1EF114_e_802_11be_EVM

Application Note
HINTS FOR IEEE 802.11BE EVM

MEASUREMENTS
Optimal hardware and measurement settings
Products:
► R&S®FSW ► R&S®SMW200A
► R&S®FSVA3000 ► R&S®FSV3000
M. Breinbauer | 1EF114 | Version 0e | 08.2022

https://www.rohde-schwarz.com/appnote/1EF114
Contents
1 Overview........................................................................................................ 4
2 Introduction to 11be ..................................................................................... 5
2.1 IEEE 802.11be Goals and Feature Summary .....................................................5
2.2 Achieve Extremely High Throughput and More ...................................................6
2.2.1 Up to 320 MHz Channel Bandwidth ....................................................................6
2.2.2 Multi-Link Operation (MLO) .................................................................................8
2.2.3 Restricted Target Wake Time..............................................................................8
2.2.4 Multi-AP Operation ..............................................................................................8
2.3 IEEE 802.11be Physical Layer ............................................................................8
2.3.1 New RU and Tone Plans .....................................................................................8
2.3.2 Multiple Resource Units per User ........................................................................9
2.3.3 Preamble / Subchannel Puncturing .....................................................................9
2.3.4 PPDU Formats ....................................................................................................9
2.3.5 EHT Preamble: Designed for the Future ...........................................................10
2.4 PHY Layer Test Requirements ..........................................................................11
2.4.1 Transmitter Requirements .................................................................................11
2.4.2 EHT Receiver Requirements .............................................................................16
3 Instrument Selection .................................................................................. 17
3.1 Signal Analyzer: ................................................................................................17
3.2 Signal Generator ...............................................................................................20
4 Factors Influencing the EVM Measurement .............................................. 22
4.1 Measurement Settings ......................................................................................22
4.2 Crest Factor of the Test Signal ..........................................................................23
4.3 EVM Independence from Modulation Order & Burst Length .............................26
4.4 4096-QAM EVM Limitation for Low-SNR Conditions ........................................28
5 How to Set Up for Best EVM Performance ................................................ 32
5.1 Signal Generator ...............................................................................................32
5.1.1 Digital Attenuation .............................................................................................32
5.1.2 Baseband Offset ...............................................................................................32
5.1.3 Optimize-EVM Feature ......................................................................................33
5.2 Signal Analyzer .................................................................................................33
5.2.1 Auto-Level Enhancements and Optimize EVM Feature ....................................33
5.2.2 R&S®FSW-K18 Option with IQ Averaging ........................................................35
5.2.3 Modulation Measurement Optimizer Tool..........................................................35
Rohde & Schwarz | Application Note Hints for IEEE 802.11be EVM Measurements 2
6 Other 11be Transmitter Test Items ............................................................ 36
7 Literature ..................................................................................................... 38
8 Ordering Information .................................................................................. 38
1 Overview
IEEE 802.11be Extremely High Throughput (EHT), also known as Wi-Fi 7, is the latest amendment of the
IEEE 802.11 standard and is still under development. This amendment focuses mainly on improved
throughput. To do so, the most notable changes currently implemented are:
► New modulation scheme: 4096-QAM (4K-QAM)
► Larger bandwidth: 320 MHz
► Support for 16x16 MU-MIMO
► Enhanced resource allocation in OFDMA
This document provides some technical background and guidance on how to optimize the measurement of
one of the key performance parameters of WLAN transmitters - the Error Vector Magnitude (EVM).
2 Introduction to 11be
The content of this chapter 2 is to a large extent an excerpt from [1], a Rohde & Schwarz Technology
Introduction on 802.11be - with some additions mainly in chapter 2.4.1.3.
IEEE 802.11 is the IEEE working group that develops the Wireless Local Area Network (WLAN)
specifications that are behind the Wi-Fi technology. In the beginning, IEEE 802 created 11a/b/g that enabled
wireless communication for everyone. The working group continues to enhance the 802.11 MAC and PHY
layer to, for example, improve user experience, increase throughput, utilize resources more efficiently and/or
support new use cases. These enhancements are written as amendments to the base 802.11 standard and
many have been published in the last 20+ years.
The 802.11 working group is currently developing the next generation of Wi-Fi, 802.11be (Wi-Fi 7). 802.11be
is also known as EHT (Extremely High Throughput). The high-level goals of EHT are defined in the 802.11be
PAR (Project Authorization Request) and include enabling at least one operation mode that can support at
least 30 Gbps and at least one operation mode to improve latency and jitter to better support time sensitive
networks. EHT is supported in the unlicensed bands between 1 and 7.125 GHz.
EHT re-uses many of the concepts and techniques used at the HE PHY layer but makes enhancements such
as assigning multiple resource units to a single user, increasing bandwidth support to 320 MHz, supporting
up to 16 spatial streams and adding support for 4096 QAM modulation. At the MAC layer, EHT introduces
several significant features to 802.11 such as Multi-Link Operation (MLO).
The current two 802.11be “Core” Documents are the 802.11be draft amendment version 1.0 and the
802.11be Specification Framework Document (SFD) version 23.
2.1 IEEE 802.11be Goals and Feature Summary
802.11n (HT) 802.11ac (VHT) 802.11ax (HE) 802.11be (EHT)

Supported bands 2.4, 5 GHz 5 GHz 2.4, 5, 6 GHz 2.4, 5, 6 GHz
Channel 20, 40 20, 40, 80, 80+80, 20, 40, 80, 80+80, 20, 40, 80, 160, 320
Bandwidth (MHz) 160 160
Subcarrier 312.5 312.5 78.125 78.125
Spacing (kHz)
Symbol Time (us) 3.2 3.2 12.8 12.8
Cyclic Prefix (us) 0.8 0.8, 0.4 0.8, 1.6, 3.2 0.8, 1.6, 3.2
MU-MIMO No Downlink Uplink & Downlink Uplink & Downlink
Modulation OFDM OFDM OFDM, OFDMA OFDM, OFDMA
Data Subcarrier BPSK, BPSK, BPSK, BPSK,
Modulation QPSK, QPSK, QPSK, QPSK,
16-QAM, 16-QAM, 16-QAM, 16-QAM,
64-QAM 64-QAM, 64-QAM, 64-QAM,
256-QAM 256-QAM, 256-QAM,
1024- QAM 1024-QAM,
4096-QAM
Coding BCC (Mandatory) BCC (Mandatory) BCC (Mandatory) BCC (Mandatory)
LDPC (Optional) LDPC (Optional) LDPC (Mandatory) LDPC (Mandatory)
PHY parameter values for 11n, 11ac, 11ax and 11be
802.11be aims to specify MAC and PHY features that will meet the high throughput and low latency
requirements for applications like video conferencing, remote working, gaming and cloud computing. The
11be PHY is in general very similar to the PHY defined in the highly successful 11ax standard.
Further, 11be has an eye towards future compatibility. 802.11 PHY always consider backwards compatibility
with legacy 802.11 generations, but 11be introduces concepts to make forward compatibility achievable. For
example, 11be introduces a new preamble field called the universal SIG (U-SIG) that will be used in
802.11be and all future 802.11 generations as well as more precise terms/rules for reserved bits.
The 11be MAC layer introduces significant changes and new features such as multi-link operation, multi AP
(Access Point) support, restricted target wait time (TWT) and 1024 bit block acknowledgement (1K BA) to
meet the targeted low latency applications.
2.2 Achieve Extremely High Throughput and More

In order to achieve the targeted extremely high throughput, the EHT physical layer supports wider bandwidth,
more spatial streams and a higher modulation scheme. MLO (Multi Link Operation) provides the possibility to
increase throughput by using several physical links in parallel.
2.2.1 Up to 320 MHz Channel Bandwidth
The IEEE 802.11be standard amendment 8 (EHT) covers carrier frequency operation between 1 and 7.250
GHz ensuring backward compatibility and coexistence with legacy IEEE 802.11 devices operating at 2.4
GHz, 5 GHz and 6 GHz bands. The channel allocation is defined in the following documents:
─ 2.4 GHz Band, 802.11-2016, chapter 19.3.15.2:
Channel center frequency = 2407 + 5 x nch (MHz), nch = 1, 2, …13
─ 5 GHz Band, 802.11-2016, chapter 19.3.15.3:
Channel center frequency = Channel starting frequency + 5 x nch (MHz), nch = 1, … 200
Channel starting frequency is defined as dot11ChannelStartingFactor x 500kHz or is set to 5GHz
where dot11ChannelStartingFactor is false.
─ 6 GHz Band, 802.1ax, 27.3.23.2:
Channel center frequency = Channel starting frequency + 5 × nch (MHz), nch = 1, …, 233
Channel starting frequency is defined as dot11ChannelStartingFactor x 500kHz. For example, a
channel center frequency of 5.955 GHz is indicated by dot11ChannelStartingFactor = 11900 and
nch = 1. A channel center frequency of 5.935 GHz is indicated by dot11ChannelStartingFactor =
11850 and nch = 2.
The channel starting frequencies depending on the operating classes are defined in the Annex E of these
documents.
The 802.11be amendment extends the previously existing standards. EHT supports 20, 40, 80, 160 and 320
MHz channel bandwidths. 320 MHz bandwidth is new in 802.11be and is made possible due to updated
regulatory rules in many countries that now allow unlicensed device operation (subject to regional
restrictions) in the 6 GHz band. In some regions like US and Canada full spectrum from 5925 to 7125 MHz is
available while in other regions like Europe only the lower part from 5945 to 6425 MHz is available. In
general, three device categories are considered in the spectrum regulations:
► low power indoor (LPI) devices with a power (EIRP) limit of around 250 mW
► very low power devices (VLP) with a power (EIRP) limit of around 25 mW
► standard power devices for indoor and outdoor with an AP applying automatic frequency coordination
(AFC)
The Wi-Fi Alliance provides a global spectrum tracker of current status for unlicensed usage in the 6GHz
band via its 'Countries Enabling Wi-Fi 6E' page: https://www.wi-fi.org/countries-enabling-wi-fi-6e
Because 802.11 wants to use this newly allocated spectrum to its fullest, EHT seeks to encourage high
throughput and low latency applications in the 6 GHz band. Therefore, for example, EHT does not allow 20
MHz only devices to operate in the 6 GHz band. Additional EHT device bandwidth requirements for the 3
operating bands are listed below:
► An 11be AP must support:
─ 160 MHz operating channel width in the 6 GHz band
─ 80 MHz operating channel width in the 5 GHz band
─ 20 MHz operating channel width in the 2.4 GHz band
► 11be non-AP STA must support 80 MHz channel in 5 / 6 GHz (unless it is a 20 MHz only device)
► A 20 MHz only device may only operate in the 2.4 and 5 GHz band
The figure below shows the example channelization based on the FCC regulations in the US. Because there
are an odd number of 160 MHz channels, EHT defines a set of overlapping 320 MHz channels. This provides
a way to efficiently use the spectrum since each 160 MHz can be part of a 320 MHz channel.
The center frequencies are determined using the equation from the 802.11 standard:
channel center frequency = channel starting frequency + 5 × n ch (MHz)
where
nch = channel center frequency number
31, 95, and 159 for 320 MHz-1 or
63, 127, and 191 for 320 MHz-2 channels
channel starting frequency = 5.950 GHz
To make it easier for signaling and for a STA (Station) to identify which of the overlapping 320 MHz channels
it is hearing, 11be defines two types of 320 MHz channelization to identify two non-overlapping channel sets:
320 MHz-1 and 320 MHz-2
320 MHz channel allocation based on FCC in the 6 GHz band
802.11be non-contiguous (e.g. 80+80 MHz) operation is not used because, in part, it was found that the
80+80 MHz non-contiguous channelization defined in 802.11ax is not commonly used. In addition, the
802.11be MLO MAC feature can be used to achieve the same result as the non-contiguous channels in the
PHY.
To achieve extremely high throughput, 802.11be will also support up to 16 spatial streams across all
scheduled stations for DL/UL MU-MIMO and SU-MIMO. For EHT MU-MIMO transmissions the maximum
number of spatial streams per STA is 4 to maximum eight users.
2.2.2 Multi-Link Operation (MLO)
A new MAC feature called MLO is introduced in EHT. A device capable of MLO (called an MLD or Multi-Link
Device) can establish multiple links on different channels with another MLD.
2.2.3 Restricted Target Wake Time
The target wake time (TWT) feature was introduced in 802.11ah to support low power IoT applications by
allowing STAs - after negotiation with the AP - to go into the sleep mode outside the wake time period. It also
allows the AP to distribute the wake period of different STAs over time in order to minimize contention.
802.11be extends the TWT capability to give the AP mechanisms to provide more predictable latency to
support time sensitive application requirements.
2.2.4 Multi-AP Operation
In 802.11be also a couple of features are under discussion to improve efficiency of operation of adjacent
access points. By this means spectrum resources can be used more efficiently and the throughput can be
further improved.
2.3 IEEE 802.11be Physical Layer
2.3.1 New RU and Tone Plans
802.11ax (HE) introduced OFDMA to 802.11 and users were allocated RU (Resource Units) that were 26,
52, 106, 242, 484 or 996 tones in size.
The EHT tone plan and resource unit (RU) locations for a 20 and 40 MHz PPDU (Physical Layer Protocol
Data Unit) are the same as for the 802.11ax 20 and 40 MHz PPDU. The EHT 80 MHz PPDU RU and tone
locations, however, are slightly different from HE. When HE developed the tone plan, the focus was to use
the spectrum as efficiently as possible. Therefore, they used as many subcarriers as possible to transmit
data. For example, a middle RU was defined that straddled DC and no null carrier was used between two
242-tone RUs.
One consequence of this design is that the 242-tone RU do not align with a 20 MHz subchannel boundary
and the middle 26-tone RU falls into two 20 MHz sub-bands. This causes problems in case of preamble
puncturing, for example, where a punctured 20 MHz subchannel punctures tones from an adjacent RU,
causing some degradation to the adjacent channel RU. The following figure shows the misalignment of the
20 MHz boundaries and the 242-tone RU.
HE 80MHz tone plan showing 2 examples where a punctured 20 EHT 80 MHz tone plan solving the misalignment
MHz subchannel impacts subcarriers of adjacent resource units
The EHT 80 MHz plan solves this issue with a minimal change to the HE 80 MHz plan. This small change
means that the same RU sizes can be used in EHT as was used in HE, but it eliminates the middle 26-tone
RU and aligns the 242-tone RU with 20 MHz channel boundaries.
2.3.2 Multiple Resource Units per User
While much of the 802.11be PHY is the same or very similar to 11ax, a key differentiator for 802.11be is the
capability to allocate more than one resource unit to a single user. Assigning multiple RUs per user provides
scheduling flexibility to take advantage of frequency diversity and to allocate resources efficiently within the
spectrum. The drawback of this additional flexibility is the potential need to increase overhead to describe all
of the possible RU combinations. EHT circumvents this by defining rules to limit which RUs may be
combined, focusing on those combinations that provide the most benefit. The basis of the rules is the EHT
classification of the RU as a small or a large size RU. Small size RUs are less than 20 MHz, i.e. 26, 52, 106
tone RUs. A large size RU is 20 MHz or larger, i.e. 242-tone (20 MHz), 484-tone (40 MHz), 996-tone
(80 MHz) RUs. A small size RU and a large size RU are not used together in MRU. That is, MRUs either
contain only 2 small size RU or 2 large size RU.
2.3.3 Preamble / Subchannel Puncturing
802.11 operates in unlicensed bands that can be utilized by other networks and technologies. In addition,
incumbent users are present in the newly allocated 6 GHz. It is important that 802.11 networks limit any
interference into already occupied channels. At the same time, however, it is important to utilize spectrum as
efficiently as possible and make use of the wider channel bandwidths defined for 802.11be (and 802.11ax).
802.11be (and ax) therefore can utilize a wide bandwidth channel (for example 160 MHz) for transmission
but ‘puncture’, or not transmit, in a subchannel that is already in use. This is called preamble puncturing.
2.3.4 PPDU Formats
In 802.11 data is transmitted over the wireless medium using PHY frames which are also called PPDU (PHY
layer Protocol Data Unit). The PPDU contain the data to be transmitted along with a preamble prepended to
the data. The preamble consists of several fields which are used to aid reception (e.g. automatic gain control
and timing synchronization) and provide information that the receiver will need to know to demodulate the
packet. The field names are listed in the following table.
Field Description Field Description
L-STF Legacy Short Training Field EHT-SIG EHT Signal Field
L-LTF Legacy Long Training Field EHT-STF EHT Short Training Field
L-SIG Legacy Signal Field EHT-LTF EHT Long Training Field
RL-SIG Repeated Legacy Signal Field Data Data
U-SIG Universal Signal Field PE Packet Extension Field
PPDU Field Descriptions
Two PPDU formats are defined in EHT:

► Multi User PHY Protocol Data Unit (EHT MU PPDU)
► Trigger Based PHY Protocol Data Unit (EHT TB PPDU).
The EHT MU PPDU can be sent to a single user or to multiple users. The related EHT-SIG field, along with
the U-SIG, provides RU/MRU allocations and other information the STA(s) need to know in order to
understand the EHT MU Packet. When the MU PPDU is sent to multiple users, the transmission can be
OFDMA or MU-MIMO. Further, an RU that is 242 tones or larger in an OFDMA transmission may use MU-
MIMO to send the RU to up to 8 users.
EHT Multi User PDDU Format
An AP uses a control frame called a trigger frame to assign resources and solicit a response from one or
more STAs. The STA(s) use the EHT TB PPDU to respond to the trigger from the AP. The EHT TB frame
format is very similar to the EHT MU PPDU. However, the TB PPDU does not contain the EHT-SIG preamble
field. In addition, the EHT-STF field is two times longer than in the EHT MU PDDU in order to improve
performance and reliability for the uplink transmissions.
EHT Trigger Based PDDU Format
2.3.5 EHT Preamble: Designed for the Future
The preamble is an important part of the design. It is used to provide information such as MCS (Modulation
and Coding Scheme) that is needed for the receiver to decode the transmitted data. It is also used to provide
backwards compatibility with previous PHY versions. However, the preamble never directly conveyed the
PHY version of the packet. Instead auto detection / spoofing mechanisms were defined for the receiver to
determine the PHY version implicitly. As the number of PHYs has increased, the auto detection algorithms
have become more complex.
EHT will solve this problem by introducing the universal sig (U-SIG) field. The U-SIG comes right after the
RL-SIG and is 2 OFDM symbols in length. The U-SIG will be present in EHT and all future 802.11 PHYs and
contains version independent and version dependent bits. The version independent bits are the first 20 bits of
the U-SIG and will have the same location and definition for EHT and all future PHYs. The first 3 bits (bits 0
to 2) of the U-SIG are used to identify the PHY version which will greatly simplify auto detection for EHT and
future 802.11 generations. The next 3 bits indicate the spectrum occupancy of the PPDU (e.g. 80 MHz BW).
The 7th bit signals the link direction (i.e. uplink or downlink). The next 6 bits identify the BSS (Basic Service
Set) in use via the BSS color and the 7 TXOP bits provides information on how the long the PPDU uses the
medium. The remainder of the U-SIG bits/fields (not described in the table) will depend on the PHY version
and PPDU type.
U-SIG filed content for EHT MU PDDU and EHT TB PDDU
To provide flexibility and prepare for possible new capabilities, EHT classifies the reserved bits in the EHT
preamble as 'disregard' or 'validate'. This classification helps a receiver determine the appropriate action if it
comes across a bit value that is not used in a PHY it supports. Disregard means it doesn't matter/ignore this
bit and continue reception. Validate means the device should check if the bit matches a known value and if it
does not, the device should terminate reception.
2.4 PHY Layer Test Requirements
2.4.1 Transmitter Requirements
2.4.1.1 Transmit Spectral Mask
The spectrum mask requirements for the 11be 20/40/80/160 MHz bandwidth transmissions are the same as
in 11ax. The spectrum mask for the 11be 320 MHz bandwidth is a scaled version of the 11ax mask.
Measurements are made using RBW=100 KHz and VBW=7.5 KHz. No other analyzer settings are specified.
2.4.1.1.1 Spectrum Mask for a PPDU with Punctured Channel(s)
Punctured subchannels are used in EHT (and HE) to avoid transmitting on subchannels that are not available
or are occupied by other/incumbent users. To limit the leakage from the occupied subchannel(s) into the
punctured subchannel(s), EHT applies an additional mask so that an overall spectrum mask for a PPDU
containing punctured channel(s) is formed by combining the transmit spectral mask defined in the previous
section with a puncture mask. The puncture masks are based on the puncture masks defined in the ETSI
BRAN (Broadband Radio Access Networks) EN 301 893 standard.
2.4.1.2 Spectral Flatness
Spectral flatness provides a way to measure whether the subcarriers have a similar amount of power. This is
done by determining the average energy of a range of subcarriers and verifying that no individual subcarriers
energy in that range deviates by more than the value specified. EHT spectral flatness requirements for 20,
40, and 80 MHz non-punctured PPDU are the same as in HE.
2.4.1.3 Transmitter Modulation Accuracy
2.4.1.3.1 Transmit Center Frequency Leakage
This measures the amount of energy that ‘leaks’ through and appears at the RF LO frequency. This
measurement is needed because, depending on the type of receiver used, too much power leakage at this
frequency may lead to poor demodulator performance. Further, if the power level is too high, a receiver may
false trigger on the signal.
2.4.1.3.2 Transmitter Constellation Error
The transmitter constellation error - also called EVM (Error Vector Magnitude) - is an important figure of merit
for a transmitter in a digital modulation system. It provides a way to measure how close to an ideal
constellation point the device is able to transmit.
The EVM requirements depend on the specific transmission parameters used by a WLAN device. These are
adapted according to the current quality of the RF environment. All the allowed combinations of e.g.
modulation and coding rate are listed in the Modulation on Coding Scheme (MCS) tables under a unique
MCS index.
MCS Modulation Coding EVM of EVM of EVM of
EHT MU PDDU EHT TB PDDU EHT TB PDDU
transmit power transmit
larger than MCS 7 power equal or less than
maximum power MCS7 max power
0 BPSK 1/2 - 5 dB -13 dB -27 dB
1 QPSK 1/2 - 10 dB -13 dB -27 dB
2 3/4 -13 dB -13 dB -27 dB
3 16-QAM 1/2 -16 dB -16 dB -27 dB
4 3/4 -19 dB -19 dB -27 dB
5 64-QAM 2/3 -22 dB -22 dB -27 dB
6 3/4 -25 dB -25 dB -27 dB
7 5/6 -27 dB -27 dB -27 dB
8 256-QAM 3/4 -30 dB -30 dB -30 dB
9 5/6 -32 dB -32 dB -32 dB
10 1024-QAM 3/4 -35 dB -35 dB -35 dB
11 5/6 -35 dB -35 dB -35 dB
12 4096-QAM 3/4 -38 dB -38 dB -38 dB
13 5/6 -38 dB -38 dB -38 dB
14 BPSK-DCM-DUP 1/2 -5 dB N/A N/A
15 BPSK-DCM 1/2 -5 dB -13 dB -27 dB
ETH transmitter constellation error specification
As stated in chapter 36.3.19.4.1 of the 802.11be draft amendment (also see table above), transmitters need
to fulfill the -38 dB EVM requirement for 4096-QAM (practically required additional margin is not considered
here). This means the measurement equipment must be able to accurately measure EVM values around and
below -38 dB.
A signal analyzer cannot be assumed to be an ideal measurement equipment by itself, but it also contributes
to the measured EVM. There are in principle two different types of EVM contributions: on the one extreme
there are coherent contributions. These sum up well-defined depending on their specifics in a way either
increasing the overall EVM or possibly even lower the measured EVM. On the other side there are
incoherent, noise-like contributions. These random contributions always increase the overall measured EVM
according to the rooted sum of squared EVM contributions (RSS).
The sources for residual EVM of signal analyzers are mainly not coherent to the signal which should be
measured and thus the influence on the measured EVM value can be calculated according to the RSS
formula. If the signal analyzers residual EVM shows a 10dB margin to the targeted EVM value of -38 dB ("it is
able to measure down to -48 dB EVM"), the error in the measured EVM value due to the signal analyzers
contribution is around 0.41 dB or 0.06% - as shown in the following table.
EVM to EVM to Assumed SA EVM SA EVM Total EVM Total EVM
measure measure SA EVM contribut. contribut. EVM error EVM error
[dB] [%] Margin [dB] [%] [%] [%] [dB] [dB]
[dB]
-38 1,259 0 -38 1,259 1,780 0,521 -34,990 3,010
-38 1,259 2 -40 1,000 1,608 0,349 -35,876 2,124
-38 1,259 4 -42 0,794 1,489 0,230 -36,545 1,455
-38 1,259 6 -44 0,631 1,408 0,149 -37,027 0,973
-38 1,259 8 -46 0,501 1,355 0,096 -37,361 0,639
-38 1,259 10 -48 0,398 1,320 0,061 -37,586 0,414
-38 1,259 12 -50 0,316 1,298 0,039 -37,734 0,266
-38 1,259 14 -52 0,251 1,284 0,025 -37,830 0,170
-38 1,259 16 -54 0,200 1,275 0,016 -37,892 0,108
-38 1,259 18 -56 0,158 1,269 0,010 -37,932 0,068
-38 1,259 20 -58 0,126 1,265 0,006 -37,957 0,043
-38 1,259 22 -60 0,100 1,263 0,004 -37,973 0,027
-38 1,259 24 -62 0,079 1,261 0,003 -37,983 0,017
-38 1,259 26 -64 0,063 1,261 0,002 -37,989 0,011
-38 1,259 28 -66 0,050 1,260 0,001 -37,993 0,007
-38 1,259 30 -68 0,040 1,260 0,001 -37,996 0,004
Influence of the signal analyzers residual EVM margin on the measurement result, fixed target EVM
Influence of the signal analyzers residual EVM margin on the measurement result, target EVM -38 dB
If a fixed margin of 10 dB for the instruments residual EVM is assumed, the EVM contributions at different
target EVM values calculate as follows:
Target Real SA EVM SA EVM SA EVM Total EVM Total EVM

EVM EVM Margin contr. contr. EVM error EVM error
[dB] [%] [dB] [dB] [%] [%] [%] [dB] [dB]
-30 3,162 10 -40 1,000 3,317 0,154 -29,586 0,414
-32 2,512 10 -42 0,794 2,634 0,123 -31,586 0,414
-34 1,995 10 -44 0,631 2,093 0,097 -33,586 0,414
-36 1,585 10 -46 0,501 1,662 0,077 -35,586 0,414
-38 1,259 10 -48 0,398 1,320 0,061 -37,586 0,414
-40 1,000 10 -50 0,316 1,049 0,049 -39,586 0,414
-42 0,794 10 -52 0,251 0,833 0,039 -41,586 0,414
-44 0,631 10 -54 0,200 0,662 0,031 -43,586 0,414
-46 0,501 10 -56 0,158 0,526 0,024 -45,586 0,414
-48 0,398 10 -58 0,126 0,418 0,019 -47,586 0,414
-50 0,316 10 -60 0,100 0,332 0,015 -49,586 0,414
-52 0,251 10 -62 0,079 0,263 0,012 -51,586 0,414
-54 0,200 10 -64 0,063 0,209 0,010 -53,586 0,414
-56 0,158 10 -66 0,050 0,166 0,008 -55,586 0,414
-58 0,126 10 -68 0,040 0,132 0,006 -57,586 0,414
-60 0,100 10 -70 0,032 0,105 0,005 -59,586 0,414
Influence of the signal analyzers residual EVM margin on the measurement result, fixed EVM margin
The procedure for calculating EVM is very similar to IEEE802.11ax. The test is performed using a minimum
of 20 PPDUs with at least 32 data symbols containing random data if the occupied RU has 26 tones. If the
occupied RU has more than 26 tones, then the PPDUs shall be at least 16 data symbols long. EVM
calculation is done using compensation of both estimated frequency offset and sampling offset drift. The
result is determined by averaging over the subcarriers, frequency segments, EHT PPDUs and spatial
streams.
Test equipment used for this measurement should have a residual EVM of 10 dB or less. This means that the
analyzer should be capable of measuring lower than -48 dB for the 4096 QAM case. The following figure
shows a screenshot from the R&S®FSW Signal and Spectrum Analyzer achieving -50 dB EVM for an EHT
PPDU using 4096 QAM modulation in a 320 MHz channel.
4096QAM WLAN constellation on 320 MHz channel and EVM
The residual EVM of a signal analyzer is usually evaluated by directly connecting it to a high-end signal
generator. Then the EVM is measured for the frequency and signal of interest over different power levels.
The result is a typical bathtub curve EVM vs. Power. This curve is dominated by noise on the low-power side
and non-linearities on the high-power side.
Typical bathtub curve measuring EVM versus signal power
To minimize the EVM measurement uncertainty from test system contributions over a wide range of power
levels, the RF front-end of both instruments needs to be adapted continuously - ideally using an automatic
levelling algorithm. These algorithms generally have to do a trade-off between best possible performance
and the required time to achieve this optimal levelling.
Both instruments - the generator and the analyzer - contribute to the measured residual EVM.
2.4.1.3.3 Unused Tone Error
For EHT TB PPDU, the EVM requirements need to account for multiple STAs transmitting at the same time.
The AP will see the noise from the multiple sources as a total cumulative noise and network performance will
decrease if this noise becomes too large. In addition, a STA transmitting power unintentionally outside of its
allocated RU will negatively affect the EVM of other STAs. Therefore, an EHT TB PPDU must also meet an
EVM requirement for unused tones to measure if the STA is causing interference to adjacent RUs.
2.4.2 EHT Receiver Requirements
The 802.11be receiver testing requirements and limits are similar to those defined in the 802.11ax
specification and cover the following test items:
─ Receiver minimum input sensitivity
─ Adjacent and nonadjacent channel rejection
─ Receiver maximum input level
─ Trigger based PPDU precorrection specifications
─ Transmit power accuracy and RSSI measurement
─ Carrier frequency offset (CFO) error and timing drift
3 Instrument Selection
To optimally cover all the very different markets and applications for signal analyzers and signal generators,
these instruments can be configured with a whole lot of different options to best suit the customer
requirements. The chosen configuration can strongly influence the performance parameters of the setup.
Comparing the performance of instruments with respect to certain parameter is only feasible, if the possibly
different optioning of the instruments is kept in mind.
3.1 Signal Analyzer:

When selecting the best signal analyzer for a test setup to ensure lowest possible residual EVM contribution,
the following points have to be considered.
The R&S®FSW Signal and Spectrum Analyzer is called FSW in the following.
Lowest frequency option possible (8/26GHz)

A higher maximum input frequency is always associated with a more complex RF front-end of the instrument.
Typically, some switching is required to provide best performance over the wider frequency range. This adds
some loss to the signal path. Additionally, the elements of the front end must be designed to handle wider
frequency spans. This in general increases the challenges for the design.
The displayed average noise level (DANL) from the datasheet can be used as an indicator for good EVM
measurement performance:
Frequency DANL R&S FSW8 DANL R&S FSW13/26 DANL R&S FSW43 DANL R&S FSW50 DANL R&S FSW67
[GHz] [dBm] [dBm] [dBm] [dBm] [dBm]
1 GHz -150/-154 -149/-154 -149/-154 -149/- -149/-
<3 GHz -152/-156 -151/-156 -150/-155 -150/- -150/-
>3 GHz -152/-156 -151/-156 -150/-155 -150/- -150/-
8 GHz -152/-156 -149/-154 -148/-152 -148/- -144/-
DANL, nominal/typ. values, preamplifier off, instrument without R&S®FSW-B13, noise cancellation off, RF attenuation = 0 dB,
termination = 50 Ω, normalized to 1 Hz RBW, see datasheet
The values in the above table are valid for the spectrum mode. The values for the IQ-mode can differ but are
not listed separately in the datasheet. The figure below shows an example measurement of the DANL in IQ-
mode for two specific R&S®FSW models with R&S®FSW-B320 option.
DANL in IQ mode for 320MHz measurement bandwidth; different R&S®FSW models with pre-amplifier off, 15dB and 30dB gain
Therefore - as a rule of thumb - instrument models with a frequency range "just enough" for the
measurement task should be preferred.
In addition, it has to be noticed that the auto-levelling routine of the WLAN-11be application is optimized for
the most relevant models R&S®FSW8/13/26 so far. The optimization for the models R&S®FSW43/50/67
might follow, even though we do not recommend these models for WLAN applications. The optimal levelling
of the RF frontend (preamplifier gain, attenuation, reference-level) is absolutely crucial for the residual EVM
measurement. Manual optimization of the levelling is always possible.
Preamplifier option
The signal analyzer should be equipped with the R&S®FSW-B24 preamplifier option. The switchable
preamplifier helps to optimally level the signal in a wider power range. Multi-stage preamplifiers provide an
even more precise levelling compared with single-stage preamplifiers.
As stated in the previous section, R&S®FSW8 or R&S®FSW26 are recommended for 802.11be
measurements.
If models with higher frequency range are used, it should be kept in mind that a new, redesigned and
optimized preamplifier model with an enhanced microwave frontend is available for the R&S®FSW43,
R&S®FSW50 and R&S®FSW67. New instruments starting with the following serial numbers are equipped
with this enhanced frontend by default (if R&S®FSW-B24 is ordered):
Type From Serial Number
R&S®FSW43 102199
®
R&S FSW50 101672
R&S®FSW67 101701
Serial numbers for FSW with new preamplifier option by default
With the latest firmware, the new enhanced R&S®FSW-B24 option will show B24U in the option list to clearly
identify the new hardware. Upgrade kits are available for instruments with material number 1331.5003.XX,
which already have a previous version R&S®FSW-B24 installed.
Option R&S®FSW-B25
If this option is installed (electronic attenuator), the specified DANL of the datasheet is degraded by typ. 1dB.
Option R&S®B25 should only be chosen, if a heavy use of the mechanical attenuator is expected (e.g. in
automated test systems). It can then help to reduce the number of switching cycles of the mechanical
attenuator which extends the lifetime of the mechanical relays.
Smallest bandwidth option possible

Instruments equipped with wider instantaneous bandwidth options, are generally also more susceptible to
noise in the receive path. If the R&S®FSW-B1200/-B2001/-B800R/-B4001/-B6001/-B8001 option is installed,
the specified DANL in the datasheet has to be corrected by typ. +2dB.
Comparison of EVM versus Power measurements with different bandwidth options
Therefore, the R&S®FSW-B320 and R&S®FSW-B512 are the recommended bandwidth options for 11be
measurements. Instruments equipped with R&S®FSW-B1200 show slightly higher residual EVM values as
expected based on the DANL specification.
In addition, it has to be noticed that the auto-levelling routine of the WLAN-11be application is optimized for
the most relevant options B160/320/512 so far. The optimization for the B1200 option will follow. The optimal
levelling of the RF frontend (preamplifier gain, attenuation, reference-level) is absolutely crucial for the
residual EVM measurement. Manual optimization of the levelling is always possible.
Latest Firmware with "Optimize EVM" function

The R&SFSW-K91BE IEEE 802.11be Measurements option was first available in the firmware version 4.90
and enhanced in the firmware version 5.00.
The firmware release 5.10 introduces the new auto-levelling feature Optimize EVM in combination with the
bandwidth options R&S®FSW-B320 and R&S®FSW-B512. This extends the optimization of the RF frontend
settings by an iterative search for the minimum residual EVM. If activated, this helps to find the best
instrument settings with respect to the signal conditioning in a fast, repeatable and automatic manner.
Optimize EVM drop down menu
3.2 Signal Generator

For a measurement of the residual EVM of a signal analyzer, a high-quality signal source is required to
create the test signal. Same as the signal analyzer, also the R&S®SMW200A Vector Signal Generator can
be adopted to the customer requirements with a lot of different hardware options. To achieve the best
possible signal fidelity for 11be measurements, the following points should be considered.
The R&S®SMW200A Vector Signal Generator is called SMW in the following.
Latest RF hardware
Since its introduction, the SMW has been established as the leading high-end vector signal generator in the
market. Rohde & Schwarz continuously improves the instrument to keep it state of the art. One of these
upgrades relates to the RF frequency options. To distinguish older from the latest hardware, the new
frequency options use the following new nomenclature:
RF path A: SMW-B10xx
RF path B: SMW-B20xx
In contrast to that, the older frequency options had 3-digit option numbers.
To achieve best EVM performance, instruments with the new RF options should be used.
Low noise options

The SMW with the new RF hardware can be equipped with different types of low phase noise options,
providing different levels of phase noise performance.
Phase noise performance level Required options for RF path A Required options for RF path B
Standard performance R&S®SMW-B10xx frequency option R&S®SMW-B20xx frequency option
Low phase noise R&S®SMW-B10xx frequency option and R&S®SMW-B20xx frequency option and
R&S®SMW-B709 R&S®SMW-B719
Improved close-in phase noise R&S®SMW-B10xx frequency option and R&S®SMW-B20xx frequency option and
performance R&S®SMW-B710 R&S®SMW-B720
Ultra-low phase noise R&S®SMW-B10xx frequency option and R&S®SMW-B20xx frequency option and
R&S®SMW-B711 R&S®SMW-B721
Possible option combinations for an SMW with two RF paths
As phase noise in the signal also directly affect the residual EVM of a test set up, ideally, the signal generator
should be equipped with the Ultra-low phase noise option.
Latest Firmware with "Optimize EVM" function

From release firmware 5.00 and higher, the optimization of the preexisting excellent EVM performance of the
SMW is simplified by the addition of a one-button-click automated procedure.
Optimize EVM button in the I/Q Modulator menu of the signal generator
This procedure includes the previously existing Adjust I/Q Modulator Current Frequency and Optimize for
current settings functions plus some other adjustments to deliver best EVM by a mouse-click or automatically
after any change of the signal, frequency or power. If activated, a small icon is shown next to the frequency
field on the top of the instrument.
These optimizations do require some extra steps which add-up to the settling time of the generator.
4 Factors Influencing the EVM Measurement
In addition to the selected instrument hardware and firmware (see chapter above), also specific settings in
the measurement application and the signal parameters do affect the measured EVM values.
4.1 Measurement Settings

The following table gives an overview of standard-conform and EVM-optimized settings for channel
estimation and tracking in the WLAN measurement application. If non-standard EVM corrections are
activated during a test, the EVM result might be artificially improved and thus not usable to check compliance
against the standard.
Setting Standard conform setting Setting for best EVM Comment
802.11be, 36.3.19.4.4.
CHANNEL ESTIMATION
Channel Estimation Range Preamble Payload
Interpolation Wiener / None Wiener
Wiener Relative Delay Spread ON/OFF, tuned manually ON, tuned manually The optimum value depends on the channel
properties. Lower values increase smoothing.
TRACKING
Preamble Channel Estimation n.a. for 11be n.a. for 11be Only available for 802.11ac and n
Phase ON ON
Timing ON ON
Level OFF ON 802.11ax places different target EVM values
for amplitude drift compensation on and off
and 1024QAM; not applicable for 11be
I/Q Mismatch Compensation OFF ON
Pilots for Tracking According to standard According to standard If the pilot generation algorithms of the DUT
/ Detected has a problem, mode "Detected" might allow
synchronization
EHT-LTF Symbol Duration 6.4 μsec, 12.8 μsec 12.8 μsec Setting of waveform / signal generator
Important settings for standard conform EVM measurements
The 802.11 standard allows to "…Estimate the complex channel response coefficient for each of the
subcarriers…" for the EVM measurement. The channel estimation is done in frequency domain based on
known pilots in the Long Training Field in the preamble and missing subcarrier estimations are obtained by
interpolation.
The physical layer for EHT (802.11be) provides support for 6.4μs (2xLTF), and 12.8μs (4xLTF) LTF symbol
durations. This can be configured in the R&S®SMW200A 802.11 WLAN option, while the R&S®FSW WLAN
option automatically detects the LTF duration. Longer LTF durations allow for better equalizer training e.g. by
averaging the measurement values for the repeated LTFs.
The channel estimation interpolation setting is only available for 802.11ax and be. It applies filtering to the
channel, which reduces signal noise and influences the interpolation of gaps in the signal due to unused
subcarriers and thus the smoothing of the equalizer function during channel estimation.
The channel estimation is done on a per-subcarrier basis. Assuming independent noise on each subcarrier,
some averaging can reduce the noise in the frequency response calculation. This can reduce the EVM value
- except under low-noise conditions or when the real frequency response is unusually rough, which would
require a very "un-smooth" correction also.
The filter for the Wiener interpolation is defined by the Wiener Relative Delay Spread. By default, the filter is
determined automatically. Decrease this setting to finetune the EVM result if there is negligible delay spread -
for example for a wired connection.
The 802.11be standard provides the following additional specifics about the EVM measurement:
"… Local oscillator leakage that can potentially show up at the center frequency of the EHT PPDU tone plan
and within ±3 neighboring subcarriers shall be excluded from the computation of the transmitter modulation
accuracy test…"
"…The test shall be performed over at least 20 PPDUs … If the occupied RU has 26 tones, the PPDUs under
test shall be at least 32 data OFDM symbols long. For occupied RUs that have more than 26 tones, the
PPDUs under test shall be at least 16 data OFDM symbols long. Random data shall be used for the
symbols…"
While the 802.11b/g DSSS standards distinguish between Peak Error Vector and PPDU EVM, this is not the
case for 802.11be and the other OFDM based WLAN versions.
4.2 Crest Factor of the Test Signal

Like other communication standards, WLAN utilizes high order modulation and multiple carrier techniques.
The complexity of these signals can lead to high crest factors (peak voltage to average voltage ratio). High
crest factors in turn are associated with two problems in the signal conditioning chain:
─ High peak voltages challenge the linearity of power amplifiers (compression) and thus can cause
intermodulation effects.
─ Since the whole RF chain - including the D/A and A/D converters - has to be configured to handle
the peak voltages, the average voltages are converted with relatively low resolution. This leads to
higher quantization noise and thus reduced SNR.
Principal relationships of AD-converter and waveform parameters in a schematic spectrum-analyzer plot
Therefore, the crest factor is - together with the signal bandwidth - the main signal parameter which
influences the achievable residual EVM value. Performance comparisons always need to be based on
waveforms with similar characteristics regarding these two parameters.
The Overload indication of signal analyzers reacts on the peak in the measured signal. For 802.11ax and
11be signals, this peak is typically located in the preamble area:
Screenshot of the WLAN application showing an IF-Overload warning
In addition, the crest factor of a specific 11be waveform can also depend quite significantly on settings like
the Scrambler Init value which was set on the signal generator. The following figure shows EVM vs. Power
results for similar 11be waveforms mainly differing in the crest-factor value:
EVM versus Power measurements for signals with different crest-factors
The preamble part of the WLAN signal, which is responsible for the increased crest factor in the above
example, does not affect the EVM measurements. EVM is defined to only evaluate the payload of the
transmission. Furthermore, robust modulation schemes are used in the preamble part. These modulations
schemes can be demodulated correctly even if they are clipped slightly.
A clipping feature is implemented directly in the R&S®SMW-K147 IEEE 802.11be option of the signal
generator. This feature aims to clip all preamble parts exceeding the payload max peak and can be found in
the 11be as well as in the 11ax implementation. It harmonizes the crest factor between preamble and
payload.
Navigate to the General tab of the WLAN option, open the Filter/Clipping Settings menu, select the Clipping
tab and finally activate the Clip Signal Fields to Payload Max Peak option.
Clipping settings in the IEE802.11 WLAN menu of the signal generator
This immediately reduces the crest factor of the signal. At the same time, the achievable SNR for the signal
increases by approximately the same amount. The resulting improved EVM performance can be observed on
the signal analyzer.
Clip Signal Fields to Payload Max Peak OFF Clip Signal Fields to Payload Max Peak ON
Crest Factor 13.01dB Crest Factor 11.76dB

EVM All -49.02dB EVM All -49.63dB
Signal without and with "Clip Signal Fields to Payload Max Peak" activated
Since the crest factor is dependent on many different parameters of the signal, the EVM gain achievable by
this option is different for every signal.
To ensure valid measurements, it is generally recommended to absolutely avoid overload conditions on the
signal analyzer (indicated by the red OVLD status display). As for 11be signals the peak signal is typically
located in the preamble area, while the EVM is calculated over the payload fields, a small overload can result
in a better EVM value, because only some parts of the preamble are clipped slightly. But this always needs
special attention and validation.
4.3 EVM Independence from Modulation Order & Burst Length

Modulation Order
OFDM systems add up a large number of individually modulated subcarriers at the transmitter. Without crest-
factor-reduction these summed-up signals have power characteristics similar to additive white gaussian noise
with high peak-to-average power ratios above 10dB.
The crest factor of M-QAM modulated signals does not constantly increase with growing modulation M order
but saturates at sqrt(3) or 4.8dB for high modulation orders (theoretical value for quadratic constellation and
equally likely constellation points).
Also, the properties of the underlaying modulation are "scrambled" by the IFFT (Inverse Fast Fourier
Transformation) operation which is applied to switch from the frequency-domain to the time-domain signal.
These two facts help to explains why the modulation-order of the individual subcarriers does not affect the
final crest factor of the time-domain signal much.
This is shown exemplarily in the following table, where a 320MHz wide 11be waveform with different
modulation schemes for the data stream was analyzed. Clip Signal Fields to Payload Max Peak is enabled
on the generator to measure independent of the specifics of the signaling fields:
Modulation Measured CF [dB] Measured EVM All @-10dBm set power [dB]
QPSK 11.62 -49.7
16 QAM 11.79 -49.8
64 QAM 11.65 -49.7
256 QAM 11.63 -49.6
1024 QAM 11.96 -49.8
4096 QAM 11.88 -49.7
EVM results for an 11be 320MHz signal using different subcarrier modulations at a fixed power level; Optimize EVM activated on signal
generator and signal analyzer; >~ 100 symbols; Clipping Signal Fields to Payload Max Peak set on generator;
A high signal length of around 100 symbols was chosen to increase the probability of high instantaneous
signal amplitudes in the captured signal.
Burst Length
The number of symbols in a PPDU does not affect the average EVM, as the EVM vs. Symbol result should
not drift over time (i.e. versus the symbol number).
EVM versus Symbol plot showing no drift over time
As the timing synchronization is based on the preamble symbols, one could expect a sampling point drift for
very long burst length. But as time tracking is required for a standard conform EVM measurement, a possible
drift would be compensated. Disabling the time tracking in the above measurement directly results in a visible
drift - if the reference clocks of the instruments are not coupled.
EVM versus Symbol plot: drift over time due to disabled time tracking and uncoupled reference clocks
The standard demands a minimum of 16 symbols. If less than 5 symbols are found in the payload, a warning
is displayed on the signal analyzer:
Short Payload warning for signals with less than 5 symbols
For such short payload lengths, the compensation of certain error terms (frequency/phase offsets) and the
phase training become less accurate.
4.4 4096-QAM EVM Limitation for Low-SNR Conditions

To increase the maximum possible throughput, 802.11be introduced the 4096 QAM modulation of the
subcarriers. Up to 802.11ax, 1024 QAM was the maximum applied modulation order.
With the increasing number of constellation points, these points get closer and closer in the constellation
diagram.
Typical 4096 QAM constellation
To calculate the EVM value, each sampled point has to be assigned to an ideal constellation point. Based on
the "distance" to this ideal point, the EVM is calculated.
As the option R&S®FSW-K91 does not have additional information about the measured signal, this decision
can only be based on the vicinity of the sampling point to a constellation point. The closest "neighbor"
constellation point will be selected for the EVM calculation.
The deviation of the sampling points from the ideal constellation points can be caused e.g. by non-linearity of
an amplifier, spurious interferers, IQ gain imbalance or by noise added to the intended signal. These effects
can create different types of deformations of the constellation.
Assuming only wideband, uncorrelated noise as a source of increased EVM for the following, the "spread" of
the sampled points increases with decreasing signal-to-noise ratio and might be in the range or even bigger
than the "distance" of the constellation points:
EVM -46dB EVM -39dB EVM -36dB
Top-right segment of a typical 4096 QAM constellation with decreasing signal to noise ratio
With decreasing SNR, this approach leads to more and more wrong assignments of the sampled data to
simply the closest constellation point instead of the correct one. Therefore, the measured EVM value will be
unrealistically good under low SNR conditions and saturate at some point.
In opposite to the R&S®FSW-K91 WLAN option (nearest constellation point decision), the R&S®FSW-K18
Amplifier option calculates the EVM based on the known reference waveform. This means, that in this option
the calculation of the EVM is based on the knowledge of the sent signal. Therefore, no wrong decisions are
taken also under bad signal to noise conditions and the EVM value does not saturate at some point.
Comparing both approaches clearly shows the difference in the low signal power region:
EVM versus Power plot: nearest-constellation-point EVM calculation (FSW-K91) versus known-reference EVM calculation (FSW-K18)
COARSE ESTIMATION OF THE LIMITATIONS FOR NEAREST-POINT EVM CALCULATION
The following applies for EVM calculations based on a nearest-constellation-point-decision for QAM
modulated signals, assuming only fully quadratic constellations.
From a certain distance of a real sampling point from the ideal constellation point (let's assume the red dot in
the figure below should be the correct constellation point), the sampling point will be wrongly assigned to a
neighbor constellation point if no further information about the real waveform is available.
Sampling points with a distance x/2 like the red diamond, will be assigned to a wrong constellation point if the
angle is 0, 90,180 or 270 degrees relative to the red dot.
Sampling points with a distance bigger than the red circle will always be assigned to a wrong constellation
point, independent of their angle.
This is valid only for sampling points "inside" the constellation. If the sampling point lies outside the
constellation area (like the green dot), bigger distances are possible without a wrong decision, as fewer
neighbor points exist. This affects modulations with lower order in the following estimation more than high-
order modulations - assuming a more or less equal distribution of sampling points across all the possible
constellation points.
16 QAM Example
Distances bigger than this can lead to wrong
decisions, depending on the angle
Distances bigger than this lead to 100% wrong
nearest-point decisions
2𝑎 √2
𝑥= =
𝑁 𝑁
𝑁 = √𝑀𝑜𝑑𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑂𝑟𝑑𝑒𝑟 − 1
1
𝑎2 + 𝑎2 = 1 → 𝑎 = √
2
1 2 1 2
𝑦 2 = ( 𝑥) + ( 𝑥)
2 2
1
EVM closest-point decision for different sampling point locations
𝑦 = √ 𝑥2
2
From the above sketch, the following numbers can be derived for the estimated worst case EVM:
QAM x [linear] X/2 [dB] y [linear] y [dB] Worst case EVM [dB] measured while
Order Possibly wrong decision 100% wrong decision increasing AWGN level on generator
16 0.4714 -12.5 0.3 -9.5 -12
256 0.0943 -26.5 0.066 -23.5 -21
1024 0.0456 -32.8 0.0323 -29.8 -29
4096 0.0224 -39.0 0.0159 -36 -35
Estimation of worst case EVM boundaries for closest-point-decision based EVM; only considering points inside the constellation
These number can only be used to give a rough idea of the possible EVM range for different QAM
modulations. A true signal always has a certain amplitude distribution. Also, other effects like the type of
distortion might affect the EVM measurement.
In the last column, numbers of real EVM measurements are provided for comparison. To simulate low SNR
scenarios, the signal was charged with an increasing level of Additive White Gaussian Noise. The highest
EVM reading, that was obtained under increasing noise level, is given in the table.
5 How to Set Up for Best EVM Performance
So far, the following important topics for accurate EVM measurements were discussed in this document:
─ Instrument selection (model, option, firmware version)
─ Signal/waveform selection (crest factor)
─ Correct measurement parameters
In this chapter some additional points will be highlighted, especially with respect to the optimal hardware
settings of the signal generator and signal analyzer.
To cover a wide range of frequencies, bandwidths, standards and applications, the generator as well as the
analyzer comprise complex hardware for optimal signal conditioning under changing circumstances and
requirements. This hardware has to be configured appropriately to achieve the best possible performance.
5.1 Signal Generator
5.1.1 Digital Attenuation
The generator offers the option to add a negative digital attenuation to the baseband. This amplification in the
digital domain can be used to trade-in a bit of linearity versus an improved SNR. Similar to the effect
discussed in chapter 4.2, this may lead to an overall better EVM result while degrading signal parameters
which are less important for this specific measurement.
With wider bandwidths, also the noise in the signal path is increasing in general. This leads to a certain
bandwidth dependency of the optimal setting. It can be advantageous to increase the baseband signal level
for wider bandwidths while decreasing it for low-bandwidth signals.
5.1.2 Baseband Offset
Every mixer-based hardware shows carrier leakage at the center frequency. To further optimize the signal
characteristics, it can be helpful to shift the baseband signal with a user-defined baseband frequency offset to
a different center frequency. To allow this, the complex I/Q bandwidth of the shifted useful signal must not
exceed the total available baseband bandwidth of the signal generator.
Shifting the baseband signal to a different center frequency can eliminate the carrier leakage impact.
Principle of a baseband frequency offset
When a baseband frequency offset is applied, the RF spectrum shows the wanted signal offset from the LO
leakage peak and a suppressed sideband signal at the image frequency.
To shift the generated baseband signal, select Baseband > Baseband Offsets > Frequency Offset. This offset
has to be taken in to account when calculating the correct receive frequency for the analyzer.
0MHz BB offset, generator set frequency 2.4GHz 400MHz BB offset, generator set frequency 2.0GHz
Effect of a baseband frequency offset
5.1.3 Optimize-EVM Feature
From release firmware 5.00 and higher, the optimization of the preexisting excellent EVM performance of the
R&S®SMW200A Vector Signal Generator is simplified by the addition of a one-button-click automated
procedure.
Please see chapter 3.2 for additional details.
5.2 Signal Analyzer
5.2.1 Auto-Level Enhancements and Optimize EVM Feature
The signal conditioning in the RF front-end of signal analyzers is crucial to achieve the best performance with
respect to image-suppression, noise-floor, dynamic range and other RF-key parameters.
Precise signal levelling is especially important for complex measurements like Error-Vector-Magnitude. To
minimize the measurement uncertainty from test system contributions over a wide range of different power
levels, the RF front-end needs to be adapted continuously according to the signal characteristics, signal
power and frequency - ideally using an automatic levelling algorithm.
The firmware release 5.00SP3 introduces an improved auto-level algorithm for the 802.11be measurement
application in combination with the bandwidth options R&S®FSW-B320 and R&S®FSW-B512.
Firmware 5.10 rolls-out the additional auto-levelling feature Optimize EVM.
Optimize EVM drop down menu
The Optimize EVM drop down menu defines whether an optional, iterative search is performed to determine
the required settings for minimum residual EVM with increased accuracy. If enabled, the reference level,
preamplifier and, optionally, attenuation are configured:
─ Off: switches off the additional auto-level functionality; in this case, the default auto-levelling is NOT
using data from the demodulated signal
─ Full: An optional iterative search for minimum residual EVM is performed for the available
preamplifier and attenuation settings using also results from several EVM measurements; therefore,
the R&S®FSW-K91 has to be configured to correctly demodulate the signal for this mode;
─ PA only: Only the preamplifier settings are optimized using also results from several EVM
measurements with different settings; the attenuation setting is not changed; the R&S®FSW-K91
has to be configured to correctly demodulate the signal for this mode;
For signals with 160MHz or 320MHz bandwidth the Low Noise Attenuation optimization mode is set
automatically.
The following figure gives an example of the effect of the Optimize-EVM feature on an EVM vs. Power
bathtub curve:
EVM versus power; random set-power sequence during measurement; 320MHz 802.11be signal; Optimize-EVM Off, On and Pa-Only
The improved auto-levelling is especially important around the preamplifier-switching points (~-32dBm in the
above figure) but also helps at other power settings to achieve an optimal EVM result.
5.2.2 R&S®FSW-K18 Option with IQ Averaging
The R&S®FSW-K18 Amplifier option was introduced already in chapter 4.4 to analyze the EVM numbers for
low SNR scenarios. The IQ-Averaging feature of this option can be utilized to further analyze and separate
several contributors to the residual EVM by e.g. averaging out some of the uncorrelated noise sources.
Special attention and validation is required when IQ-averaging is applied to correctly contextualize the
achieved EVM results. They might not be in accordance to the standard test requirements.
5.2.3 Modulation Measurement Optimizer Tool
The Rohde & Schwarz Modulation Measurement Optimizer (MMO) tool was developed to support the
automatic optimization of the signal analyzer RF front-end settings to minimize the instrument contribution to
the residual EVM.
https://www.rohde-schwarz.com/de/applikationen/modulation-measurement-optimizer-application-
note_56280-1168512.html
The 11be application in the latest firmware versions with optimized auto-levelling / Optimize EVM feature
(see chapter 3.1) is already able to automatically determine optimized settings for the RF front end for best
EVM for the recommended instruments.
The MMO tool might be useful though for hardware/firmware which is not supported by the new auto-level
algorithm or to gain further insights in the challenges of auto-levelling.
6 Other 11be Transmitter Test Items
Spectrum Emission Mask (SEM) with punctured channel
Signal setup with punctured 40MHz block
If there are punctured subchannels in the signal bandwidth, the overall spectral mask is constructed by
overlaying the standard mask for the unpunctured signal and the puncture masks for the specific puncture
case (802.11be amendment chapter 36.3.19.2).
The Multi-SEM mode (using several Sub-Blocks for the mask definition) of the SEM measurement in the
Spectrum mode can be used to create a combined mask for punctured signals:
SEM measurement: RU Size 3x994+484 MRU Index 4, 40MHz puncturing 111x1111
Spectral flatness
The Spectrum Flatness trace of the R&S®FSW-K91 WLAN option is derived from the magnitude of the
estimated channel transfer function. Since this estimated channel is calculated from all payload symbols of
the PPDU, it represents a carrier-wise mean gain of the channel.
Spectrum Flatness measurement
Transmit center frequency leakage
Center frequency leakage measurement
Center frequency leakage is the portion of the transmitter energy that leaks through the transmitter
components. This unwanted energy appears in the modulated signal at the signal center frequency which
can lead to bad demodulation performance - depending on the receiver implementation.
Often OFDM-based receiver systems utilize some way to remove the carrier leakage. IEEE mandates that
the transmitter center frequency leakage does not exceed certain limits.
The center frequency leakage can be observed either directly in the FFT-spectrum or as I/Q Offset
measurement in the result summary window.
IQ offsets are typically removed during demodulation in the measurement applications before the
constellation diagram is plotted. The following figure illustrates the effect of IQ offsets on a 16 QAM signal
using the generic R&S®FSW-K70 Vector Signal Analysis option.
Constellation without IQ offset compensation Spectrum degraded by IQ offset
Example of a 16 QAM signal with 10% I and Q offset
7 Literature
[1] IEEE 802.11be Technology Introduction, White Paper: Rohde & Schwarz, 2022.
[2] White paper: IEEE 802.11ax technology introduction (search for "1MA192"), Rohde & Schwarz.
[3] Rohde & Schwarz, "Even Faster, An outline of the upcoming 802.11be WLAN standard," NEWS, no.
224, pp. 19-21, 2021.
8 Ordering Information
Designation Type Order No.
WLAN 802.11be measurements R&S®FSW-K91BE 1350.6730.02
WLAN 802.11be measurements R&S®FSV3-K91BE 1346.4966.02
IEEE 802.11be measurements R&S®VSE-K91BE 1345.1428.06
IEEE 802.11be R&S®SMW-K147 1413.6677.02
IEEE 802.11be (WinIQSIM2™) R&S®SMW-K447 1413.6683.02
Rohde & Schwarz
The Rohde & Schwarz electronics group offers
innovative solutions in the following business fields: test
and measurement, broadcast and media, secure
communications, cybersecurity, monitoring and network
testing. Founded more than 80 years ago, the
independent company which is headquartered in
Munich, Germany, has an extensive sales and service
network with locations in more than 70 countries.
www.rohde-schwarz.com
Certified Quality Management
ISO 9001
Rohde & Schwarz training

www.training.rohde-schwarz.com
PAD-T-M: 3572.7186.00/01.00/EN
Rohde & Schwarz customer support

www.rohde-schwarz.com/support
R&S® is a registered trademark of Rohde & Schwarz GmbH & Co. KG

Trade names are trademarks of the owners.
1EF114 | Version 0e | 08.2022
Application Note | Hints for IEEE 802.11be EVM Measurements
Data without tolerance limits is not binding | Subject to change
© 2022 Rohde & Schwarz GmbH & Co. KG | 81671 Munich, Germany
www.rohde-schwarz.com

1EF114_e_802_11be_EVM

Uploaded by

Copyright:

Available Formats

1EF114_e_802_11be_EVM

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

1EF114_e_802_11be_EVM

Uploaded by

Copyright:

Available Formats

Application Note

HINTS FOR IEEE 802.11BE EVM

M. Breinbauer | 1EF114 | Version 0e | 08.2022

2.1 IEEE 802.11be Goals and Feature Summary

802.11n (HT) 802.11ac (VHT) 802.11ax (HE) 802.11be (EHT)

PHY parameter values for 11n, 11ac, 11ax and 11be

2.2 Achieve Extremely High Throughput and More

2.2.1 Up to 320 MHz Channel Bandwidth

320 MHz channel allocation based on FCC in the 6 GHz band

2.2.3 Restricted Target Wake Time

2.2.4 Multi-AP Operation

2.3 IEEE 802.11be Physical Layer

2.3.1 New RU and Tone Plans

2.3.3 Preamble / Subchannel Puncturing

2.3.4 PPDU Formats

PPDU Field Descriptions

Two PPDU formats are defined in EHT:

EHT Multi User PDDU Format

EHT Trigger Based PDDU Format

2.3.5 EHT Preamble: Designed for the Future

U-SIG filed content for EHT MU PDDU and EHT TB PDDU

2.4.1 Transmitter Requirements

2.4.1.1 Transmit Spectral Mask

2.4.1.1.1 Spectrum Mask for a PPDU with Punctured Channel(s)

2.4.1.2 Spectral Flatness

2.4.1.3 Transmitter Modulation Accuracy

2.4.1.3.1 Transmit Center Frequency Leakage

2.4.1.3.2 Transmitter Constellation Error

ETH transmitter constellation error specification

-38 1,259 2 -40 1,000 1,608 0,349 -35,876 2,124

-38 1,259 4 -42 0,794 1,489 0,230 -36,545 1,455

-38 1,259 6 -44 0,631 1,408 0,149 -37,027 0,973

-38 1,259 8 -46 0,501 1,355 0,096 -37,361 0,639

-38 1,259 10 -48 0,398 1,320 0,061 -37,586 0,414

-38 1,259 12 -50 0,316 1,298 0,039 -37,734 0,266

-38 1,259 14 -52 0,251 1,284 0,025 -37,830 0,170

-38 1,259 16 -54 0,200 1,275 0,016 -37,892 0,108

-38 1,259 18 -56 0,158 1,269 0,010 -37,932 0,068

-38 1,259 20 -58 0,126 1,265 0,006 -37,957 0,043

-38 1,259 22 -60 0,100 1,263 0,004 -37,973 0,027

-38 1,259 24 -62 0,079 1,261 0,003 -37,983 0,017

-38 1,259 26 -64 0,063 1,261 0,002 -37,989 0,011

-38 1,259 28 -66 0,050 1,260 0,001 -37,993 0,007

-38 1,259 30 -68 0,040 1,260 0,001 -37,996 0,004

Target Real SA EVM SA EVM SA EVM Total EVM Total EVM

-32 2,512 10 -42 0,794 2,634 0,123 -31,586 0,414

-34 1,995 10 -44 0,631 2,093 0,097 -33,586 0,414

-36 1,585 10 -46 0,501 1,662 0,077 -35,586 0,414

-38 1,259 10 -48 0,398 1,320 0,061 -37,586 0,414

-40 1,000 10 -50 0,316 1,049 0,049 -39,586 0,414

-42 0,794 10 -52 0,251 0,833 0,039 -41,586 0,414

-44 0,631 10 -54 0,200 0,662 0,031 -43,586 0,414

-46 0,501 10 -56 0,158 0,526 0,024 -45,586 0,414

-48 0,398 10 -58 0,126 0,418 0,019 -47,586 0,414

-50 0,316 10 -60 0,100 0,332 0,015 -49,586 0,414

-52 0,251 10 -62 0,079 0,263 0,012 -51,586 0,414

-54 0,200 10 -64 0,063 0,209 0,010 -53,586 0,414

-56 0,158 10 -66 0,050 0,166 0,008 -55,586 0,414

-58 0,126 10 -68 0,040 0,132 0,006 -57,586 0,414

-60 0,100 10 -70 0,032 0,105 0,005 -59,586 0,414