitude errors. 10

11 Silicon Challenges CMOS Technology Trends PA Efficiency Silicon PA Performance 60 GHz chipset solutions have been developed 11

12 Passives and Models used Build models for inductors and transformers Cap depend on L Cap depend on L,k,n Core layout for Best Fmax B. Martineau, et al., ESSCIRC 2007 B. Heydari, et al., ISSCC 2007 Transistor model with extraction 12 12

13 Diff Gain Boosting and Wide-Bandwidth 1nH SE Pros: Achieve CC boosting performance with SE design Reduce current consumption for LNA improve NF Use feedback and Transformer(TF) match (BW) Cons: Feedback sensitivity to parasitic and variation SE TF suffer from high common mode capacitance 13

14 Built in Self Test (BIST) for Power Gain and Phase Remove the expensive external setup Over the air testing is complex and potentially less accurate V in_ ch1 V out _ V in_ ch 2 BIST measure Pout,LNA gain and TX phase difference BIST accuracy in phase and power over PVT is 5deg and +/-1dB 14

15 low cost mmw packaging LTCC Assembly Routing Antenna Routing phase simulation Routing IL simulation Equal delay routing with quadrant symmetry Route loss ~2-2.5dB and phase match ~10deg 15

16 Full assemblies LTCC aperture coupling patch Mounted on FR4 RFIC RFIC End-fire on BT LTCC BOTTOM LTCC phased array module LTCC TOP slot-loop antenna on HDI PCB 16 16

17 Architecture options (TX) 17

18 Digital Beamformer MIMO implementation Reduce die size, speed up design time, flexibility & power PS will have excellent resolution MIMO is done at BB before the RF chain 18

19 RF DAC IQ combining challenge Over the air combining loses 3dB on the antenna gain 19

20 LO distribution challenge BB distribution will allow maximum flexibility in array structure and reduction in design cycle BB distribution will shrink the die size and also reduce distribution loss 20

21 RX dynamic range ADC + mixer may need to cope with increased dynamic range Average Side Lobe Level define the spatial filter limit Relative Amplitude [db] deg 45deg ideal Angle scan [Deg] 21

22 Synchronization between channels Mean power degradation SNR due to amp error SNR due to phase error Initial phase shift is predefined or calibrated The amplitude noise may be higher w/o track Up to 20deg std random error between chain can be tolerated (around 1psec) limit of the PN of each chain PLL 22

23 Large array & FDD challenge Create a phased array with shared TX/RX antenna space 23

24 TX-RX Cancelation options Transmit power 20dBm Receive power -90dBm Total cancelation 110dB! Can cancel only partially in the RF FEM and rest at IF, BB Complexity of the FEM canceling? 24

25 Thank You 25