5G Test and Mesuarment Rohde Schwarz Jan 2020 v4

eBook
End-To-End 5G Test
Solutions
January 2020
S P O N S O R E D B Y
Table of Contents
3

Introduction
Pat Hindle
Microwave Journal, Editor
4

Pillars of 5G: Spectral & Energy Efficiency
Dr. Corbett Rowell
Rohde & Schwarz
11

NR and LTE Coexistence Through Dynamic
Spectrum Sharing
Andreas Roessler
Rohde & Schwarz, Munich, Germany

17

Ericsson Performs 5G Coverage/Performance Verification
Using R&S Drone-Powered Solution
Rohde & Schwarz
18

Overcoming the Challenges of mmWave, On-Wafer
Load-Pull Measurements for 5G
Richard Hilton and Steve Dudkiewicz
Maury Microwave Corp., Ontario, Calif.
22

Testing 5G mmWave Devices Over the Air
Rohde & Schwarz
23 Optimizing the Perennial Doherty Power Amplifier

Gareth Lloyd
30

One Box Test Solutions for 5G
Rohde & Schwarz
Munich, Germany
32

Software and Hardware Near-Field Transformations
for 5G OTA Testing
Benoît Derat, Corbett Rowell, Adam Tankielun and Sebastian Schmitz
2
Introduction
End-to-End 5G Test Solutions
More than 50 commercial 5G networks were deployed in 2019. As 5G rollouts accelerate in 2020,
manufacturers will need to pay close attention to testing and verification to streamline their processes
and still deliver high quality products. This eBook covers almost everything you need to know about
testing and verification for 5G from devices to systems from an RF perspective. Rohde & Schwarz
authored most of these informational articles with Microwave Journal plus an additional article from
Maury on load pull.
The first couple of articles discuss important features of 5G such as spectral and energy efficiency
along with Dynamic Spectrum Sharing. Then specific test solutions are discussed such as drone testing
5G systems and on wafer load pull measurements at mmWave frequencies. The next couple of articles
cover device testing and digital optimization for Doherty power amplifiers. Finally, the last two articles
address 5G test systems including a one box solution and hardware/software for 5G OTA testing.
We thank Rohde & Schwarz for publishing all of these educational articles and sponsoring this
eBook so that you can download it for free. As 5G gains momentum next year, device and system
manufacturers will need to be well versed in testing and verification techniques for 5G, so we hope this
eBook will be helpful in this area.
Pat Hindle, Microwave Journal Editor
3
Pillars of 5G: Spectral & Energy
Efficiency
Corbett Rowell
5G deployment is driven by two key factors that are typically in conflict with each other: system
capacity (spectral efficiency) and system cost (energy efficiency). Spectral efficiency describes
how much capacity can be provided and is typically measured in bps (bits per second) per Herz
of frequency, whereas energy efficiency characterizes how much it costs to run the network for a
given capacity.
I
n the past generations of mobile technology, the while simultaneously decreasing the costs of running
cost of the provision of higher capacity was almost the network.
directly proportional, as this involved building more
base stations or increasing the spectral bandwidth (BW) HOW MUCH DOES IT COST TO RUN A CELLULAR
inside the network. While this was roughly sustainable NETWORK?
in the past, the demands for 10 to 100× the capacity of Although the base stations have become more ener-
4G networks make this approach a path to bankruptcy, gy efficient from 2G to 4G, the costs of providing more
as it is unlikely the consumers of the increased capacity capacity through network densification have increased
are willing to increase their expenditure by 10 to 100×. significantly (see Figure 2). Most of the cost of setting
As illustrated in Figure 1, in order to advance the mo- up and operating a cellular network is in the remote
bile networks, the industry needs to solve the problem provision of air conditioning and site rental for the base
of how to increase capacity across the entire network stations.1-2 In terms of the initial CAPEX, the air condi-
tioning is over 50 percent, leaving the
Increased Capacity, Increased Expenses Optimal Network remaining for the base station equip-
BS Locations ment. Similarly, in terms of recurring
BS Locations
OPEX, electricity is almost 50 percent
Low Data Rates
on Edges of the costs. Most of the electricity is
for the operation of the remotely dis-
Cell Edge tributed air conditioning network used
Uniform Experience
Regardless of Location to cool the baseband processors (ra-
dio units are typically air-cooled and
Traffic Traffic
? do not require additional air condi-
Voice Expenses Voice tioning). The actual transmission of en-
Dominated Dominated ergy is only 7 percent of the OPEX. If
Growth
Growth
Revenue Revenue more base stations are deployed, the

Expenses
30 percent of site rent scales accord-
Mobile Data
Explosion
Mobile Data ingly, making it unattractive to simply
Explosion
deploy more base stations (this could
Time Time
become a problem for 5G FR2 with re-
s Fig. 1 5G Business Case. Drive profit by reducing expenses (energy efficiency). duced cell sizes compared to 5G FR1).
4
From the analysis of the power consumption, it is clear dependent of the Radio Access Network and can be used
that most of the costs are due to the distributed and re- to control a mixed generation cellular network.
mote deployment of air conditioning for the baseband The type of information being transmitted also influ-
processing portion of a base station. China Mobile pro- ences the energy efficiency of the network. As shown
posed to centralize the baseband processing in a manner in Figure 2, different types of data have different data
similar to data storage facilities for the internet. Figure 3 packet to signaling packet ratios (DSR). A low DSR rep-
illustrates the architecture of the baseband cloud where resents a low usage of the channel to transmit data; for
each baseband for a base station becomes a virtual ma- example, text messages which represent 60 percent of
chine inside the cloud (C-RAN). Even traditionally separate all network traffic have a DSR between 1 to 3 whereas
network appliances such as gateways can be integrated photos and videos are more energy efficient as this data
into the cloud as virtual machines. By centralizing the requires fewer signaling packets. 5G FR1 addresses this
baseband processing, the remote air conditioning be- problem by adjusting the subcarrier spacing to allow
comes centralized, thereby significantly decreasing both different types of data to more efficiently use the avail-
OPEX and CAPEX. In addition, it is easier to realize CoMP able channel capacity.
where separate base stations transmit to a mobile phone
(network MIMO) with a centralized control, increasing the
spectral efficiency at the same time. This architecture is in-
BS1: GSM BS2: LTE ... BS3: 5G
Phy/Mac Phy/Mac Phy/Mac
Cellular Network Energy Consumption (China) RTOS RTOS ... RTOS
2G GSM Hypervisor
830,000 Base Stations
General Purpose Processor Platform
80 GWH (96 KWH per BTx)
3G TD-SCDMA
Virtual Base Station Pool (Real-Time Cloud BBU)
4G TD-LTE High Bandwidth Optical Transport Network

s Ba l
es
ckh au
ckh
Ac
c au
Ac l Ba
ce
Service Ratio (%) Packet Data to Signaling
ss
Acc
Type Size (kB) Ratio (DSR)

B C
ess
Text/IM 60 1 1 to 3 s
c es
Voice 35 10 Ac
Photo 4 150 65 to 375 A Fiber Transport
Video 1 1500
Distributed Configurable Wideband RRU
Radio Access Network Energy Consumption Centralized Control/Processing

• Centralized Processing Resource Pool that can Support 10~1000 Cells
3%
Collaborative Radio
• Multi-Cell Joint Scheduling and Processing
O&M
21% Real-Time Cloud
• Target to Open IT Platform
Air-Con Equipment Electricity
51% 46% • Consolidate the Processing Resource into a Cloud
41%
Site Rent • Flexible Multi-Standard Operation and Migration
31%
Tx Clean System Target
7% • Less Power Consuming –15% Capital Costs
–50% Operating Costs
• Lower OPEX
CAPEX OPEX –70% Power Consumption
• Fast System Roll-Out
Biggest CAPEX/OPEX Expense is Air Conditioning

CMRI, ”C-RAN: The Road Towards Green RAN“ Dec. 2013 Architecture Equipment Air Con Switching Battery Transmission Total
Example: China Mobile Network in 2013 Consumed Traditional 0.65 kW 2.0 kW 0.2 kW 0.2 kW 0.2 kW 3.45 kW
Over 15 Billion KWH Cloud Radio 0.55 kW 0.01 kW 0.2 kW 0.1 kW 0.0 kW 0.86 kW
Source: IEEE Communications Magazine, Feb.2014
s Fig. 2 Power Consumption in a Cellular Network. s Fig. 3 Centralized Baseband Processing.
5
More Channels = MIMO (5G FR1) Shannon Channel Information Capacity SNR Increase (log2 Increase)
Linear Increase
Capacity (Bits/Seconds)
x1(t) M = 4 Transceivers
x2(t)
C = W log2 (1 + γ)
x3(t)
x4(t) Signal BW (Hz) SNR (S/N)
Massive MIMO
Larger Signal BW = mmWave (5G FR2)

×××× Linear Increase
××××
××××
××××
Fiber
Coverage High Capacity
Mobility, Reliability Massive Throughput
Distributed
BU FR1: Sub-6 GHz FR2 mmWave: 20-90 GHz
s Fig. 4 Spectral Efficiency.
WHAT DETERMINES THE CAPACITY OF A and 5G FR2 uses the mmWave frequencies to obtain
NETWORK AND HOW CAN IT BE INCREASED? larger capacities.
In the early 1900s, two researchers working indepen- 2. Increase the number of channels: MIMO utilizes the
dently derived a relatively simple equation that serves multipath scattering inside the network to concur-
as a Moore’s Law for the wireless industry: the Shannon- rently transmit on several channels at the same time.
Hartley theorem. This theorem gives an upper bound to Similar to the channel BW, network capacity also
the amount of information that can be transmitted over scales linearly with this effect, but with an upper limit
the wireless channel where the individual channel capaci- determined by the correlation (or similarity) of the
ty is dependent on only two parameters: channel BW and multipaths inside the network. 5G FR1 relies on scal-
the signal-to-noise-ratio (SNR). While the capacity scales ing up MIMO to provide increased data rates.
linearly with the channel bandwidth, it only scales at log2 3. Increase the output power of the network: Due to the
for the signal to noise ratio: presence of the noise in the SNR, the asymptotic log
Cj = BWlog2(1 + SNR) scaling of the SNR, and health/safety concerns of high
From the Shannon-Hartley theorem, there are three ba- electromagnetic energy, this method has its limits. One
sic methods to increase network capacity (see Figure 4): safer method of increasing the SNR throughout the
1. Increase the channel BW: In 4G, carrier aggregation network is the use of femtocells in areas of decreased
is used to increase the available signal bandwidth coverage. If too many omnidirectional femtocells are
deployed in a single area, however,
Traditional Base Station Antenna Massive MIMO Antenna Array the interference between the femto-
cells provides an upper limit to the
Wasted Power
capacity gain of the network. By tar-
geting the energy to a specific user,
however, the energy efficiency of the
..
Wasted Power . network can be increased—this is
referred to as “beamforming” and it
is a key technology for both 5G FR1
Number of UEs: 1
Number of Antennas = 1 120 Antennas per UE and FR2 base stations.
Number of BS Transmit 1 120
Antennas IMPROVEMENT OF
1 1 ENERGY EFFICIENCY WITH
Normalized Output Power Pant = =1 Pant = =1
of Antennas Mt Pt2 BEAMFORMING
Mt Mt In a traditional cellular network, a
Normalized Output Power
of Base Station Ptotal = ∑P i =
ant 1 Ptotal = ∑ P ˜ 1/1000
i
ant
cell is associated with a base station
i=1 i=1 that transmits energy throughout a
Source: IEEE Signal Processing Magazine, Jan. 2013
wide area (typically 120 degrees an-
Improve Energy Efficiency: Beamforming gular arc in front of the base station).
While some of this energy is received
s Fig. 5 Beamforming and Energy Efficiency. by users within the base station’s cell,
6
Beamsteering Sidelobe Suppression
10 10
Principle of Beamforming & Beamsteering ∆φ = 0 ∆φ = 90
∆φ = 90 Linear Taper
0 0
1. Fixed Antenna Spacing d
2. Choose Direction ϴ Broadside
3. Set Phase Shifts ∆φ –10 –10
Gain (dB)
Gain (dB)
To Far-Field –20 –20
ϴ
–30 –30
2π
∆ϕ =
λ d sin ϴ –40 –40
φ1 φ2 φ1
φM –50
–50 0 50
–50
–50 0 50
Antennas d . . . Angle (°) Angle (°)
Phase Shifters . . .
Attenuators . . .
s Fig. 6 Principles of Beamforming.
Single Transceiver + Antenna nth TRx Module

Circulator
Probe m Antenna or Switch
th
Receive RF Chain Rn
Measurement
Equipment p Cm g
m
Transmit RF Chain Tn
From Analog... ...To Digital ...To Hybrid
Analog Beamforming (ABF) Digital Beamforming (DBF): 5G FR1 Hybrid Beamforming (HBF): 5G FR2
Data Digital
Digital Baseband RF •
Stream 1 Baseband RF Chain
•
•
Chain Processing
Processing
•
•
• •
K Data •
•
•
•
•
•
Data Stream Baseband •
•
•
•
• Streams • • •
Processing Data Digital Digital

•
•
• Stream n Baseband RF Baseband RF •
•
Chain Processing Chain •
Processing
Digital Beamforming Analog
Beamforming
s Fig. 7 Beamforming Architectures.

the vast majority of this energy is absorbed into the desired direction, there is also energy going in other
environment (buildings, people, trees, cars, etc.). This directions (sidelobes and backlobes). This additional
wasted power represents a reduced energy efficiency energy represents interference to other users inside a
and higher network OPEX (see Figure 5). If the single base station cell. This effect can be mitigated by either
base station antenna is replaced by 120 antennas that ensuring the adjacent users are in the nulls of the main
target the energy to individual users, the required base beam or by weighting the individual antennas with an
station power is decreased to 0.1 percent of the original amplitude distribution to lower the energy in the sid-
output power.3 This reduction, however, is theoretical. elobes (see Figure 6).
Practically, the output power for the same capacity only There are three types of beamforming architecture
decreases to 30 percent of the original power due to that have direct impact on the energy efficiency of the
the efficiency and losses of the RF components inside base station and the UE (see Figure 7):
the base station. • Analog Beamforming (ABF): The traditional way to
In order to form a beam, a set of periodically form beams is to use attenuators and phase shifters
spaced antennas can form a beam in any direction as part of the analog RF circuit where a single data
by only changing the phase differences between the stream is divided into separate paths. The advantage
antennas (see Figure 6). Typical antenna array spacing of this method is that there is only one RF chain (LNA,
is half wavelength, leaving the beam angle (θ) directly filters, switch/circulator) required. The disadvantage
related to the phase difference between the antennas: is the loss from the cascaded phase shifters at high
θ ∝ sin-1(∆φ). While the beam can focus its energy in a power.
7
• D
igital Beamforming (DBF): DBF assumes there is a compared to an LTE network while decreasing costs by
separate RF chain for each antenna element. The beam a factor of 10 (assuming that there are eight transceivers
is then “formed” by matrix-type operations in the base- per user with beamforming). While 5G FR1 has limited
band where artificial amplitude and phase weighting is frequency spectrum available, 5G FR2 uses large amounts
performed. For frequencies lower than 7 GHz in 5G FR1, of spectrum above 24 GHz. The spectral efficiency of 5G
this is the preferred method since the RF chain com- FR2 (assuming HBF with eight transceivers per antenna
ponents are comparatively inexpensive and can com- array) is comparable to LTE at 10 bps/Hz, but with higher
bine MIMO and beamforming into a single array. For energy efficiency compared to LTE4 (see Figure 8).
frequencies of 28 GHz and above, the PAs and ADCs In conclusion, the combination of spectral and energy
are very lossy for standard CMOS components. If exotic efficiency allows operators to deploy new networks that
materials, such as GaAs and GaN are used, the losses can simultaneously boost capacity while lowering OPEX.
decrease at the expense of higher cost. Networks of the future will combine the different solu-
• Hybrid Beamforming (HBF): HBF combines DBF with ABF tions of FR1 and FR2 into single networks, allowing the
in order to allow the flexibility of multiple radio transceiv- FR1 to deliver high dates in a wide area network with in-
ers plus beamforming while reducing the cost and losses building penetration while FR2 serves as data off-load-
of the beamforming unit (BFU). Each data stream has ing, hotspots and extreme network densification. This
its own separate analog BFU with a set of M antennas. network deployment will not only affect consumers and
If there are N data streams, then there are N×M anten- equipment vendors, but also has a fundamental impact
nas. The analog BFU loss due to phase shifters can be on the entire test & measurement industry.
mitigated by replacing the adaptive phase shifters with a
selective beamformer such as a Butler matrix. One pro- IMPACT ON TEST & MEASUREMENT INDUSTRY
posed architecture uses the digital BFU to steer the direc- The new base stations required for 5G result in a new
tion of the main beam while the analog BFU steers the measurement paradigm with the use of over-the-air (OTA)
beam within the digital envelop. testing for both antenna and transceiver performance char-
acterization.
OPTIMAL NETWORKS: SPECTRAL AND ENERGY
EFFICIENCY 5G BASE STATION ARCHITECTURE
The combination of C-RAN, MIMO, new spectrum The combination of beamforming and MIMO into a
and beamforming allows 5G to both increase capac- single array leads to a massive MIMO base station as mul-
ity while decreasing costs compared to traditional and tiple sets of antennas are required for both beamforming
current cellular networks. The Shannon-Hartley theorem (same data vector for each antenna) and MIMO (differ-
can be reformulated to take into account the energy ent data vectors for each set of beamforming antennas).
efficiency of the channel.2 Using the constraints in the Designing these base stations so that both spectral and
performance of the base stations and the network, the energy efficiency can be increased is complex and re-
joint optimal spectral-energy efficiency of 2G and 4G quires very tight integration of all the components (see
networks can be calculated at 4 bps/Hz for GSM and 8 Figure 9):
bps/Hz for LTE. Note, in real networks, these values are • Beamforming Architecture: This depends on the avail-
often lower at 4 bps/Hz for LTE. ability of components both in terms of losses (energy
The combination of MIMO and digital beamforming efficiency) and cost.
in 5G FR1 can lead to an increase of capacity of over 3×
12 x 107
1 TRx + DBF
10 8 TRx + DBF
2G & 4G Capacity
EE (bits/J)
8 5G Optimal Capacity for DBF

1 x 106 Massive MIMO
6 (8 TRx Per User)
FR1 SE: 30 bps/Hz
1 x 105
EE: 6 × 107 bits/Joule
EE (bits/J)
1 x 103
10 20 30 40 50
SE (bps/Hz)
4 x 107
GSM
LTE Macro
LTE Small Cell 3 5G Optimal Capacity for
Hybrid Beamforming
EE (bits/J)
FR2 mmWave (50) Antennas)

2 4 6 7 8 1 TRx + ABF (8 TRx Per Array)
SE (bps/Hz) 8 TRx + ABF SE: 10 bps/Hz
EE: 3.5 × 107 bits/Joule
2G Optimal Capacity: 2-4 bps/HZ
4G Optimal Capacity: 8-10 bps/HZ
Sources: IEEE Communications Magazine, Feb, 2014 & Jan. 2015
5 10 15 20
SE (bps/Hz)
s Fig. 8 Optimized Networks: Spectral and Energy Efficiency.
8
• W
ideband Power Amplifier and
Filter Banks: The increase of the •••
Antenna
number of frequency bands, carrier Beamforming Mutual
aggregation across wide frequency Architecture Self-
Coupling
bands will require large numbers of Calibration
Circuit
both filters and power amplifiers.
The power amplifiers will require Wideband Adaptive
Power Amplifier Self-Calibration
pre-distortion or exotic materials to and Filter Banks
operate at higher efficiencies.
•
Antenna Mutual Coupling: Sim-
ply packing more antennas into a mmWave = RFIC RFIC Clock
space will decrease the capacity non-CMOS Synchronization
Components
and increase the losses in the base FPGA LO
station.
• Clock Synchronization: For a large Fiber Transceivers Fiber Multiplexing Heat Dissipation
massive MIMO array, the clocks will Digital I/Q
require synchronization over separate
PCB boards. Clock drift will lead to s Fig. 9 Massive MIMO Architecture.
uncertain phase changes (due to fre-
quency drift) between the antennas
and decrease the effect of beamform- Active Antenna System (Massive MIMO or mmWave)
Baseband
ing.
• Adaptive Self-Calibration: Due to the Dig
large numbers of components, chip- I/Q Traditional
sets, clocks and amplifiers, together
with the dependency of phase on
Baseband
temperature conditions inside the
base station, the output phase at each
antenna can vary significantly from the 64-128 Antennas 8-128 RF Transceivers FPGA + Fiber TRx 5G
desired phase. Therefore, an adap-
tive self-calibrating circuit is required s Fig. 10 New Measurement Paradigm for 5G.
where the phase and amplitude offset
of each signal is measured and is then Antenna Array (30 cm) at 28 GHz Reactive Near Field Region
pre-distorted so that optimal beam-
forming can be achieved.
• Fiber Transceivers: The output of the
massive MIMO base station is typical-
D = Radiating Apeture Size
ly baseband that is then transmitted
through fiber to either a local base-
band unit or into the C-RAN. Conse-
quently, a real-time FPGA is required
to translate the output baseband data Reactive Near
Field Region
Radiated Near Field Region
Phase & Magnitude
Far Field
Magnitude
from the RFIC to the baseband proto-
col for fiber.
• Heat Dissipation: The combination Any Object in this Region Becomes
√‾λ
D3 2D2 Part of Antenna System & Interferes
of up to hundreds of antennas, thou- 0.62 = 16.0 m with the Measurements
λ
sands of components and dozens of
RFICs/FPGAs within a confined space s Fig. 11 Antenna Electromagnetic Fields.
leads to significant heat and thermal
issues. As these units are deployed without externally between the RF test ports and the measurement instru-
provided air cooling in areas of large temperature varia- ments (i.e., vector signal analyzer and signal generators).
tion, large heatsinks are required, significantly increasing The antenna performance was typically measured OTA
the weight of the massive MIMO unit. using CW waveforms with a vector network analyzer.
Due to the highly integrated architecture of the mas-
TEST & MEASUREMENT OF 5G BASE STATIONS sive MIMO base station, there is no longer direct access
AND DEVICES to the individual RF paths. This implies a fundamental
Traditionally, the performance of a base station was change of measurement philosophy by moving away
measured by separating the antenna portion from the from highly predictable cable measurements of the RF
radio/RF portion. The modulated performance of the transceivers to a more chaotic environment consisting of
RF transceivers could be measured directly with cables OTA measurements (see Figure 10).
9
Passive Measurements Active Measurements OTA Test Solutions
2D/3D Antenna Characterization RF Transceiver Characterization
System and System and

AUT Control AUT Control
R&S®AMS32 OTA Performance R&S®AMS32 OTA Performance

Measurement Software Measurement Software
R&S®PWC200 Plane Wave Converter
R&S®CONTEST R&S®CONTEST
R&S®SMW200A R&S®ATS800 CATR Rack

Vector Signal Generator Based Antenna
Test System
R&S®ZNA Vector Network
Analyzer
R&S®ATS1800 CATR Conformance

R&S®FSW Signal and Chamber System
Spectrum Analyzer
s Fig. 12 Plane Wave Converter and CATR
OTA measurements are significantly more complex as well. For this reason there will be a high demand for
than cable measurements due to the different physical OTA chambers and measurement equipment to not
properties of the radiated fields (see Figure 11) in the only measure the strict radiative properties of anten-
near-field and far-field regions of the device under test nas, but substituting traditional conducted transceiver
(DUT). Due to both the time and spatial-varying proper- measurements as well. Rohde & Schwarz, with its wide
ties of the modulated signals, measurements must be range of anechoic chambers and measurement equip-
performed in the far-field of the DUT (planar waves), re- ment expertise, is well situated to deliver solutions even
sulting in either very large anechoic antenna chambers or for future customer requirements.5n
indirect far field chambers such as the plane wave con-
verter (PWC) or compact antenna test range (CATR). A References
1. CMRI, “C-RAN: The Road Towards Green RAN,” December 2013.
CATR uses a reflector to convert a spherical wave to a 2. I. Chih Lin, C. Rowell et al., “Towards Green and Soft: A 5G Perspective,”
planar wave distribution in the near field of the reflector IEEE Communications Magazine, February 2014.
and the PWC uses an array to generate a planar wave 3. F. Rusek et al., “Scaling Up MIMO: Opportunities and Challenges with
distribution in the near field of the array (see Figure 12). Very Large Arrays,” IEEE Signal Processing Magazine, January 2013.
4. H. Shuangfeng et al., “Large Scale Antenna Systems with Hybrid Analog
Due to the elimination of RF test ports and the use of and Digital Beamforming for Millimeter Wave 5G,” IEEE Communica-
frequencies in the millimeter region, OTA will become tions Magazine, January 2015.
an essential tool for characterizing the performance of 5. “Antenna Array Testing - Conducted and Over the Air: The Way to 5G,”
Rohde & Schwarz White Paper, 2016.
not just the antenna arrays of an active antenna system
of a massive MIMO array, but the internal transceivers
10
NR and LTE Coexistence Through
Dynamic Spectrum Sharing
Andreas Roessler
All 5G NR FR1 deployments are currently based on time-division duplex (TDD) and use unpaired
frequency bands, typically in the 3.5 GHz frequency range. The second phase of 5G NR network
deployments, anticipated for 2020, will use frequency-division duplex (FDD) mode, as almost 90
percent of the spectrum below 8 GHz is organized as paired frequency bands, where downlink
and uplink use different frequencies. All the targeted frequency bands, however, are already in
use by 4G LTE. By adding spectrum sharing capabilities to the 5G NR standard, this spectrum
can be accessed while in use, enabling coexistence between 4G LTE and 5G NR. The sharing
of the spectrum allows network operators a smooth transition from LTE to 5G without the need
for spectrum refarming. This article discusses the required feature sets, summarized as dynamic
spectrum sharing (DSS), and analyzes related test and measurement challenges for network and
device testing.
T
he first 5G NR networks are on-air using network the first 5G NR network deployments took advantage of
deployment option 3X and utilizing E-UTRA NR the underutilized TDD frequency bands, including 3.5
Dual Connectivity (EN-DC) with a split bearer GHz. The first generation of 5G modems and, subse-
setup. In this deployment scenario, the so-called non- quently, the first generation of 5G mobile devices do
standalone (NSA) mode, an LTE anchor is required to only support the TDD mode for FR1. FR1 FDD is some-
exchange control and signaling information. In addition thing that the industry is still working to commercialize.
to LTE signaling, the anchor is also required to config-
ure, add, modify and release the connection to the 5G THE NEED FOR DSS
NR radio access network (RAN). In this setup, the LTE Not all service providers own spectrum licenses with-
base station (eNB) takes on the role of the master cell in a TDD band. To take advantage of 5G with optimized
group, where the 5G base station (gNB) becomes the quality of service, to lower latencies and to further ad-
secondary cell group. Both RANs connect to the exist- dress the new market verticals (e.g., automotive and
ing LTE core network, the evolved packet core. industrial), a network operator must transition to stand-
According to the 3GPP standard, carrier aggregation alone (SA) mode, in which the 5G RAN is connected
can be activated for each cell group. However, today's to the 5G core network (5G-CN, option 2). There are
5G deployments at sub-8 GHz frequencies, also called several intermediate steps (options 4, 5 and 7) defined
frequency range 1 (FR1, from 410 MHz to 7.125 GHz), that lead toward a standalone deployment. Which path
combine multiple LTE carriers with typically one NR car- an operator follows is up to its 5G deployment strategy.
rier. A vast majority of these networks worldwide use 3.5 For a detailed description of these options and other
GHz, with a carrier bandwidth of up to 100 MHz, apply- fundamental aspects of the fifth generation of wireless
ing 4 x 4 MIMO and using TDD mode. communication, please refer to Kottkamp et al.2
Due to various local regulatory requirements, the 3.5 Due to the occupation of its FDD-based spectrum
GHz band comprises three different frequency bands assets, service providers are forced to choose between
from a standardization perspective. These are n77 the acquisition of new spectrum or refarm spectrum al-
(Asia), n78 (Europe) and n48 (USA).1 As the majority of ready in use. Both options are costly. Therefore, the 5G
frequency bands worldwide are FDD and used by LTE, NR standard offers the possibility of adapting to existing
11
s Fig. 1 NR signal configuration within the MBSFN subframe.
LTE deployments and sharing the spectrum used exclu- part of the evolved multimedia broadcast multicast ser-
sively by LTE today. The enabling feature is called DSS, vices (eMBMS) functionality; eMBMS is part of 3GPP’s
which is part of the overall mechanism allowing NR and release 9 set of technical specifications.
LTE to coexist while using the same spectrum. Six out of ten subframes forming the LTE radio frame
Another feature that enables NR LTE coexistence is can be configured by the network to become MBSFN
the decoupling of downlink and uplink transmissions, subframes. Based on the standard, this could be sub-
necessary due to coverage issues in the uplink direction frames 1, 2, 3 and 6, 7 and 8. To minimize the impact
at the FR1 midrange frequency bands of 3.5 GHz and on the performance of LTE, typically only one subframe
above. This feature results in a function called supple- out of the six possible subframes is configured to be an
mental uplink (SUL) that allows a device to switch its up- MBSFN subframe. The applied configuration is broad-
link transmission from the midrange spectrum to the low cast by the LTE network with system information block
band spectrum based on received signal quality. This type 2 (SIB2). This is the same SIB that informs a 5G-
feature is not discussed in this article, as the industry is capable terminal that the LTE serving cell can connect
currently focusing on DSS. the handset to the 5G RAN. A standard LTE terminal
In the long term, DSS enables network operators to would read in the MBSFN configuration from SIB2 and
provide a coverage layer for 5G using the lower fre- ignore the subframes configured for broadcast. Initially,
quency bands, typically targeting frequencies below 1 DSS is tested based on NSA mode; thus, the 5G hand-
GHz. DSS needs infrastructure updates, advertised as set would have two radios active, LTE and 5G NR.
software-only updates, but also requires the second- The LTE portion will follow the same principles as an
generation of 5G chipsets and handsets taking advan- LTE-only device. However, the 5G NR part of the hand-
tage of these new modems. The rollout of DSS is ex- set, scanning the targeted frequency band for sharing,
pected early to mid-2020. will detect the transmitted SSB within the open LTE sub-
frame on the desired frequency channel. As DSS is in-
THE AWAKENING OF MBSFN tended to provide a coverage layer for 5G, typically fre-
DSS has an impact on both LTE and 5G NR standards. quency bands below 1 GHz are shared between LTE and
The effect on LTE is marginal, as it is hard to change a 5G NR. Initially, a subcarrier spacing of 15 kHz is used
successfully deployed technology to enable its succes- for 5G NR to avoid interference between both technolo-
sor. A 5G NR device needs to detect the synchronization gies, resulting in the same number of subframes used
signal blocks (SSB) to access the network. To maintain in LTE and 5G NR. Based on the mapping principles for
synchronization in time and frequency, these SSBs need SSB for 15 kHz, defined as case A, the targeted frequen-
to be sent periodically by the network, with a gap de- cy band below 1 GHz leads to a maximum of four SSBs
fined to transmit the SSB on an already occupied fre- transmitted by the 5G network.
quency channel used by LTE. The ideal feature to allow Typically, the SSBs are transmitted by the network in
this gap in a continuous LTE transmission is to use mul- the first half-frame (5 ms) of a radio frame. As it is not
timedia broadcast single frequency network (MBSFN) possible in LTE to configure subframe 0 for MBSFN,
subframes. Defined initially to enable broadcast via an subframe 1 is configured as the MBSFN subframe, so it
LTE network and, therefore, make the transfer of content can carry SSBs. However, an MBSFN subframe is never
more efficient than unicast transmissions, this feature is entirely empty. There is a non-MBSFN region defined
12
that can be one or two OFDM symbols long, dependent uler in the LTE base station is not mapping any data into
on overall signal bandwidth. This region is intended to such a subframe, the LTE CRS would still be present,
carry the control channels for LTE, such as the physical maintained and transmitted.
hybrid ARQ channel, the physical control format indica- To enable NR to use these subframes, rate-matching
tor channel and the physical downlink control channel around LTE CRS is adopted by the standard. Several fac-
(PDCCH). Therefore, any NR transmission can only start tors impact the rate-matching algorithm. The first pa-
at OFDM symbols 2 or 3 within an MBSFN subframe. rameter is required to align the subcarrier positioning
The requirement to transmit control information for for 5G NR related to LTE. The value corresponds to the
NR, to schedule the reception of NR’s physical data offset to point A divided by 15 kHz. Second, the band-
shared channel (PDSCH) and the necessity to map a width and number of antenna ports used by LTE must
demodulation reference signal for the data channel to be known, as the MIMO mode dictates the mapping of
the beginning of the subframes lead to the configura- CRS per antenna port. As mentioned earlier, DSS tar-
tion illustrated in Figure 1. Out of the total 14 OFDM gets frequencies below 1 GHz; therefore, typically two
symbols forming a subframe of 1 ms duration for subcar- antenna ports (2 x 2 MIMO) are used. Lastly, the fac-
rier spacing of 15 kHz in one 5G NR slot, only 12 OFDM tor vshift represents the impact of the physical cell iden-
symbols are available for NR transmission. For proper tity (PCI; vshift = PCI mod 6), which defines the starting
demodulation of the PDSCH and to enable mobility, a point (subcarrier) for the mapping of the LTE sequence
second symbol carrying the PDSCH demodulation refer- used for generating CRS. Figure 2 displays all required
ence signal (DMRS) for the data channel is required. This parameters to configure the LTE CRS rate-matching
additional symbol is symbol 9, as shown in Figure 1.3 algorithm on the R&S®SMW200A vector signal genera-
tor, intended to perform physical layer testing of the
INCREASING CAPACITY AND IMPROVING 5G NR 5G modem.
WHILE DSS IS CONFIGURED
With just one subframe available for 5G NR, the tech-
nology operates under its potential. Therefore, DSS ad-
ditionally enables the use of subframes that are dedicat-
ed to LTE and not configured for MBSFN via two distinct
features:
1. Depending on the MIMO mode (2 x 2, 4 x 4),
standard LTE subframes include cell-specific reference
signals (CRS) mapped to certain resource elements in
the time-frequency grid. An LTE terminal uses CRS for
channel estimation and to maintain full synchroniza-
tion in time and frequency. Assuming a simple scenario s Fig. 2 LTE CRS coexistence setting for the R&®SMW200A
vector signal generator.
where, based on the current load situation, the sched-
s Fig. 3 NR using standard LTE subframes with CRS rate-matching and additional DMRS position.
13
In a real network, this information is sig-
naled to the device via the dedicated radio
resource control (RRC) connection. As NSA
mode is the initial deployment mode for
DSS testing, this information is sent over
the established LTE connection to the de-
vice. In SA mode, it would be sent over the
NR RRC connection.
2. The second feature is the support of
an additional position for the mapping of
the PDSCH DMRS. Based on a standard LTE
subframe, with the LTE control channel and
CRS present, and assuming the scheduler
does not schedule any PDSCH, the remain-
der of the subframes are available to 5G NR.
Therefore, the control resource set (CORE- s Fig. 4 Dynamic sharing of a subframe between LTE and 5G NR.
SET) and the NR PDSCH with rate-matching
active, including DMRS, are mapped on the symbol offset of 2, which, under certain conditions, is
available resource element (see the example in Figure not favorable when applying DSS between LTE and 5G
3). Due to the CORESET occupying OFDM symbol 2, NR. PDSCH mapping type B overcomes this drawback.
the first PDSCH DMRS is assigned to symbol 3. The po- Figure 4 shows a configuration example for LTE with a
sition information (l0) of the OFDM symbol (2 or 3) that 5 MHz channel bandwidth, where NR is transmitted on
carries the first DMRS is indicated with the master infor- the upper 6 out of 25 resource blocks.
mation block carried by the physical broadcast channel
as part of the SSB. To support mobility, proper chan- COORDINATION, COORDINATION AND, AGAIN,
nel estimation is a prerequisite that can only be guaran- COORDINATION
teed if at least two symbols within a slot carry PDSCH All described physical layer features require coordi-
DMRS. According to 3GPP TS 38.211 V15.6.0 (2019- nation among both RANs as in a dual connectivity ap-
06),3 for the provided example, this additional symbol proach; both (LTE and NR) use two independent sched-
to carry DMRS would be OFDM symbol 11. However, ulers. The resulting E-UTRA NR cell resource coordina-
from an LTE perspective, CRS is still present and trans- tion procedure (see Figure 5) that can be triggered by
mitted in this symbol; therefore, the additional position both nodes over the Xn interface is defined in 3GPP TS
of the DMRS must move from symbol 11 to symbol 12 38.423 V15.4.0 (2019-07), NG-RAN.4
(see Figure 3). This feature is a device capability, which The process allows the coordination of scheduling
means the device signals its support of this functionality resources in the frequency and time domains at the me-
to the network during the initial registration process. dium access control layer in both base stations. When
In general, this change is only applicable if three initially testing DSS in NSA mode, the coordination pro-
conditions are fulfilled: First, the device needs to have cess is triggered by the eNB. In its request message, the
submitted the support of the capability to the network; eNB sends the data traffic resource indication toward the
second, the network has configured the device with the gNB. This information element contains the information
rate-matching parameter for LTE CRS via RRC; third, the if the sharing is in uplink only or for uplink and downlink.
first position of the PDSCH DMRS is set to l0 = 3. For the latter case, two individual bitmaps are provided
with the message, containing a bitmap between 6 and
WHERE IS THE DYNAMIC SHARING ELEMENT? 17,600 bits long. Each position in this bit string stands
Up to now, what has been discussed is a semi-static for a physical resource block (PRB) pair that is reserved
configuration of both LTE and NR to enable the use of for E-UTRA if it is set to 1 and is not used for E-UTRA
specific subframes for NR when LTE is not present or when it is set to 0. The bit string may span multiple con-
mechanisms that allow NR to transmit in LTE subframes tiguous subframes. The first position of the data traffic
not used by LTE but where essential LTE signals com- resources information element corresponds to the re-
ponents are still sent. The question is whether there is
a way for LTE and NR to share a subframe and for both
to transmit control information (PDCCH and CORESET)
and data (PDSCH)? The answer is yes. E-UTRA – NR Cell Resource
So far, the discussion has focused on the fundamental Coordination Request
mapping of the data channel and corresponding DMRS E-UTRA – NR Cell Resource
for 5G NR based on the so-called PDSCH mapping type Coordination Respone
A. But the 3GPP standard also defines PDSCH mapping
type B. The difference between these two is that type
eNB gNB
A defines the mapping relative to the slot start, where
X2
type B represents the mapping relative to the begin-
ning of the PDSCH within the slot. PDSCH mapping
type A allows, according to the standard, a maximum s Fig. 5 E-UTRA NR cell resource coordination procedure.
14
ceiving node’s subframe 0. The length of the
bitmap is an integer multiple of the band-
width. Assuming the signal bandwidth to be
5 MHz (25 PRB), a total of 704 subframes
are addressed through the bitmap. The bit-
map length doesn’t provide the flexibility to
address all subframes within radio frames,
which makes it necessary to provide infor-
mation for which the system frame number
within the bitmap becomes active. Figure 6
is an example of the bitmap. Based on this
configuration, resource blocks 15 to 24 in
subframe 0 (light blue), resource blocks 5 to
24 in subframe 1 (green), resource blocks 3
to 17 (brown) and resource blocks 8 to 22
(purple) in subframe 703 can be used for 5G
NR transmission and reception.
Other resources in LTE are protected
through the exchange of the protected E-
UTRA resource indication message between
the eNB and the gNB during the setup of
the X2 interface, which is a separate but
mandatory prerequisite. As explained ear- s Fig. 6 Data traffic resource pattern for NR LTE resource coordination (5 MHz,
lier, LTE configures MBSFN subframes to al- 25 PRB).
low the transmission of SSBs for NR. These
subframes are protected from the above configuration
by additional information provided by the data traffic
resource indication as a reserved subframe pattern.
TESTING FOR DSS

DSS is a powerful feature, but it requires extensive
testing for several reasons. This applies to lab-based
LTE and 5G user equipment testing as well as carrying
out network performance measurements using scanners
(sensitive receivers) and devices to estimate coverage
and end-to-end (E2E) performance.
The activation of DSS within the network should not
create any interference for the exiting LTE deployment.
LTE-only devices must not suffer any interference or im-
pact when configuring MBSFN subframes within the net-
work. Further, with MBSFN active, end-to-end through-
put tests are required to ensure minimal impact on LTE s Fig. 7 R&S CMW500 and R&S CMX500 for testing mobile
performance. While 5G NR, including SSB, are transmit- devices supporting 5G NR, LTE and legacy technologies.
ted within MBSFN subframes, receiver sensitivity tests
for LTE devices must be favorable, to ensure sensitivity
requirements are still met by the device when 5G NR is
present within the channel at defined subframes.
A 5G NR capable device must be able to synchro-
nize in time and frequency with the 5G RAN when SSBs
are transmitted within MBSFN configured subframes.
When 5G NR is sent in non-MBSFN subframes using
an LTE CRS rate-matching pattern for NR’s PDSCH, a
data throughput test is adequate to verify correct imple-
mentation of the stack features. Advanced device test-
ing includes dynamic scheduling procedures that mimic
the described E-UTRA NR resource coordination proce-
dure, where data is scheduled according to data traffic
resource indication, including the validation of PDSCH
mapping types A and B. The new R&S CMX500 mobile
radio tester platform from Rohde & Schwarz, in combi-
nation with the R&S CMW500 wideband radio commu- s Fig. 8 R&S TSME network scanner and R&S ROMES drive
nication tester, is an example of an appropriate tool to test software for assessing 5G NR and LTE network perfor-
mance.
carry out these extensive test scenarios on mobile de-
vices supporting LTE and 5G NR (see Figure 7).
15
Another change: with release 16, the definition of
Different LTE CRS Rate Matching Pattern
multiple LTE CRS rate-matching patterns is supported
for one device. The reason is that a network operator
may have deployed multiple LTE carriers within a fre-
LTE, 10 MHz LTE, 20 MHz LTE, 20 MHz quency band (three shown in Figure 9). These three car-
MIMO 2 × 2 MIMO 4 × 4 MIMO 4 × 4
PCI 301 PCI 304 PCI 307 riers can use different bandwidths and MIMO schemes
(e.g., 10 versus 20 MHz, 2 x 2 versus 4 x 4 MIMO). Even
if the carriers have the same bandwidths and MIMO
NR, 50 MHz schemes e.g., carriers 2 and 3), they have different PCI
MIMO 4 × 4
PCI 697 that dictate which resource elements carry the CRS and,
therefore, impact the rate-matching algorithm. 5G NR
supports wider bandwidths, e.g., 50 MHz, and could use
Frequency
the entire channel. The definition of multiple LTE CRS
s Fig. 9 Multiple LTE CRS rate-matching pattern with 3GPP rate-matching algorithms to be used by the device sup-
release 16. ports this specific deployment scenario and enables the
use of DSS functionality.
DSS has a further impact on mobile network testing
when performing coverage measurements and network SUMMARY AND CONCLUSION
optimization for LTE and 5G NR. The toolset used by DSS is the hot buzzword in the wireless industry to-
network equipment providers during initial rollout and day. It provides a service provider with the possibility
network operators during optimization and maintenance to provide a coverage layer for 5G NR at low band fre-
is based on a network scanner, a sensitive receiver and quencies. Besides, it allows the carrier to smoothly tran-
drive test software, collecting the measurements from a sition the LTE subscriber base toward 5G and roll out
passive probe. A mobile device complements the setup a standalone mode faster by avoiding the high cost of
that is used to perform end-to-end performance tests in spectrum refarming or the need to acquire new spec-
the network: file download and upload or video stream- trum licenses.
ing, for example (see Figure 8). In this setup, a scanner From a technology standpoint, DSS combines sever-
performs the same signal quality measurement on the al sophisticated features to enable coexistence between
transmitted downlink signal as the mobile device; thus, LTE and 5G NR using the same spectral band. There-
the results can be correlated and validated. In terms of fore, it is essential to verify all the features and function-
DSS, the scanner needs the capability to identify LTE alities described above, not only in the lab by testing
and NR signals and identify subframes configured as DSS-capable 5G handsets, but also in the network when
MBSFN and carry NR signals. performing drive tests for network optimization. n
OUTLOOK: DSS ENHANCEMENTS IN 3GPP References
RELEASE 16 1. 3GPP TS 38.101 V15.6.0 (2019-06), User Equipment (UE) Radio
Transmission and Reception, Part 1: Range 1 Standalone.
As shown in Figure 4, PDSCH mapping type B, as 2. M. Kottkamp, A. Pandey, D. Raddino, A. Roessler and R. Stuhl-
specified in 3GPP release 15, has the limitation that, fauth, 5G New Radio – Fundamentals, Procedures, Testing As-
at maximum, seven OFDM symbols can be allocated pects, 1st edition, Rohde & Schwarz, 2019.
when this mapping type is configured. In 3GPP release 3. 3GPP TS 38.211 V15.6.0 (2019-06), Physical Channels and Modu-
16, PDSCH mapping type B will be extended so that lation (Release 15).
4. 3GPP TS 38.423 V15.4.0 (2019-07), NG-RAN, Xn Application Pro-
nine and ten symbols can be assigned to use the slot tocol (XnAP) (Release 15).
efficiently.
16
Ericsson Performs 5G Coverage/
Performance Verification Using
R&S Drone-Powered Solution
R
ohde & Schwarz has supplied mobile network surement flights
testing tools used in drone-based network cover- conducted so far
age, performance and operation tests managed have shown the
by Ericsson. The unique procedure enables unprece- solution proce-
dented 3D accessibility, positional accuracy and repeat- dure and results
ability of the testing. It also opens up new possibilities to be extremely
to ensure end user Quality of Service (QoS) for demand- repeatable. The
ing 5G use cases such as industry 4.0, automotive and drone flights were
public safety. of various dura-
The deployment of 5G New Radio (NR) brings new tion, altitudes and
applications of cellular networks for subscribers, govern- routes, depending on the test case. Control, authentica-
ment and industry. It also makes the verification of the tion and air traffic control are considerable challenges
correct coverage, performance and operation of net- to the development of robust drone-based solutions. In
works more critical, increasing the demand for accuracy this new system they are conducted over cellular net-
and accessibility in traditional field network tests. works, eliminating the requirement for line-of-sight con-
A project team based in Jorvas, Finland and led by nection between the drone and its pilot.
Ericsson’s 5G Readiness Program RAN Technical Lead The project is a collaboration between Ericsson, Ro-
Richard Wirén, has developed together with Centria hde & Schwarz, Tampere University and Centria Univer-
University of Applied Sciences a novel system for testing sity of Applied Sciences and forms part of the Business
cellular mobile network coverage. The new system uses Finland 5G FORCE program.
mobile network testing scanners and smartphones from Wirén says, “For 5G to realize its promise, field verifi-
Rohde & Schwarz mounted on a drone that can be pro- cation of operation and quality is essential, and this de-
grammed to execute automatic tests with considerable velopment is a pioneering way to ensure our customers
flexibility, for example for precise route selection and receive the network performance they require. We are
drone speed control. This solution is especially valuable delighted to utilize test solutions from Rohde & Schwarz
for industrial use cases. It also has the advantages over that have proven themselves very reliable and are ex-
traditional walk and drive tests by providing unprec- cited that we now have access to solutions based on
edented repeatability and positional accuracy with the commercially available 5G NR UEs such as the Samsung
ability to verify beamforming and map coverage in 3D. S10 5G.”
The R&S TSMA6 network scanner is mounted on a Vice President of mobile network testing Rohde &
drone and is able to simultaneously verify important LTE Schwarz, Hanspeter Bobst, says, “We are delighted to
and 5G NR coverage metrics such as reference signal combine our industry-leading mobile network testing
received power (RSRP) and signal-to-interference-plus- know-how with Ericsson’s long tradition of network in-
noise ratio (SINR) in accordance with 3GPP standards. novations to ensure the delivery of end-user Quality of
When combined with the R&S QualiPoc Android smart- Experience as 5G NR becomes a reality.”
phone-based optimizer, IP trace, application QoS met- Future developments will focus on testing critical 5G
rics such as serving cell parameters are possible. The applications such as public safety and machine-type
solution currently uses LTE user equipment (UE) but will communications for Industry 4.0, extending the fre-
soon be further developed to include 5G UEs such as quency to extremely high frequencies of the mmWave
the Samsung S10 5G. bands and testing in an urban environment.n
The drone can be programmed to follow an exact
three-dimensional route. The over 20 successful mea-
www.mwjournal.com/articles/32866
17
Overcoming the Challenges of
mmWave, On-Wafer Load-Pull
Measurements for 5G
Richard Hilton and Steve Dudkiewicz
Maury Microwave Corp., Ontario, Calif.
Hybrid-active load-pull overcomes the challenges in mmWave power amplifier design by

removing the uncertainty of unclosed contours to enable designing for peak performance.
F
ifth-generation mobile represents the next evo- and in the 28 to
a1 S21 b2
lution in wireless communications. With an em- 30 and 37 to 39
phasis on connectivity, 5G is expected to bring GHz bands for Port Port
together data, voice, video, IoT, connected cars, high data rates. S11 S22 2
1
smart homes, smart cities, augmented reality and indus- While posing
trial automation. 5G will achieve this aggressive goal by unique challeng- b1 S12 a2
deploying technologies over multiple frequency bands, es, the mmWave s Fig. 2 Two-port S-parameter model
from low MHz to high GHz. Research in the 450 MHz bands promise to of a DUT.
to 6 GHz bands is targeting long-range communication bring many ad-
vantages, including larger bandwidth, greater capacity,
increased security and longer battery life.
A critical enabler in the 5G infrastructure is the pow-
er amplifier (PA), which must be properly designed for
optimum performance, i.e., maximizing power and ef-
POUT ficiency while maintaining appropriate linearity. A useful
Power-Added design tool for maximizing performance is load-pull.
Efficiency
LOAD-PULL TECHNIQUES
Load-pull is the process of changing the load imped-
ΓL (Imaginary)
ance presented to a device under test (DUT), commonly

a transistor, to measure its performance characteristics
under varying large-signal conditions. The impedance is
systematically changed while parameters such as output
power, gain and efficiency are measured or calculated.
Contours representing fixed performance values (e.g., x
dBm output power or y percent efficiency) are then plot-
ted to visualize the point of maximum performance, the
rate at which the performance changes and trade-offs
between various parameters (see Figure 1).
But how does load-pull work? First, consider a DUT
as a two-port network (see Figure 2). A signal a1 is in-
ΓL (Real) jected into port 1 of the DUT. A portion of the signal is
delivered to the DUT while another portion is reflected
s Fig. 1 Output power and power-added efficiency load-pull
as b1, due to the mismatch between the input imped-
contours.
18
ance of the DUT and the source impedance of the input
network. A modified signal b2 exits port 2 of the DUT Load Tuner
and is delivered to the load, while a portion of it is re-
flected back as a2, due to the mismatch between the Bias Tee
output impedance of the DUT and the load impedance
of the output network. The magnitude and phase of that ΓL
reflection, represented as ΓL, is
a (a)
ΓL = 2 (1)
b2
Load-pull changes the magnitude and phase of ΓL by
changing the reflected signal a2. Any load impedance,
which can be calculated as
⎛ 1 + ΓL ⎞
Z = Z0 ⎜ (2)
⎝ 1 − ΓL ⎟⎠
can be presented to the DUT as long as the signal a2 can
be achieved. There are two common methodologies to
vary the impedance presented to the DUT: passive load-
pull and active load-pull.
End View
PASSIVE LOAD-PULL
Passive load-pull uses mechanical impedance tuners
to change the magnitude and phase of the reflected sig-
nal a2 and vary the impedance presented to the DUT
(see Figure 3a). The magnitude and phase of the load
impedance are adjusted by varying the position of a Side View
probe (or slug) in both x and y axes along a 50 Ω airline
(see Figure 3b). The magnitude of the reflection is con-
trolled by moving the probe vertically within the airline,
while phase is controlled by moving the probe horizon-
tally along the airline. By moving the probe up and
down, left and right, it is possible to present nearly any
impedance to the DUT, as long as the magnitude of a2
remains sufficiently large so that the desired
a2
ΓL =
b2
can be achieved. It is important to note that ΓL is less than 1,
since a2 is always smaller than b2 due to losses between the
output of the DUT and the tuner.
(b)
ACTIVE LOAD-PULL
s Fig. 3 Passive tuner for performing load-pull measurements
Open-loop active load-pull (see Figure 4) does not (a) comprising a passive slide screw tuner and probe (b).
rely on a mechanical tuner to reflect part of b2 back as
a2; rather, it uses a signal generator with magnitude and
phase control to create a new signal a2. When amplified
SG
by an external amplifier, any a2 and, hence, any ΓL can Amplifier
be achieved. At first glance, active load-pull may seem Bias Tee
superior to passive load-pull since it has no theoretical ΓL

limitation; however, a practical limitation is the power re-
quired to achieve the signal a2 actually delivered to the ΓL
output of the DUT. Active tuning has several advantages
over passive tuning, including speed, as there are no me- s Fig. 4 Output network of a simple active load-pull setup.
chanical moving parts, and increased Smith chart cover-
age, as a2 is directly generated, enabling the 50 Ω amplifier and the non-50 Ω DUT causes a por-
a2 tion of the signal to be reflected back toward the ampli-
ΓL = fier; the larger the mismatch, the larger the signal that is
b2 reflected. Under extremely mismatched conditions, it is
to be greater than 1. possible that only 10 percent of the signal available will
The limitation is the maximum output power of the actually be delivered to the output of the DUT, requiring
amplifier. Referring to Figure 4, the mismatch between a large amplifier.
19
GaN transistors have output impedances of 1 to 2 Ω,
which can be represented by Γ values from 0.96 and
0.92, respectively. Figure 5 shows actual passive load-
pull measurement data for a GaN transistor on-wafer at
30 GHz with a maximum output power of 30.66 dBm.
Notice how the contours do not close, so it is uncertain
how the transistor would perform if further tuning could
ΓL (Imaginary)
be performed.
Hybrid-active load-pull overcomes this limitation in the
passive load-pull’s measurement range by adding an ac-
tive injection signal to increase a2 and, therefore, increase
Γ. A typical test setup for a hybrid load-pull system is
shown in Figure 6. The relationship between the transistor,
the system impedance, the injection power and the tuning
range is
ZL =
( ZSys + KZDut ) ( )
ΓL (Real) 2 2
Z Sys + K ZDUT − − (1 − K ) Z Sys − KZDUT
2
s Fig. 5 Passive on-wafer load-pull measurement of a GaN (7)

1− K
transistor at 30 GHz.
Hybrid-active load-pull overcomes this limitation where ZL is the impedance presented to the DUT, ZSys is
by pre-matching the DUT impedance from highly mis- the system impedance and ZDUT is the DUT’s output im-
matched to moderately mismatched, lowering the pow- pedance. K is defined as
er required to deliver the same signal a2 to the output
2 2
of the DUT. P 1 − Γ Sys Z Sys + Z 0
K = a2 ⋅ 2 2
(8)
mmWAVE LOAD-PULL Pb2 1 − Γ ZDUT + Z 0
DUT
When performing load-pull, it is preferable to be able
to close the measurement contours to ensure the DUT’s
maximum performance has been achieved. Without where Pa2 is the active tuning power injected into the
closed contours, it is possible for the optimum perfor- output of the DUT at the DUT reference plane, Pb2 is the
mance condition to be missed and the wrong conclu- DUT’s output power and Z0 = 50 Ω. The net reflection
sion formed. achievable at the DUT reference plane is
With a passive load-pull system, the net magnitude
of reflection achievable at the DUT reference plane can Z − 50
ΓL = (9)
be calculated as Z + 50
RL tuner + RL coupler + cable +probe =
RLDUT (3)
where RL is return loss RF Cable

VNA
DC Cable
⎛ VSWR tuner − 1⎞
RL tuner = −20Log ⎜ (4)
⎝ VSWR tuner + 1⎟⎠
RL coupler + cable+probe =
(
2 IL coupler + cable+probe ) (5)
Driver
Rcvr R3
Rcvr C
Rcvr R4
Rcvr D
Active
Amplifier
Amplifier
and where IL is insertion loss.
Source Tuner Load Tuner
⎛ −RLDUT ⎞
⎝⎜ 20 ⎠⎟ Bias Tee Bias Tee
ΓDUT = 10 (6)
Coupler Coupler
Assuming a typical tuner VSWR
and coupler, cable and probe loss-
es at 30 GHz, VSWRtuner = 20:1,
ILcoupler+cable+probe = 2.5 dB, the
maximum achievable magnitude of
reflection is reduced from Γ = 0.9 at Power Supply
the tuner reference plane to Γ = 0.5
at the DUT reference plane. Modern s Fig. 6 mmWave hybrid-active load-pull system (a) and setup (b).
20
With a driver amplifier output of 40 dBm and using the
same passive impedance tuner to transform the system
impedance from 50 Ω to 23.17 + j28.12 Ω, it is possible
to achieve Γ = 0.85 and successfully close the output
power contours. The contours shown in Figure 7 dem-
onstrate that a maximum output power of 31.12 dBm
can be achieved by the same GaN transistor, which is
ΓL (Imaginary)
0.46 dB or approximately 11 percent more power than
initially determined through passive load-pull with in-
complete contours.
CONCLUSION
As companies accelerate development of 5G tech-
nologies and compete for best-in-class solutions, the
optimization of power, efficiency and linearity will be-
come more essential. Small advantages of a few dB
in power or a few percentage points in efficiency may ΓL (Real)
mean the difference between best-in-class and “never
was.” Hybrid-active load-pull helps overcome the chal- s Fig. 7 Hybrid-active on-wafer load-pull measurement of a
lenges in mmWave PA design by removing the uncer- GaN transistor at 30 GHz.
tainty of unclosed contours. This enables ideal matching
and gives those that adopt the methodology an edge in
the marketplace.n
21
Advertorial
TESTING 5G MMWAVE
DEVICES OVER THE AIR Visit Rohde & Schwarz
at MWC2020
Hall 6, Booth 6B30
5G NR networks are currently in the rollout. 5G is the transform a spherical wave front from the DUT or feed
first cellular technology operating at the new cmWave antenna into a planar wave front it is known as a com-
and mmWave spectrum (frequency range 2 or FR2). pact antenna test range (CATR). This reduces the re-
It uses advanced antenna technologies with a higher quired chamber size as well as the complexity and cost
number of antennas for beamforming, which is com- of the test setup.
pensating for the increased path loss at FR2. However,
advanced antennas are integrated units, with trans- Rohde & Schwarz has several OTA solutions for mobile
ceiver frontends combined with the antenna array and device testing that make use of such a CATR setup.
without traditional RF output ports. The lack of physical These include the R&S®ATS800B benchtop antenna
connectors makes over-the-air (OTA) testing the default test system, the R&S®ATS800R rack solution and the
end-user device testing environment in FR2 which ne- R&S®ATS1800C 5G NR mmWave test chamber. All
cessitates chambers and positioning systems to also these solutions feature the high precision CATR reflec-
cover dynamic beam steering scenarios. tor to generate a big usable measurement area (the so
called quiet zone QZ) in a very compact setup and a
Generally, all OTA measurements shall be executed wide frequency range.
under far field (FF) conditions where the electromag-
netic waves have become planar. With high frequen- Get a detailed overall system overview of 5G NR in
cies, this can lead to big FF distances according to the new, free 5G eBook, written by top experts at
the Fraunhofer equation, requiring large shielded Rohde & Schwarz. The chapters include the technical
chambers. Even a 15 cm smartphone transmitting at background of the new communications technology
43.5 GHz would need a testing distance of 6.5 m due to with detailed information on subject matters such as
this. To ensure plane waves, either the distance of the use cases, architecture, access and non-access proto-
DUT from the transmitting or receiving antenna has to cols, quality of service and network slicing.
be large enough, or the far field conditions have to be
created artificially in a smaller distance by converting The eBook ‘5G New Radio - Fundamentals,
near field (NF) into FF, which is also referred to as indi- procedures, testing aspects’ is available at
rect far field (IFF).
www.rohde-schwarz.com/5G-ebook
Using an anechoic chamber for radiation shielding and
a positioning system for the DUT together with a hard-
ware transformation of the propagating waves makes
up an IFF system. If a parabolic reflector is used to
CATR antenna test

systems are convenient
and accurate solutions
for testing 5G mmWave
antennas, modules and
devices in a compact
setup.
Optimizing the Perennial
Doherty Power Amplifier
Gareth Lloyd
T
he Doherty power amplifier (PA), invented al-
most 100 years ago, is used in an increasing Carrier
PA
number of radio transmitter applications to im-
Baseband
prove energy efficiency, with numerous ways +
Doherty
to build the PA. This article begins with an overview of DAC
Combiner
+
linearization and efficiency enhancement and, against Modulator
that backdrop, highlights the associated challenges and Peaking
some of the numerous solutions. Finally, there is an al- PA
ternative design flow, illustrated with a case study pro-
viding insight into the design and how to achieve the Dual Input
best performance-cost compromise. Doherty
Classic Programmable
LINEARIZATION TECHNIQUES Doherty Split Doherty
The four key technical performance parameters in a
transmit (Tx) RF front-end (RFFE) are the efficiency, out- Bias Modulated Doherty
Doherty Outphasing
put power, linearity and bandwidth. The latter three are Continuum
ET +
often dictated by system requirements, such as a com- Doherty Doherty
munications standard. The former, (energy) efficiency, is Envelope Outphasing
Continuum Outphasing
the differentiator. All other performance parameters be- Tracking
+ ET
ing equal, a higher efficiency for a front-end is preferred. Load Multilevel
Devices used in the RFFE have imperfect linearity Modulation Outphasing
Chireix
characteristics, preventing them from being fully uti-
lized merely as drop-in components. The linearity of ER/EER LINC
a Tx RFFE can be improved by implementing a lin-
earization scheme. Typically, this will increase the raw
cost of a Tx RFFE, trading that for a combination of Envelope Efficient
PA RF PA
efficiency, linearity and output power improvement.
Baseband Baseband
Numerous linearization methods have been pub- + + Outphasing,
lished, stretching back at least to the feedforward1 DAC DAC Chireix,
+ +
and feedback2 patents. Arguably, the use of nonlinear Modulator Modulator
Isolated
predistortion dates similarly to the invention of com-
Efficient
panding.3 These schemes may be classified according RF PA RF PA
to their modus operandi (see Figure 1 and Table 1).4
One way of dividing the linearization pie is to identify s Fig. 1 Amplifier linearization options using post-source,
whether a scheme predicts or extracts its unwanted predicted/synthesized composition schemes.
23
and documented over the last 100 years. Outphasing,5
TABLE 1 envelope6 and Doherty7 transmitters, along with their
AMPLIFIER LINEARIZATION METHODS hybrids by Choi,8 Andersson9 and Chung10 are exam-
Impediment Generation ples of such techniques, except they have been primar-
Predicted/ Measured/
ily marketed for efficiency enhancement rather than as
Synthesized Extracted linearization techniques. In their purest forms, envelope
and outphasing schemes construct their signals from ef-
Digital
Predistortion
Cartesian Feedback ficiently generated, nonlinear components, using mul-
Pre-
Source
tiplication and summing of their paths, respectively. A
Analog Doherty comprises a reference path, referred to as the
Polar Feedback
Correction Predistortion
“main” or “carrier,” and an efficiency path, named the
Location Analog Post-
Feedforward “peaking” or “auxiliary.” A more comprehensive math-
Post- Distortion
ematical analysis of the Doherty design is beyond the
Source Composition Fixed Filtering scope of this article and is available in a plurality of texts.
Schemes (e.g., Bandpass) For further information, the reader is especially referred
signal and whether that unwanted correction is ap- to Cripps.11
plied before or after its creation. Classification is use-
DOHERTY IMPLEMENTATIONS
ful to understand the general properties and identify
the best approach for the application. Arguably, the most common and often quickest start-
Feedforward is an example of a measured, post-cor- ing point for a Doherty amplifier design is the “zeroth
rection scheme; feedback is a measured, pre-correction embodiment” (see Figure 2), comprising a
scheme; and predistortion is a predicted, pre-correction • Fixed RF input to the final stage power splitter.
scheme. Predictive schemes rely on the unwanted signal • Main and auxiliary amplifiers, differently biased (e.g.,
being generated, which can potentially be onerous in using class AB and class C).
wider band and lower power systems for digital predis- • Doherty combiner made from a quarter-wavelength
tortion (DPD). On the other hand, predictive schemes transmission line.
do not require that distortion exists and can, potentially, In most applications, this architecture does not pro-
eliminate distortion completely. vide sufficient power gain—at least not from a single,
Missing from these examples is a whole class of linear- final stage—and additional gain stages are cascaded
ization techniques using predictive post-correction. This ahead of the power splitter. Criticism of this most com-
family of techniques has also been heavily researched monly used implementation include
• No method for compensating gain and phase varia-
tions in any domain after the design is frozen.
Class AB
• Both the efficiency and output power are traded-off
because of the bias class. In effect, the class C bias,
an open loop analog circuit, is driving this.
• Efficiency enhancement is limited to a single stage.
With a multistage cascade, this limits the perfor-
Class AB mance improvement, especially as gain diminishes at
higher frequencies.
Class C From another perspective, the Doherty engine is an
open loop scheme, with several key functional mecha-
s Fig. 2 Simplest implementation of the Doherty amplifier.
nisms derived from the bias points of the transistors.
0 100
Imain IAUX
Doherty –0.5
Relative Output Power (dB)
Main Combiner 80
PA
–1.0
Efficiency (%)
1.0 –1.5
Balanced 60
Auxiliary Output Current
Square Law –2.0

0.8
Ideal 40
–2.5
0.6
–3.0
Auxiliary Main 20
Auxiliary 0.4
–3.5
PA
0.2 –4.0 0
0 1 2 3 4 5 6
0 Conduction Angle (Rad)
0 0.2 0.4 0.6 0.8 1.0
Input Voltage
(a) (b) (c)
s Fig. 3 Doherty amplifier challenges: combiner amplitude and phase matching (a), auxiliary amplifier current response (b) and
power-efficiency trade-off (c).
24
• Multiple gain stages inside the Doherty splitter and
Class AB
combiner.
Class Opt
• N-way Doherty.
• Intentionally dispersive splitter.
• Programmable splitter.
• Bias modulation.
• Supply modulation, i.e., adding a third efficiency
enhancement technique to the two leveraged by
Doherty.
• Envelope shaping.
s Fig. 4 Digital Doherty amplifier, where the main and auxiliary • Digital Doherty.
amplifier operating class is digitally controlled.
In addition to the different architectures available to
Once the other variables are defined (e.g., phase off- the designer, three points in the product life cycle allow
sets, splitter design, etc.), only one or two handles are adjustments. During the design phase, the design pa-
provided, upon which multiple critical adjustments rely. rameters can be modified, recognizing the parameters
will be passed to production as fixed values (e.g., the in-
Challenges put splitter design). During production, the parameters
One of the ways the Doherty improves efficiency is may be modified or tuned, typically based on measured
load modulation. The engine that drives that is the dif- data, and then frozen or fixed through programming.
ference in output currents, sourced into the combiner One example is the nominal bias voltage used to gen-
from two or more amplifiers. Since the engine can only erate the target bias current in the devices. Once the
approximate the Doherty operation, the challenge for equipment is deployed in the field, parameters may be
the designer is to enable the engine to approximate it updated, either continuously or at specific times, either
with the best, but still appropriate, cost-performance open or closed loop. Open loop concepts rely on suf-
paradigm. Some of the potential hindrances or impedi- ficiently predictable behaviors, while closed loop con-
ments to Doherty performance are 1) the amplitude and cepts might require built-in measurement and control.
phase matching of the signals incident to the combin- One example is circuitry for temperature compensation.
ing node, especially over frequency (see Figure 3a). De- These product life cycle options provide a plurality of so-
viation from the ideal degrades efficiency and output lutions with no “best” solution. It is just as important for
power. Potentially, this can be more destructive, as the the designer to be aware of the manufacturing and sup-
devices are intentionally not isolated, with the efficiency ply capabilities following the design as the design chal-
enhancement relying on their mutual interaction through lenges and trade-offs made during the design phase.
the combiner. 2) Ideally, the auxiliary path of the Doherty At the opposite end of the solution spectrum from
engine exhibits a dog leg or hockey stick characteristic the zeroth embodiment is the digital Doherty (see Fig-
(see Figure 3b). Failure to achieve the ideal is often the ure 4). This architecture is characterized by an input split
primary reason for not realizing the famous efficiency which stretches back into the digital domain, prior to the
saddle point. As the characteristic tends from the ideal digital-to-analog conversion. The ability to apply digital
to a linear response, the Doherty amplifier increasingly signal processing to the signal applied to both amplifier
behaves like its quadrature-balanced relative—albeit paths potentially gives unsurpassed performance from a
with a non-isolated combiner—especially its efficiency set of RF hardware. Compared to the standard Doherty
performance. 3) The commonly used “differential bias- implementation, the digital version can achieve 60 per-
ing” of the main and auxiliary operating in class AB and cent greater output power, 20 percent more efficiency
class C, respectively, forces the output power and ef- and 50 percent more bandwidth without degrading pre-
ficiency of both amplifiers to be degraded (see Figure dictive, pre-correction linearity.12
3c). As Cripps showed,11 the continuum of quasi-linear
amplifier classes from A to C, which theoretically oper- MEASUREMENT-AIDED DESIGN FLOW
ate with sinusoidal voltages across their sources, varies To optimize any Doherty design, it is advisable to
their respective maximum output power and efficiency build simulation environments that correlate well with
characteristics. At the same time, if biasing is used to the design, to understand trends and sensitivities. The
create the difference engine, as is the case in the classi- simulation enables a significant part of the development
cal Doherty embodiment, there is intrinsically a trade-off to be covered quickly. Inputs to the first step might in-
between output power and efficiency. Simultaneously, clude load-pull data or models for the candidate devic-
differential biasing increases the Doherty effect, yet de- es, a theoretical study of the combiner and matching
creases the achievable performance. network responses, evaluation boards with measured
data or other empirical data. Building on this starting
VARIANTS AND IMPROVEMENTS point, the design flow can be supplemented with mea-
The following variations on the basic concept may surement-aided design (see Figure 5).
be more appropriate for some applications and, with For the digital Doherty, the starting point for this ap-
the classical implementation, offer the designer perfor- proach is a Doherty comprising two input ports, input
mance and flexibility options. and output matching networks, active devices, bias net-
25
works and the Doherty combiner (see Figure 6). Mea- The measurement algorithm may be rapid or more
suring the prototype Doherty as a dual-input device exhaustive, programmed to seek the optimum values
provides greater insight into the performance limita- for desired parameters or configured to characterize
tions, trade-offs and reproducibility expected in a pro- a wide range of parameters. In a simple case, the de-
duction environment. Critical to the test set-up are two signer may want to confirm the best-case quantities and
signal paths, whose signals may be varied relative to their relative amplitude and phase balance values. More
each other. In addition to applying precise, stable and complicated, a detailed sweep to enable a sensitivity
repeatable amplitude and phase offsets to the signals, it analysis or rigorous solution space search may be war-
is advantageous to be able to apply nonlinear shaping
to at least one of the signal paths.
45
3.60
42
44
40
Frequency (GHz)
3.55
44
30
32
42
34
40
36
38
Simulation Cut
and Try
3.50 35
Empirical Eval Board 3.45

Device Model Import 30
3.40
40
42
44
2 25
Something 0
0
46
40
Load-Pull –50
44
Else
42
Amplitude –2 –100 Phase
Difference (dB) –150 Difference (º) 20
Prototype Output Side (a)
Test as Dual-Input
45.0
3.60
.2
.4
43
43
5 8
444 444.
.6
44
Choose Optimum Architecture
44
4. 44.5
Frequency (GHz)
3.55 2 44
Design Review 44.0

3.50
Specification 43.5
3.45
45
45.2
43.0
45.2
Production 3.40
42.8
45
45
45.2
2 42.5
0
44.8
45
s Fig. 5 Measurement-aided design flow for a digital Doherty 0 –50

amplifier. Amplitude –2 –100 Phase 42.0
Difference (dB) –150 Difference (º)
(b)
DSP DAC Up-Converter Main Doherty Drain Efficiency (%) PSAT (W)
Unit PA Combiner 3 3
34
33
36
30
32
31
38
28
2 2
Amplitude Difference (dB)
30
30
1 1
32 34
29
DAC Up-Converter Auxiliary

40
32
0 0
36
(a) PA
43
28
34
31
36
30
41
–1 –1
38
27
42
–2 –2
44
26
38 40
41 42
43 44
25
45
24
212
23
2
Doherty –3 –3
250 300 350 250 300 350
DUT
Phase Difference (º) Phase Difference (º)
(b) (c)
s Fig. 7 Dual-input Doherty in linear operation: measured

s Fig. 6 Simplified block diagram (a) and hardware setup (b) efficiency at 35.5 dBm (a), saturated power (b) and worst-case
for designing a digital Doherty amplifier. efficiency and power (c).
26
34
Drain Efficiency (%)
3 3 33
40
Saturated Power (W)

37
35
31
33
32
42
2 2
43
41 31
1 1
30
42
0 0 29 Low Frequency - Volume
44
–1 28 Low Frequency - Nominal

–1
High Frequency - Production
43
–2 27
–2 High Frequency - Nominal
44
45
40 26
43
45
36
46
44
–3 –3 40 41 42 43 44 45 46
250 300 350 250 300 350
Phase Difference (º) Drain Efficiency (%)
(b)
PSAT (W)
3 3 34
31
33
29
32
27
30
33
31
33
2 2
Saturated Power (W)

32
1 1
29 30
31
0 0 30
–1 –1 28 29
27 28
–2 –2
28
33
32
24
26 27
24
23
–3 –3 25
250 300 350 250 300 350 26
40 41 42 43 44 45 46
(a) Phase Difference (º)
(c) Drain Efficiency (%)
s Fig. 8 Gain and phase variation of a population of split digital Doherty amplifiers with a fixed RF input (a), saturated power and
efficiency using a look-up concept (b) and cumulative, worst-case production distribution (c).
ranted. The post-processing of these measurements can the bulk effect of these part-to-part variations can be ob-
be as simple or sophisticated as the user wishes. served, as shown in Figure 8. Figure 8a shows the drain
efficiency and saturated output power at two frequen-
CASE STUDY cies, Figure 8b shows the estimated production spread
To demonstrate the design flow and achievable re- of saturated output power and drain efficiency versus
sults, a digital Doherty PA for a 3.5 GHz, 5G New Radio the nominal values for the same two frequencies. Figure
(NR) base station was designed using a single stage un- 8c shows the cumulative production spread, aggregat-
matched GaN power transistor, the Qorvo® TQP0103. A ing the results from the two frequencies. Paradoxically,
dual-path R&S®SMW200A vector signal generator pro- in this case, most of the part-to-part variation is in the
vided the two input signals to drive the GaN amplifier. target variable, efficiency.
For measurement of dependent quantities, the single RF By adopting an alternative approach to the input
output of the amplifier was connected to an R&S®FSW splitter design, this variation can be reduced. Using a
Signal Analyzer. DC power for the devices was sourced dispersive input splitter design, meaning using different
from an R&S®HMP power supply, which measured the amplitude and phase differences at the two design fre-
DC power consumption. The amplifier was stimulated quencies, advantageously enables the stacked contour
using differentially linear and nonlinear signals, the for- plots shown in Figure 8a to, in effect, slide over one an-
mer sweeping the input power, amplitude and phase. other. Using the same part-to-part variation data with
The nonlinear tests used a variable shaping function, this dispersive splitter design yields a better result (see
amplitude dependent, at two frequencies. Output pow- Figure 9), with a higher mean efficiency and lower stan-
er, output peak-to-average power ratio, adjacent chan- dard deviation.
nel leakage ratio (ACLR) and current consumption were By directly generating signals for the two ampli-
measured, and the measurement results were analyzed fier inputs in the digital domain, the deficiencies of the
using MATLAB®.13 Doherty amplifier are significantly reduced. Addition-
Analyzing the linear measurements, efficiency at a ally, the simple part-to-part amplitude/phase variations
specified power level and saturated power were plotted shown in the linear example may be eliminated. To illus-
versus the amplitude and phase differences (see Fig- trate this, albeit not exhaustively, the auxiliary path was
ure 7), with the worst-case efficiency and output power programmed with a square law shaping function applied
shown in Figure 7c. In the basic Doherty embodiment, to both the amplitude and phase, with the phase “start”
a quasi-constant amplitude/phase split is chosen for and “end” values—the phase with zero and maximum
the operating frequency. The efficiency and saturated input amplitude—varied randomly. With a common bias
power for these amplitude/phase values can be deter- for the two amplifiers, only a trade-off between output
mined by extracting the worst-case performance at the power and efficiency remains, rather than those and the
test frequencies. Doherty difference engine magnitude.
Selecting a nominal amplitude/phase split, a pertur- To establish a baseline, driving the commonly biased
bation representing the natural variation in production amplifiers with a linearly differential signal enabled the
may be added to the evaluation. Using a look-up table, equivalent “balanced” performance to be ascertained:
27
34
Drain Efficiency (%) Low Frequency - Volume
3 33
3 Low Frequency - Nominal
Saturated Power (W)

40 32

42
37
35
31
33
2 2 High Frequency - Production
43
41 31 High Frequency - Nominal

1 1 30
42
0 0 29
44
–1 28
–1
43
27
–2 –2
44
26
40
43
36
45
46
45
44
–3 –3 40 41 42 43 44 45 46
250 300 350 250 300 350 Efficiency (%)
Phase Difference (º) (b)
PSAT (W)
3 3 34
31
29
31
33
32
27
30
33
2 2 33
Saturated Power (W)

1 32
1
29 30
31
0 0
30
–1 –1 28 29
–2 –2 27 28
26
33
32
28
24
24
23
25 27
–3 –3
250 300 350 250 300 350 26
Phase Difference (º) 40 41 42 43 44 45 46
(a) (c) Drain Efficiency (%)
s Fig. 9 Digital Doherty amplifier population using a dispersive input split: gain and phase variation (a), saturated power and
efficiency (b) and cumulative, worst-case production distribution (c).
be used to achieve the original target
46.0 output power. Taking into account the
55 Improved
45.5 expected part-to-part variation, this
3.4 GHz
3.6 GHz 45.0 reduction in device periphery might
50 be reduced further.
Efficiency (%)
Improved 44.5
PEP (dBm)
45 CONCLUSION 44.0
Reference
43.5 Significant improvements in Doherty
Reference
40 performance can be achieved by ad-
43.0
dressing the input side of the design.
3.4 GHz
35 42.5
3.6 GHz The use of either an intentionally dis-
34.0 34.5 35.0 35.5 36.0 36.5 37.0 37.5 persive or programmable input split
42.0
PAvg (dBm) 34.0 34.5 35.0 35.5 36.0 36.5 37.0 37.5
(a) (b)
can improve performance, especially
PAvg (dBm)
considering manufacturing distribu-
s Fig. 10 Efficiency vs. average output power (a) and PEP vs. average output power tions. According to peer reviewed re-
(b) for a dual-input Doherty amplifier using with square-law shaping and randomized search,12 the digital Doherty with non-
phase. linear input splitting or shaping can
the available saturated output power in this mode was achieve 60 percent more output pow-
0.5 dB higher than the differential biased case (12 per- er, 20 percent more efficiency and 50 percent greater
cent higher power). That represents the “cost” of oper- bandwidth without any degradation in predictive linear-
ating the Doherty engine using differential bias points. ization. The case study described in this article achieved
The scatter plot of random shaping functions applied to 47 percent higher output power and 11 percent greater
the auxiliary path yields the locus of performance shown efficiency over a fixed bandwidth.
in Figure 10, reflecting the distributions of average A measurement-aided methodology for extracting
power versus efficiency and peak envelope power (PEP) and understanding possible improvements was dem-
versus average power. The saturated output power is onstrated. While efficiency and saturated power served
1.7 dB higher than the conventional Doherty amplifier as examples, they do represent the two most impor-
(48 percent higher power), suggesting that 1.2 dB of tant parameters in most Doherty designs. Regardless of
the improvement (32 percent) is from better amplitude/ which Doherty architecture is used, this design method-
phase matching of the signal paths. ology provides more detailed and rigorous insight and
The 1.7 dB improvement in saturated output means improves both time-to-market and the cost-specification
the amplifier may be operated at that increased out- paradigm.n
put power without compromising headroom, and the
increase in average power is associated with a 5 point ACKNOWLEDGMENTS
increase in efficiency (from 44 to 49 percent). Alterna- The author would like to express gratitude to Jeff Gen-
tively, devices with 48 percent smaller periphery may gler, Tammy Ho Whitney and Bror Peterson at Qorvo.
28
References 8. J. Choi et al., “Optimized Envelope Tracking Operation of Doherty
1. H. S. Black, “Translating System,” U.S. Patent 1,686,792, October Power Amplifier for High Efficiency over an Extended Dynamic
9, 1928. Range,” IEEE Transactions on Microwave Theory and Techniques,
2. H. S. Black, “Wave Translation System,” U.S. Patent 2,102,671, De- Vol. 57, No. 6, June 2009, pp. 1508–1515.
cember 21, 1937. 9. C. M. Andersson et al., “A 1 to 3 GHz Digitally Controlled Dual-
3. A. B. Clark, “Electrical Picture Transmitting System,” U.S. Patent RF Input Power Amplifier Design Based on a Doherty-Outphasing
1,619,147, November 13, 1928. Continuum Analysis,” IEEE Transactions on Microwave Theory and
4. P. G. Lloyd, “Linearization of RF Front-End,” Rohde & Schwarz Techniques, Vol. 61 No. 10, October 2013, pp. 3743–3752.
GmbH & Co., November 2016, www.rohde-schwarz.com/ 10. S. Chung et al., “Asymmetric Multilevel Outphasing Architecture
appnote/1MA269. for Multi-Standard Transmitters,” RFIC 2009.
5. H. Chireix, “High-Power Outphasing Modulation,” Proceedings of 11. S. C. Cripps, “RF Power Amplifiers for Wireless Communications,”
the Institute of Radio Engineers, Vol. 23, No. 11, November 1935, Artech House, Norwood, Mass., 2006.
pp. 1370–1392. 12. Darraji et. al, “Doherty Goes Digital,” IEEE Microwave Magazine,
6. R. L. Kahn, “Single-Sideband Transmission by Envelope Elimination September 2016.
and Restoration,” Proceedings of the Institute of Radio Engineers, 13. “The Dual-Input Doherty,” Rohde & Schwarz, www.rohde-schwarz.
Vol. 40, No. 7, July 1952. com/us/campaign/premium-download-the-dual-input-doherty/
7. W. H. Doherty, “A New High Efficiency Power Amplifier for Modu- premium-download-the-dual-input-doherty_233590.html.
lated Waves,” Proceedings of the Institute of Radio Engineers, Vol.
24, No. 9, September 1936, pp. 1163–1182.
29
One Box Test Solutions for 5G
Rohde & Schwarz
Munich, Germany
T
esting 5G NR products requires new test &
measurement approaches. massive MIMO
(mMIMO), bandwidths and data rates substan-
tially higher than 4G’s and mmWave frequen-
cies require more computing power, make over-the-air
(OTA) measurements compulsory and create challeng-
ing performance requirements for the RF hardware. With
a unified testing environment, it is possible to make 5G
measurements in development, acceptance testing and
production as effectively as with 4G testing.
COMPACT TEST SOLUTION FOR 5G mmWAVE

For 5G NR testing in the mmWave bands (designated
FR2), Rohde & Schwarz offers the R&S CMPQ, a fully in-
tegrated, compact solution based on the R&S CMP200
radio communications tester combined with the R&S SIGNALING FOR NETWORK SIMULATION
CMPHEAD30 remote radio head and the R&S CMQ200 The R&S CMX500 radio communications tester adds
shielding cube. The shielding cube uses a drawer con- 5G NR signaling tests to the existing LTE test & mea-
cept enabling automated handling in a manufacturing surement solution. Users who have already invested in
environment for the frequencies from 20 to 77 GHz. The the R&S CMW500 or the R&S CMWflexx system for LTE
mechanical design ensures millions of test cycles, ensur- or legacy 2G and 3G can continue using it, adding the
ing a reliable process for mass production. R&S CMX500 as an extension box. From R&D to certifi-
The R&S CMPQ system can be customized to cre- cation testing, this will perform 5G NR RF, signaling pro-
ate solutions suited to individual needs. It can be used tocol and data throughput and application tests (e.g.,
at any time in the product life cycle, from research and IMS, VoLTE and VoNR). This allows testing 5G NR use
development through validation and quality assurance cases in non-standalone (NSA, option 3) and standalone
to production. The system is created as a robust, flex- (SA, option 2) modes in both the FR1 and FR2 bands.
ible solution that will adapt to future requirements and For pure 5G NR test environments in FR2, a single R&S
specifications. The R&S CMPQ solution provides users CMX500 radio communications tester with remote radio
with the tools needed for production testing of FR2 5G head and a shielded environment will suffice.
components. For the sub-6 GHz (FR1) bands, the so-
lution consists of the R&S CMW100 communications ONE PLATFORM WITH UNIFIED SOFTWARE
manufacturing test set and corresponding measurement The governing principle of the one platform strategy
software, the R&S CMsquares. is using the same technology, similar hardware and the
30
same software for all test solutions
(see Figure 1). This makes the test re- Signaling Sub-6 GHz + IF + mmWave Non-Signaling IF + mmWave
sults comparable in all stages of test-
ing, from R&D to production. The dif-
Technology Reuse
ferent testing approaches (signaling or ❙ Same Test Concept
non-signaling) must deliver reproduc- ❙ Result Traceability
R&S
ible and validated test results, provid- CMXHEAD30 CMX500
®
R&S ®
❙ Synergy Effects R&S® R&S®
CMPHEAD30 CMP200
ing conclusive information about the
characteristics of the devices being
tested, without also testing the test
Identical
solution. With the R&S CMPQ solu- Measurements
tion, FR2 OTA testing is as reliable as
FR1 testing.
R&S CMsquares is a new, unified
test software solution featuring a s Fig. 1 The one platform strategy enables the same measurements for FR1 and FR2,
browser-based user experience and whether testing signaling or non-signaling.
combining everything needed for 5G
NR testing. CMsquares controls all
new 5G radio communications testers
via a standardized graphical interface,
integrating test configuration, param-
eterization, measurements and test
execution in a single environment
with a dashboard and quick access
to the various applications (see Fig-
ure 2). The interactive callbox mode
rapidly connects to a device under
test, changes network parameters on
the fly, analyzes real time RF Tx and
Rx measurements and trace protocol
stack messages on all protocol lay-
ers and generates statistics for data
throughput testing, offering various
chart diagrams. It also integrates a se-
quencer mode to run preconfigured s Fig. 2 The test software R&S CMsquares controls all 5G radio communications
testers, unifying 5G NR testing.
5G NR test scripts or it can create 5G
NR test scripts from scratch using a
simple drag-and-drop approach. The challenges associated with 5G NR require high
An interactive mode and a sequencer mode can be flexibility, end-to-end data testing solutions and reli-
used in parallel, working together to provide the same able measurement methods. As a long-term partner
test results to the user, whether using the R&S CMP200 of the mobile radio communications industry, Rohde &
radio communications tester for non-signaling tests or Schwarz offers a comprehensive portfolio of innovative
the R&S CMX500 radio communications tester for 5G 5G NR test solutions.n
NR network emulation.
31
Software and Hardware
Near-Field Transformations
for 5G OTA Testing
Benoît Derat, Corbett Rowell, Adam Tankielun and Sebastian Schmitz
Increased capacity in 5G mobile communications requires rolling out massive MIMO base
stations with network and mobile terminals at both sub-6 GHz and mmWave frequencies.
Dynamic beamforming and the absence of RF test ports on the devices being tested make
over-the-air (OTA) measurement pivotal to 5G deployment. Fortunately, OTA testing solutions
employing software and hardware near-field transformations are meeting the challenges.
5
G new radio (NR) communication systems will bining modem, RF front-end and antenna. The chal-
increase the capacity of mobile radio networks lenge is to define new methods and setups for perfor-
using frequency bands in the sub-6 GHz region, mance evaluation, as RF test ports tend to disappear
called frequency range 1 (FR1) by 3GPP, and the and beam steering technologies require system-level
mmWave range (FR2). New technological approaches testing. In this context, both antenna and transceiver
selected by the industry and 3GPP promise greater performance criteria must be measured OTA: effective
bandwidth at lower operational expense. isotropic radiated power (EIRP), total radiated power
In FR1, the main innovation effort is focused on the (TRP), effective isotropic sensitivity (EIS), total isotro-
base station, with the enabling of massive MIMO tech- pic sensitivity (TIS), error vector magnitude (EVM), ad-
niques.1 4G systems use single-user MIMO, where the jacent channel leakage ratio (ACLR) and spectrum
user equipment (UE) calculates the inverse channel emission mask (SEM). Assessing these OTA raises the
matrix to extract separate data streams. 5G multi-user critical question of the required measurement dis-
MIMO (MU-MIMO) shifts the complexity from UEs to the tance. Antenna characteristics are usually measured
base station by using a pre-coding matrix. Here, each in the far field. Using direct far-field probing and ap-
data stream is received independently by separate re- plying the Fraunhofer distance criterion (R = 2D2/λ), a
ceivers. Beamforming with antenna arrays of 64 to 512 75 cm massive MIMO device under test (DUT) radiating
elements reduces interference to adjacent users using at 2.4 GHz should be evaluated in a chamber with at
MU-MIMO. In addition to facilitating the adoption of least 9 m range length. Even a 15 cm smartphone trans-
MU-MIMO to increase capacity, beamforming has other mitting at 43.5 GHz needs a 6.5 m testing distance. This
advantages. Its lower energy consumption brings a re- distance is required to create a region encompassing
duction in overall network operating costs by targeting the DUT where the impinging field is as uniform as pos-
individual UEs with their assigned signals. sible and approaches a plane wave with phase deviation
Communication systems in the FR2 range use large below 22.5 degrees, known as the quiet zone.
available bandwidths at frequencies around 28 and 39 Research shows that actual far-field behavior in the
GHz. The impact is more than 60 dB path loss at 1 m peak directivity region can start much closer than the
distance and large electromagnetic field absorption in Fraunhofer distance.2 These results proved, for exam-
nearby objects. As with FR1 systems, the solution is to ple, that the far-field EIRP or EIS of a 15 cm DUT radiat-
employ antenna arrays and beam steering, improving ing at 24 GHz can be assessed at a distance as short as
the gain on both the mobile device and base station 1.14 m. Distance reduction of about 70 percent comes
sides of the network. at the price of increased longitudinal taper error, caused
Whether for FR1 or FR2, 5G deployment relies on by the deviation of the apparent phase center from the
the performance of highly integrated solutions com- center of the measurement coordinate system. Also,
32
Planar Wavefront
Spherical Wavefront
s Fig. 2 Compact antenna test range with a roll-edge reflector

collimating a spherical wavefront into a planar wavefront.
ward larger distances and extract far-field radiation

components. From the Huygens principle, the knowl-
edge of two phasors is enough to reconstruct exactly
all six field components outside the surface. Alternative
transformation methods use spherical wave expansion,
plane wave expansion or integral equation resolution,
with techniques to improve computational efficiency or
accuracy by taking parameters such as spatial sampling
rate, scanning area or truncation into account.
Figure 1 shows a commercial system capable of
both direct far-field and near-field measurements with
spherical scanning around the DUT using a conical cut
positioner. On this system, the DUT is positioned on a
turntable rotating in azimuth, while a dual-polarized Viv-
aldi antenna is mounted at the tip of a boom rotating
in elevation. An RF test port available at the DUT con-
nects one port of a vector network analyzer (VNA); the
measurement antenna ports connect to two other termi-
s Fig. 1 Spherical measurement system (ATS 1000), capable of
nals of the VNA, enabling near-field assessment through
near-field software transformation, measuring a 28 GHz array.
measurements of complex S-parameters.
sidelobe levels cannot be evaluated accurately at short- Near-field measurement methods often rely on un-
er distances.3 While direct far-field measurements at derlying assumptions about passive or RF-fed antenna
shorter distances are not convenient for all applications, testing:
there is an incentive to do so when conditions of appli- • The antenna feed port is accessible with a signal fed
cation are verified. This is because large OTA anechoic to the antenna that is used as a phase reference.
chambers have high costs of ownership and limited dy- • The RF signal is a continuous wave signal.
namic range. Typical applications may be in the “white • Reciprocity applies so that transmit (Tx) and receive
box” case, where the antenna location within the device (Rx) patterns at the same frequency are identical.
and its aperture size are known. There are workarounds available in Tx cases where
such assumptions do not apply. For example, techniques
NEAR-FIELD TO FAR-FIELD
can address the case of a DUT transmitting a modulated
Direct far-field measurements under “white box” as- signal with no access to the antenna feed port. Hard-
sumptions may be inappropriate when the radiation ap- ware and processing implementations to retrieve the
erture is larger than the quiet zone, the antenna cannot propagation phase vary, for example using interfero-
be precisely identified within the DUT or multiple an- metric techniques or multi-port phase coherent receiv-
tennas transmit simultaneously, e.g., from two extreme ers4 with the addition of a dedicated phase reference
edges of a DUT which does not fit within the quiet zone. antenna. For systems like those in Figure 1, this antenna
The “black box” scenario must then be considered, is typically attached to the azimuth turntable. Alternative
where the radiating currents can flow anywhere within approaches include phaseless methods when the phase
the DUT. A first efficient approach to treat such cases in information is retrieved from magnitude measurements.
a compact environment is to employ software near-field However, the Rx mode is more complex. First, the
to far-field transformations (NF-FF), for which the quiet reciprocity assumption does not apply to mobile phone
zone size question becomes irrelevant. Mathematical and base station devices, as the Rx RF component chain
implementations of NF-FF may vary, but the concept is, in general, different from the Tx RF chain. For a DUT
is generally the same: at least two polarization compo- with no test port, the power available at the Rx input
nents of the electromagnetic field (E, H or a mixture of of the RF front-end generated by an impinging wave
the two) are measured in magnitude and phase over a coming from the probe antenna (here used as the trans-
surface encompassing the DUT. The measured data is mitter) cannot be straightforwardly predicted in the near
processed using functions to propagate the fields to- field. In other words, it is not possible to isolate the in-
33
The first difficulty is that these quantities depend
30 strongly on the signal-to-noise ratio (SNR) at the receiver
(a spectrum analyzer in the Tx mode or the DUT in the
27 Rx mode). This can be overcome by first assessing the
complete 3D Tx or Rx pattern to determine the peak
24 direction. Demodulation and EVM or other measure-
ments can then be conducted at this specific location.
21 –1 The question remains whether the obtained values are
–2 reliable and reflect the results obtained in the far field.
18
In the case of a single transceiver, the near-field EVM
15
must be the same as the far-field EVM so long as the
SNR is above a certain threshold dependent on modu-
12 lation scheme, e.g., better than 20 dB. For multiple in-
dependent transceivers operating simultaneously, the
9 near-field EVM may not be straightforwardly related to
–3 the far-field EVM because of positional dependence of
6 the noise figure in the near field.
3 HARDWARE NEAR-FIELD TRANSFORMATIONS

–4
–5 –6 Alternative testing methods enable OTA assessment
0 in the near field without a software transformation, rath-
0 3 6 9 12 15 18 21 24 27 30
er a hardware-based one. The idea is to physically cre-
50 cm Range Length Spherical System ate far-field conditions in a specified quiet zone region
1.5 m Range Length Spherical System within a short range. This is known as “indirect far field.”
42 cm x 42 cm Reflector CATR 2 dB
A compact antenna test range (CATR) uses a mirror to
transform a spherical wave into a planar wave and vice-
s Fig. 328 GHz, 2 dB amplitude taper quiet zones for a roll-
versa. Using Fermat’s principle of least time, a planar
edge compact range system vs. direct far-field systems.
wave can be focused on a single point using a parabolic
mirror. If a measurement antenna is placed at this focal
point, using the reciprocity principle, a plane wave can
be generated as the parabolic mirror reflects a certain
planar component of the incoming spherical wave from
the measurement (or feed) antenna into the quiet zone
where the DUT is placed (see Figure 2).
The error inside a CATR system comes from two
main sources: the mirror geometry—the edge treat-
ment and smoothness of the surface which limit the fre-
quency range—and the feed antenna characteristics. If
the reflector is built with a simple parabolic section, the
sharp edges cause diffraction, which significantly con-
taminates the quiet zone by producing ripples as large
as 2 dB. Techniques to mitigate this phenomenon in-
clude serrations and rolled edges to scatter the energy
s Fig. 4 R&S PWC200 showing the PWC antenna array and
away from the quiet zone. The size and shape of the
calibration array mounted on a great-circle cut positioner. serrated/rolled edges determine the lowest operating
frequency, where the surface roughness determines the
trinsic receiving properties of the DUT in the far field upper frequency. Feed antenna pattern characteristics
from near-field coupling effects resulting from the test have a direct impact on the size of the quiet zone, as the
setup. There is also no access to a phase reference, so mirror essentially projects the radiation pattern of the
the NF-FF software transformation becomes inapplica- feed antenna onto the quiet zone. The reflector size with
ble. Therefore, EIRP can be evaluated accurately in the serrated/rolled edges is generally at least 2× the DUT/
near-field using NF-FF software but not EIS. quiet zone size, where a reflector with sharp edges is 3
to 4× the size of the quiet zone. The optimum reflec-
TRANSCEIVER PERFORMANCE MEASUREMENTS tor to DUT separation distance is (5/3)× the focal length
Another key question is the OTA evaluation of radio of the reflector. An optimal focal length can be derived
transceiver performance, such as EVM, ACLR or SEM. from the manufacturing shape tolerances with a range
Software NF-FF approaches are designed for process- of roughly 0.3 to 1 for the ratio of focal length to para-
ing periodic portions of the RF signal (the carriers) that bolic diameter.
determine propagation. However, this part of the signal Since the quiet zone size is dependent on the reflec-
is of no interest to assess these performance parame- tor characteristics instead of range length, it is much
ters, so the challenge is to extract information from the easier to create a large quiet zone inside small enclo-
carrier modulation. sures. Figure 3 shows the measured magnitude quiet
34
Antennas and Propagation.3 It comprises an array of 156
wideband Vivaldi antennas with a beamforming network
of phase shifters and attenuators, located at the rear.
This PWC array is 1.8 m wide and creates a spherical
quiet zone of 1 m diameter at a distance as short as 1.5
m in the frequency range from 2.3 to 3.8 GHz. In the
setup of Figure 4, the DUT (here the calibration antenna)
is mounted on a combined axis positioner, enabling full
spherical measurement. The calibration antenna is used
for evaluating the appropriate compensations of the in-
dividual RF channels of the PWC array, as well as deter-
mining the path loss of the entire test system. The PWC
system is reciprocal and has only one RF input/output
s Fig. 5EVM of a single 20 MHz carrier measured using the which can either be connected to a signal generator, a
R&S PWC200. spectrum analyzer or a VNA, enabling measurement of
zone size of 27 cm at 28 GHz of a CATR, similar to the devices with or without RF test ports.
one shown in Figure 2, with a 42 cm × 42 cm reflector. Figure 5 shows the single-carrier EVM measured with
This CATR setup fits within a chamber as small as 2 m the PWC for an OFDM signal with five 20 MHz carri-
× 1.5 m × 0.85 m. A direct far-field measurement sys- ers in the range of 2.35 to 2.45 GHz, using a Rohde &
tem featuring the same quiet zone size would require a Schwarz vector signal generator. The output power is
14.5 m range. 5 dBm and fed into a 60 cm × 60 cm patch array DUT.
Such technologies are of great interest for testing The demodulation is carried out by a Rohde & Schwarz
UEs or base stations operating in 5G NR FR2, promising vector signal analyzer connected to the PWC, where
a significant decrease in the size of test environments. the measurement span is 30.72 MHz. The EVM is as low
In addition, CATR has the same capabilities as a far-field as 0.41 percent, roughly corresponding to the internal
system, i.e., instantaneous and direct measurements of EVM of the measurement instruments. EVM results were
RF transceiver performance in both Tx and Rx. As the below 0.5 percent for the other four carriers, showing
path loss of such a system only occurs between the lim- that the PWC adds negligible EVM to the measurement
ited region where waves propagate between the feed setup.
and the reflector, the dynamic range of a CATR system is
better than a direct far-field approach. Using Figure 3 as SUMMARY
an example, the CATR system has a focal length of 0.7 Near-field techniques employing software transforma-
m compared to the equivalent far-field range length of tions are suitable for evaluation of EIRP and TRP quantities.
14 m, resulting in a path loss difference of 26 dB. When Rx or demodulation is involved with a DUT using
multiple non-identical RF transceivers, methods utilizing
PLANE WAVE SYNTHESIS hardware field transformations such as CATR and PWC
A CATR reflector is typically built using a solid piece overcome the limitations of software NF-FF. They also pro-
of aluminum to maintain the strict surface geometry vide compact and reliable alternatives to direct far-field
requirements. The 5G FR2 DUT size requirements al- measurements, making them well-suited for 3GPP RF con-
low for compact and rather light reflectors (20 to formance testing of UEs and base stations.n
40 kg). In the 5G FR1 range, reflector weight significantly
increases, up to hundreds of kilograms for base station References
1. C. L. I, C. Rowell, S. Han, Z. Xu, G. Li and Z. Pan, “Toward Green
DUTs. The cost, fabrication time and handling of large and Soft: A 5G Perspective,” IEEE Communications Magazine,
heavy mirrors becomes prohibitive. A lightweight and Vol. 52, No. 2, February 2014, pp. 66–73.
cost-effective alternative is to use an “electronic ver- 2. B. Derat, “5G Antenna Characterization in the FF,” IEEE EMC &
sion” of the CATR mirror.3 By combining the radiation APEMC 2018, Singapore, May 2018.
3. C. Rowell and A. Tankielun, “Plane Wave Converter for 5G Mas-
of multiple antennas assembled in a phased array and sive MIMO Base Station Measurements,” 12th EuCAP2018, Lon-
fed with pre-determined signal magnitude and phase, a don, U.K., April 2018.
plane wave within a defined quiet zone is created. A ver- 4. Derat et al., “A Novel Technology for Fast and Accurate Specific
sion of this near-field focusing technique was used for Absorption Rate Measurement (SAR),” iWAT, Karslruhe, Germany,
March 2013.
several years for the measurement of large phased array 5. Rohde & Schwarz, “2D Compact Range for Testing of AAS Base
radars at MIT Lincoln Labs and has been proposed as an Stations,” TSG-RAN WG4 #87, R4-1806605, May 2018.
OTA measurement baseline for base stations by 3GPP.5
Figure 4 shows the plane wave converting (PWC)
system reported at the 2018 European Conference on
35
5G NEW RADIO – FUNDAMENTALS,
PROCEDURES, TESTING ASPECTS
Online book with technical details
and implementation aspects of 5G
Five Rohde & Schwarz experts for wireless communications technologies
give insights into 5G NR. In this eBook readers
► explore all relevant details of the new technology –
from fundamentals to procedures in line with the 3GPP specification
► discover key implementation aspects by describing not only “how”
the technology was specified, but also “why” and
► learn which test and measurement equipment is required to develop,
manufacture and service 5G NR products
Register now
www.rohde-schwarz.com/5g-ebook
Visit us at
Hall 6
Booth 6B30

5G Test and Mesuarment Rohde Schwarz Jan 2020 v4

Uploaded by

Copyright:

Available Formats

5G Test and Mesuarment Rohde Schwarz Jan 2020 v4

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

5G Test and Mesuarment Rohde Schwarz Jan 2020 v4

Uploaded by

Copyright:

Available Formats

What are the main topics covered in the document?

What are the main topics covered in the document?

What testing techniques are discussed for evaluating 5G systems?

What testing techniques are discussed for evaluating 5G systems?

eBook

23 Optimizing the Perennial Doherty Power Amplifier

End-to-End 5G Test Solutions

Pat Hindle, Microwave Journal Editor

Revenue Revenue more base stations are deployed, the

Cellular Network Energy Consumption (China) RTOS RTOS ... RTOS

4G TD-LTE High Bandwidth Optical Transport Network

Type Size (kB) Ratio (DSR)

Radio Access Network Energy Consumption Centralized Control/Processing

Biggest CAPEX/OPEX Expense is Air Conditioning

s Fig. 2 Power Consumption in a Cellular Network. s Fig. 3 Centralized Baseband Processing.

x4(t) Signal BW (Hz) SNR (S/N)

Larger Signal BW = mmWave (5G FR2)

s Fig. 4 Spectral Efficiency.

s Fig. 6 Principles of Beamforming.

Single Transceiver + Antenna nth TRx Module

From Analog... ...To Digital ...To Hybrid

Processing Data Digital Digital

Chain Processing Chain •

s Fig. 7 Beamforming Architectures.

8 5G Optimal Capacity for DBF

FR2 mmWave (50) Antennas)

s Fig. 8 Optimized Networks: Spectral and Energy Efficiency.

2D/3D Antenna Characterization RF Transceiver Characterization

System and System and

R&S®AMS32 OTA Performance R&S®AMS32 OTA Performance

R&S®SMW200A R&S®ATS800 CATR Rack

R&S®ATS1800 CATR Conformance

s Fig. 12 Plane Wave Converter and CATR

TESTING FOR DSS

Hybrid-active load-pull overcomes the challenges in mmWave power amplifier design by

ance presented to a device under test (DUT), commonly

s Fig. 5 Passive on-wafer load-pull measurement of a GaN (7)

where RL is return loss RF Cable

CATR antenna test

Square Law –2.0

Empirical Eval Board 3.45

Prototype Output Side (a)

Design Review 44.0

s Fig. 5 Measurement-aided design flow for a digital Doherty 0 –50

Amplitude Difference (dB)

DAC Up-Converter Auxiliary

s Fig. 7 Dual-input Doherty in linear operation: measured

Saturated Power (W)

–1 28 Low Frequency - Nominal

Amplitude Difference (dB)

Saturated Power (W)

Saturated Power (W)

Amplitude Difference (dB)

41 31 High Frequency - Nominal

Amplitude Difference (dB)

Saturated Power (W)

COMPACT TEST SOLUTION FOR 5G mmWAVE

s Fig. 2 Compact antenna test range with a roll-edge reflector

ward larger distances and extract far-field radiation

3 HARDWARE NEAR-FIELD TRANSFORMATIONS