Fundamentals of Real Time Spectrum Analysis

Primer
Fundamentals of Real-Time
Spectrum Analysis
Fundamentals of Real-Time Spectrum Analysis
Primer
Table of Contents
Chapter 1: Introduction and Overview . . . . . . . . 3-8 Digital Phosphor Display . . . . . . . . . . . . . . . . . . . . .26
The Evolution of RF Signals . . . . . . . . . . . . . . . . . . . . . . .3 The DPX Display Engine . . . . . . . . . . . . . . . . . . . . .27
Modern RF Measurement Challenges . . . . . . . . . . . . . . .4 The DPX Transform Engine . . . . . . . . . . . . . . . . . .29
A Brief Survey of Instrument Architectures . . . . . . . . . . .5 DPX Density Measurements . . . . . . . . . . . . . . . . . .29
The Swept Spectrum Analyzer . . . . . . . . . . . . . . . . .5 Timing and Triggers . . . . . . . . . . . . . . . . . . . . . . . . . . . .30
Vector Signal Analyzers . . . . . . . . . . . . . . . . . . . . . .7 Real-Time Triggering and Acquisition . . . . . . . . . . .31
Real-Time Spectrum Analyzers . . . . . . . . . . . . . . . .7 Triggering in Systems with Digital Acquisition . . . . .32
Trigger Modes and Features . . . . . . . . . . . . . . . . . .32
Chapter 2: How Does the Real-Time Spectrum RSA Trigger Sources . . . . . . . . . . . . . . . . . . . . . . .33
Analyzer Work? . . . . . . . . . . . . . . . . . . . . . . . . . 9-37 Constructing a Frequency Mask . . . . . . . . . . . . . . .34
RF/IF Signal Conditioning . . . . . . . . . . . . . . . . . . . . . . . .9 Demodulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
Input Switching and Routing Section . . . . . . . . . . .10 Amplitude, Frequency and Phase Modulation . . . .35
RF and Microwave Sections . . . . . . . . . . . . . . . . . .10 Digital Modulation . . . . . . . . . . . . . . . . . . . . . . . . . .36
Frequency Conversion/IF Section . . . . . . . . . . . . . .11 Power Measurements and Statistics . . . . . . . . . . .37
Digital Signal Processing (DSP) Concepts . . . . . . . . . . .12
Chapter 3: Real-Time Spectrum Analyzer
Digital Signal Processing Path in Real-Time
Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . 38-46
Spectrum Analyzers . . . . . . . . . . . . . . . . . . . . . . . .12
IF Digitizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13 Real-Time Spectrum Analyzer Measurement Types . . . .38
Corrections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13 Frequency Domain Measurements . . . . . . . . . . . . .38
Digital Downconverter (DDC) . . . . . . . . . . . . . . . . .14 DPX Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . .39
IQ Baseband Signals . . . . . . . . . . . . . . . . . . . . . . .14 Time Domain Measurements . . . . . . . . . . . . . . . . .39
Decimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15 Modulation Domain Measurements . . . . . . . . . . . .42
Decimation Filtering . . . . . . . . . . . . . . . . . . . . . . . .15 Code Domain Measurements . . . . . . . . . . . . . . . . .44
Transforming Time Domain Waveforms Statistical Domain Measurements . . . . . . . . . . . . .45
to the Frequency Domain . . . . . . . . . . . . . . . . . . . . . . .16 Chapter 4: Applications Using the RSA . . . . . . . 47-49
Real-Time Spectrum Analysis . . . . . . . . . . . . . . . . .17
Pulse Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . .47
Discovering and Capturing Transient Events . . . . . .17
Radar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
Relating RSA to Swept Spectrum Analyzer . . . . . .18
RFID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
RBW on the Real-Time Spectrum Analyzer . . . . . .18
Spectrum Management/Surveillance . . . . . . . . . . . . . . .48
Windowing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
Radio Communications . . . . . . . . . . . . . . . . . . . . . . . . .48
Discrete Fourier Transforms (DFT) in the
Power Amplifier Testing . . . . . . . . . . . . . . . . . . . . . . . . .49
Real-Time Spectrum Analyzer . . . . . . . . . . . . . . . .20
Microphonics and Phase-Hits Analysis . . . . . . . . . . . . .49
Digital Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .22
EMI/EMC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49
Finite Impulse Response (FIR) Filters . . . . . . . . . . .22
Frequency Response vs. Impulse Response . . . . .22 Chapter 5: Terminology . . . . . . . . . . . . . . . . . . . . 50-51
Numerical Convolution . . . . . . . . . . . . . . . . . . . . . .23 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50
DPX® Technology: a Revolutionary Tool Acronym Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . .51
for Signal Discovery . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
2 www.tektronix.com/rsa
Primer
Chapter 1: Introduction and Overview

The Evolution of RF Signals be found in almost any imaginable location: consumer
Engineers and scientists have been looking for innovative electronics in homes, medical devices in hospitals, industrial
uses for RF technology ever since the 1860s, when James control systems in factories, and even tracking devices
Clerk Maxwell mathematically predicted the existence of implanted underneath the skin of livestock, pets, and people.
electromagnetic waves capable of transporting energy As RF signals have become ubiquitous in the modern world,
across empty space. Following Heinrich Hertz’s physical so too have problems with interference between the devices
demonstration of “radio waves” in 1886, Nikola Tesla, that generate them. Products such as mobile phones that
Guglielmo Marconi, and others pioneered ways of operate in licensed spectrum must be designed not to
manipulating these waves to enable long distance transmit RF power into adjacent frequency channels and
communications. At the turn of the century, the radio cause interference. This is especially challenging for complex
had become the first practical application of RF signals. multi-standard devices that switch between different modes
Over the next three decades, several research projects of transmission and maintain simultaneous links to different
were launched to investigate methods of transmitting network elements. Devices that operate in unlicensed
and receiving signals to detect and locate objects at great frequency bands must be designed to function properly in
distances. By the onset of World War II, radio detection the presence of interfering signals, and are legally required
and ranging (also known as radar) had become another to transmit in short bursts at low power levels. These new
prevalent RF application. digital RF technologies that involve the combination of
Due in large part to sustained growth in the military and computers and RF include wireless LANs, cellular phones,
communications sectors, technological innovation in RF digital TV, RFID and others. These, combined with new
accelerated steadily throughout the remainder of the 20th advances in Software Defined Radio (SDR) and Cognitive
century and continues to do so today. To resist interference, Radio (CR) provide a new path forward and will fundamentally
avoid detection, and improve capacity, modern radar change spectrum allocation methodologies resulting in
systems and commercial communications networks have increased efficiency in the way that the RF spectrum,
become extremely complex, and both typically employ one of the scarcest commodities, is utilized.
sophisticated combinations of RF techniques such as To overcome these evolving challenges, it is crucial for
complex and adaptive modulation, bursting and frequency today’s engineers and scientists to be able to reliably
hopping. Designing these types of advanced RF equipment detect and characterize RF signals that change over time,
and successfully integrating them into working systems are something not easily done with traditional measurement
extremely complicated tasks. tools. To address these problems, Tektronix has designed
At the same time, the increasingly widespread success the Real-Time Spectrum Analyzer (RSA), an instrument
of cellular technology and wireless data networks that can discover elusive effects in RF signals, trigger on
combined with the advancing state of semiconductor those effects, seamlessly capture them into memory, and
technology and packaging has caused the cost of basic analyze them in the frequency, time, modulation, statistical
RF components to drop significantly over time. This has and code domains.
enabled manufacturers outside of the traditional military This document describes how the RSA works and provides
and communications realms to embed relatively simple a basic understanding of how it can be used to solve many
RF devices into all sorts of commodity products. RF measurement problems associated with modern RF signals.
transmitters have become so pervasive that they can
www.tektronix.com/rsa 3
Primer
Modern RF Measurement Challenges Capturing burst transmissions, glitches,

switching transients
Given the challenge of characterizing the behavior of today’s
RF devices, it is necessary to understand how frequency, Characterizing PLL settling times, frequency drift,
amplitude, and modulation parameters behave over short microphonics
and long intervals of time. Traditional tools like Swept Capturing spread-spectrum and frequency-hopping signals
Spectrum Analyzers (SA) and Vector Signal Analyzers (VSA) Monitoring spectrum usage, detecting rogue transmissions
provide snapshots of the signal in the frequency domain
Testing and diagnosing transient EMI effects
or the modulation domain. This is often not enough
information to confidently describe the dynamic nature Characterizing time-variant modulation schemes
of modern RF signals. Isolating software and hardware interactions
Consider the following challenging measurement tasks: Each measurement involves RF signals that change over
Discovery of rare, short duration events time, often unpredictably. To effectively characterize these
signals, engineers need a tool that can discover elusive
Seeing weak signals masked by stronger ones
events, effectively trigger on those events and isolate them
Observing signals masked by noise into memory so that the signal behavior can be analyzed
Finding and analyzing transient and dynamic signals in the frequency, time, modulation, statistical, and
code domains.
Primer
Fa Fb Fa Fb
Figure 1-1. The Swept Spectrum Analyzer steps across a series of frequency segments, often missing important transient events that occur outside the current sweep band
highlighted in tan segment Fb on the right.
A Brief Survey of Instrument Architectures high dynamic range, its disadvantage is that it can only
calculate the amplitude data for one frequency point at a
To learn how the RSA works and understand the value of
time. This approach is based on the assumption that the
the measurements it provides, it is helpful to first examine
analyzer can complete at least one sweep without there
two other types of traditional RF signal analyzers: the
being significant changes to the signal being measured.
Swept Spectrum Analyzers (SA) and the Vector Signal
Consequently, measurements are only valid for relatively
Analyzer (VSA).
stable, unchanging input signals. If there are rapid changes
The Swept Spectrum Analyzer in the signal, it is statistically probable that some changes
will be missed. As shown in Figure 1-1, the SA is looking
The swept-tuned, superheterodyne spectrum analyzer is
at frequency segment Fa while a momentary spectral event
the traditional architecture that first enabled engineers to
occurs at Fb (diagram on left). By the time the sweep arrives
make frequency domain measurements several decades
at segment Fb, the event has vanished and is not detected
ago. Originally built with purely analog components, the SA
(diagram on right). The SA architecture does not provide a
has since evolved along with the applications that it serves.
reliable way to discover the existence of this kind of transient
Current generation SAs include digital elements such as
signal, thus contributing to the long time and effort required to
ADCs, DSPs, and microprocessors. However, the basic
troubleshoot many modern RF signals. In addition to missing
swept approach remains largely the same and is best suited
momentary signals, the spectrum of impulse signals such
for observing controlled, static signals. The SA makes
as those used in modern communications and radar may be
power vs. frequency measurements by downconverting the
misrepresented as well. SA architectures cannot represent
signal of interest and sweeping it through the passband of a
the occupied spectrum of an impulse without repetitive
resolution bandwidth (RBW) filter. The RBW filter is followed
sweeps. One also needs to pay special attention to sweep
by a detector that calculates the amplitude at each frequency
rate and resolution bandwidth.
point in the selected span. While this method can provide
Primer
Display
Low-Pass Resolution Envelope Video
RF Downconverter Bandwidth Detector Bandwidth
Attenuator Filter (SLVA) Filter
Y
Input
YIG
Pre-Selector X
Swept Tuned
Local
Oscillator
Sweep
Generator
a) Swept Tuned Spectrum Analyzer (SA)
Display
Low-Pass
RF Downconverter
Digital
Attenuator IF Filter P X-Y
Filter
Micro-
Input ADC Memory
Processor
Band-Pass Local Acquisition Bandwidth Post Capture Processing

Oscillator
b) Vector Signal Analyzer (VSA)
Modern FFT-Based Analyzers

RF Downconverter Real-Time Digital Post Capture
Low-Pass Downconvert
Attenuator IF Filter & Filter Capture Displays
Amp./Phase DDC/
Input ADC Memory
Corrections Decimation
X-Y
Ext
Band-Pass
Local
Oscillator 100 or 300 Msps
r
Free i
Processor
Live Signal
Micro-
Run
Processing
110 MHz
Acquisition Bandwidth DPX
Real Time Engine
Real-Time Display X-Y

IQ out trigger
(option 05) Analysis
Real-Time Bandwidth Display Processing
c) Real-Time Spectrum Analyzer (RSA6000 Series + Option 110)
Figures 1-2 a, b, c. Simplified Block Diagram of Swept Spectrum Analyzer (a), Vector Signal Analyzer (b), and Real-Time Spectrum Analyzer (c).
Primer
Figure 1-2a depicts a typical modern SA architecture. Even Real-Time Spectrum Analyzers
though modern SA have replaced analog functionality with
The term “real-time” is derived from early work on digital
digital signal processing (DSP), the fundamental architecture
simulations of physical systems. A digital system simulation
and its limitations remain.
is said to operate in real-time if its operating speed matches
Vector Signal Analyzers that of the real system which it is simulating.
Analyzing signals carrying digital modulation requires vector To analyze signals in real-time means that the analysis
measurements that provide both magnitude and phase operations must be performed fast enough to accurately
information. A simplified VSA block diagram is shown in process all signal components in the frequency band of
Figure 1-2b. interest. This definition implies that we must:
A VSA digitizes all of the RF power within the passband Sample the input signal fast enough to satisfy Nyquist
of the instrument and puts the digitized waveform into criteria. This means that the sampling frequency must
memory. The waveform in memory contains both the exceed twice the bandwidth of interest.
magnitude and phase information which can be used by Perform all computations continuously and fast enough
DSP for demodulation, measurements or display processing. such that the output of the analysis keeps up with the
Within the VSA, an ADC digitizes the wideband IF signal, and changes in the input signal.
the downconversion, filtering, and detection are performed Discover, Trigger, Capture, Analyze
numerically. Transformation from time domain to frequency The Real-Time Spectrum Analyzer (RSA) architecture is
domain is done using FFT algorithms. The VSA measures designed to overcome the measurement limitations of the
modulation parameters such as FM deviation, Code Domain SA and VSA to better address the challenges associated
Power, and Error Vector Magnitude (EVM and constellation with transient and dynamic RF signals as described in the
diagrams). It also provides other displays such as channel previous sections. The RSA performs signal analysis using
power, power versus time, and spectrograms. real-time digital signal processing (DSP) that is done prior
While the VSA has added the ability to store waveforms to memory storage as opposed to the post-acquisition
in memory, it is limited in its ability to analyze transient processing that is common in the VSA architecture. Real
events. In the typical VSA free run mode, signals that time processing allows the user to discover events that are
are acquired must be stored in memory before being invisible to other architectures and to trigger on those events
processed. The serial nature of this batch processing allowing their selective capture into memory. The data in
means that the instrument is effectively blind to events memory can then be extensively analyzed in multiple
that occur between acquisitions. Single or infrequent domains using batch processing. The real-time DSP
events cannot be discovered reliably. Triggering on these engine is also used to perform signal conditioning,
types of rare events can be used to isolate these events calibration and certain types of analysis.
in memory. Unfortunately VSAs have limited triggering
capabilities. External triggering requires prior knowledge
of the event in question which may not be practical. IF
level triggering requires a measurable change in the total
IF power and cannot isolate weak signals in the presence
of larger ones or when the signals change in frequency
but not amplitude. Both cases occur frequently in today’s
dynamic RF environment.
Primer
Frame Frame Frame Frame
VSA
Not Real-Time Missed
Input
FFT FFT
Time Sampled
Time
FFT FFT FFT FFT

Frequency
Real-Time Domain
RSA
Processing Time < Acquisition time!
Figure 1-3. VSA processing vs. Real-Time Spectrum Analyzers real-time engine processing.
The heart of the RSA is a real-time processing block as Triggering in the frequency domain with Frequency Mask
shown in Figure 1-2c (on page 6). Similar to the VSA, a wide Trigger (FMT)
capture bandwidth is digitized. Unlike the VSA, the real-time Triggering on user specified bandwidths with filtered
engine operates fast enough to process every sample power trigger
without gaps as shown in Figure 1-3. Amplitude and phase
Real-time demodulation allowing the user to “listen” to a
corrections that compensate for analog IF and RF responses
particular signal within a busy band
can be continuously applied. Not only can the data stored in
memory be fully corrected, but this enables all subsequent Digital IQ streaming of digitized data allows the uninterrupted
real-time processing to operate on corrected data as well. output of the signal for external storage and processing
The real-time engine enables the following features that The real-time engine not only enables signal discovery and
address the needs of modern RF analysis: trigger, but it also performs many of the repetitive signal
Real-time correction for imperfections in the analog processing tasks, freeing up valuable software-based
signal path resources. Like the VSA, the RSA offers post-acquisition
DPX® Live RF display allows the discovery of events analysis using DSP. It can perform measurements in
missed by swept SAs and VSAs multiple time-correlated domains that can be displayed
simultaneously.
DPX Density™ measurements and triggering defined by the
persistency of a signal’s occurrence
Advanced time-qualified triggering, such as runt triggering,
usually found in performance oscilloscopes
Primer
LF or IQ section (if present)

LP Filter
Input Switching
I
Input ADC
Digitized
Baseband
LP Filter
Q
Input ADC
ADC
RF LF/RF
LF Step Clock
Input Switch Attenuator
Cal
Switch Mixer Final IF
Mixer
ADC Digitized
IF
Final LO
RF Step Image Reject 1st IF Real-Time
Attenuator Filters 1st LO BW ADC
Internal Clock
Alignment
Source RF/uW Section Frequency Conversion/IF Section
Figure 2-1. Real-Time Spectrum Analyzer RF/IF block diagram.
Chapter 2: How Does the Real-Time A real-time digital signal processing (DSP) engine enables
processing with no gaps.
Spectrum Analyzer Work?
Sufficient capture memory and DSP power to enable
This chapter contains several architectural diagrams of the continuous real-time acquisition over the desired time
main acquisition and analysis blocks of the Tektronix Real- measurement period.
Time Spectrum Analyzer (RSA). Some ancillary functions
An integrated signal analysis system that provides multiple
have been omitted to clarify the discussion.
analysis views of the signal under test, all correlated
Modern RSAs can acquire a passband, or span, anywhere in time.
within the input frequency range of the analyzer. At the heart
of this capability is an RF downconverter followed by a wide- RF/IF Signal Conditioning
band intermediate frequency (IF) section. An ADC digitizes Figure 2-1 shows a simplified RSA RF/IF block diagram.
the IF signal and the system carries out all further steps Signals with frequency content anywhere in the frequency
digitally. DSP algorithms perform all signal conditioning range of the RSAs are applied to the input connector. Once
and analysis functions. signals enter the instrument, they are routed and conditioned
Several key characteristics distinguish a successful real-time in accordance with the needs of the analysis selected by
architecture: the user. Variable attenuation and gain is applied. Tuning is
RF signal conditioning that provides a wide-bandwidth IF achieved using multi-stage frequency conversion and a
path and high dynamic range. combination of tunable and fixed local oscillators (LO).
Analog filtering is done at the various IF frequencies. The
The use of band-pass filters, instead of YIG preselection last IF is digitized with an A/D converter. All further processing
filters, enabling simultaneous image-free frequency is performed using DSP techniques. Some RSA models
conversion and wideband measurements across the have optional baseband modes where the input signal is
entire input frequency range of each product. digitized directly, without any frequency conversions. The DSP
An ADC system capable of digitizing the entire real-time for baseband signals follows a similar approach as is used
BW with sufficient fidelity and dynamic range to support with RF signals.
the desired measurements.
Primer
Input Switching and Routing Section Step attenuators used by RSAs vary by model in their
design. They typically can be programmed to attenuate
The input switching and routing section distributes the input
from 0 to greater than 50 dB in steps of 5 or 10 dB.
waveforms to the various signal paths within the instrument.
Some RSA models include a separate DC coupled Image Reject Filter versus YIG Preselection Filters
baseband path for increased dynamic range and accuracy RSAs provide image-free frequency conversion from the RF
when analyzing low frequency signals as well as DC coupled and microwave signals at their input to the final IF. This is
accomplished by placing a variety of filters in front of the
IQ baseband paths. RSAs also include internal alignment
first mixer. The various RSA models use multi-stage mixing
sources. These alignment sources, which produce signals
schemes incorporating broadband filters that allow image-
with properties that are specifically tailored for the RSA
free conversion of the entire acquisition bandwidth with
(PRBS, calibrated sinusoids, modulation references, etc.) repeatable, specified amplitude flatness and phase linearity.
are used in self-alignment procedures that correct for
YIG preselector filters introduce significant distortions when
temperature variations in system parameters such as:
wideband signals are to be measured. The narrowband
Gain preselector must be bypassed when measuring such signals,
Amplitude flatness across the acquisition bandwidth if accurate measurements (particularly phase measurements)
Phase linearity across the acquisition bandwidth must be made.
Time alignment These tunable filters are inherently narrow band. There are
significant phase variations across the filter passband, which
Trigger delay calibration
get worse as the signal approaches the edges of the filter.
The self-alignment processes, when combined with calibrations
Even if an attempt can be made to correct these variations
using external equipment performed at the factory or the
with calibration, the tuning mechanism itself somewhat
service center, are at the heart of all critical measurement
defeats such calibrations. The tuning is done by varying a
specifications of RSAs.
magnetic field imposed on the YIG crystal. When this magnetic
RF and Microwave Sections field is tuned to a different frequency and back to the first
one, the magnetic hysteresis in the magnet structure causes
The RF/Microwave section contains the broadband circuitry that
an inability to return to the precise frequency that was tuned
conditions the input signals so that they have the proper level
before the first tuning change.
and frequency content for optimal downstream processing.
This causes variations in the phase calibration with each
Step Attenuator
tuning change. And if this were not enough, there are small
The step attenuator is a device composed of resistive
variations in amplitude and phase that sweep through the
attenuator pads and RF/µW switches that decreases the
passband as the tuning is swept across large frequency
level of broadband signals by a programmed amount.
bands. These small variations usually change with
The step attenuator performs two functions:
temperature.
1. It reduces the level of RF and microwave signals at the
input to a level that is optimum for processing. The step
attenuator also protects the input from damage due to
very high level signals by absorbing excessive RF power.
2. It presents a broadband impedance match over the entire
frequency range of the instrument. This impedance match
is crucial in maintaining accuracy in measuring RF signals.
For this reason, most instrument specifications are stated
for the condition of 10 dB or more input attenuation.
Primer
Frequency Conversion/IF Section

All RSA models can analyze a broad
band of frequencies centered anywhere
in the analyzer’s frequency range.
This is done by converting the band of
interest to a fixed IF where it is filtered,
amplified and scaled. This IF signal is
then digitized. Real-time and batch
processing are then used to perform
multi-domain analysis on the signals
of interest.
Multi-Stage Frequency Conversion
Figure 2-2. (Left) This is representative of a single signal that can be represented across the microwave frequency
The goal of the frequency conversion
range in a traditional spectrum analyzer when the YIG preselection filter is removed. (Right) Restoring the preselection
filter can then show the single signal. section is to faithfully convert signals in
the desired band of frequencies to an IF
Often, to avoid distortions and bandwidth limitations, spectrum suitable for analog-to-digital conversion. Tuning is accomplished
analyzers that utilize a harmonic mixing architecture for swept by selecting the frequencies of local oscillators (LO) in a
and FFT based analysis techniques need to by-pass the YIG multiple conversion superheterodyne architecture as shown
preselection filter. Figure 2-2 demonstrates the potential in Figure 2-1 (on page 9). Each frequency conversion stage
problem of non-image free frequency conversion that can contains a mixer (analog multiplier) followed by IF filtering and
occur when preselector filters are removed during swept amplification. The choices of IF frequencies, filter shapes,
operation at microwave frequencies. Multiple responses gains, and levels differ depending on RSA model and indeed
across a wide spectrum can appear, masking the true intended are changed within each model as a function of instrument
signal of interest. Restoring the YIG preselection filter will settings in order to optimize performance in several areas as
remove these responses. Traditional spectrum analyzers listed below:
cannot show wideband signals and image-free frequency
Spurious responses due to mixer and filter imperfections
conversion signals without distorting the response.
Dynamic range (smallest and largest signals that can be
More importantly, images frequencies exist that can convert
viewed simultaneously without errors)
spectral energy to the instrument’s intermediate frequency
(IF). This image acts as a spectral window allowing not only Amplitude flatness across the real-time bandwidth
one window to convert, but opening multiple windows due to Phase linearity across the real-time bandwidth
the harmonic mixing utilized the traditional spectrum analyzer. Delay match between the signal and trigger paths
Preamplifier Internal Alignment Sources
Some RSA models include options for a selectable preamplifier
The performance achieved in RSAs for some characteristics
that adds gain to the signal path prior to the image reject filter.
mentioned in the previous bulleted list far exceeds what is
This option improves the noise figure of the RSAs and is
practical with analog components. Filter responses, delays
useful for analyzing very weak signals. Adding gain at the
and gains vary over temperature and can be different for
input, of course, limits the largest signal that can be analyzed.
individual instruments. RSAs performance is achieved by
Switching this amplifier out of the signal path returns the
actually measuring gains, filter shapes and delays and using
analyzer’s range to normal.
DSP to compensate for the measured performance. The
frequency response and gain variations of the wideband
RF components is measured at the factory with calibrated
Primer
Real-Time Digital Signal Processing

Real-Time
Down- Post Acquistion
Amp/Phase converter IQ Out
(Option) Display
DDC/
ADC Correction
Decimation
Capture Analysis
SW
External
Free
Trigger Micro-
Run Processor
Filtered Power
Level
Freq
Live Signal Processing
Mask DPX
Pixel
DPX
Buffer
Figure 2-3. Real-Time Spectrum Analyzer Digital Signal Processing Block Diagram.
equipment, traceable to National Metrology Institutes such Digital Signal Processing Path in Real-Time
as NIST, NPL, PTB. This equipment is also used to calibrate Spectrum Analyzers
the internal alignment sources which in turn provide signals
Tektronix RSAs use a combination of analog and digital
that adjust for the signal path conditions at the time and place
signal processing (DSP) to convert RF signals into
where the RSA is used. RSAs use two kinds of internal signals:
calibrated, time-correlated multi-domain measurements.
A highly accurate, temperature stable sinusoidal signal is This section deals with the digital portion of the RSAs
used to set the signal path gain at a reference frequency,
signal processing flow.
typically 100 MHz. This signal is the internal RF level
reference. It sets the accuracy in measuring RF power Figure 2-3 illustrates the major digital signal processing blocks
at the center of the acquisition bandwidth. used in the Tektronix RSAs. A band of frequencies from the
A calibrated broadband signal is used to measure the RF input is converted to an analog IF signal that is bandpass
amplitude and phase response across the real-time filtered and digitized. Corrections are applied to the sampled
acquisition BW. This signal is the internal channel data correcting for amplitude flatness, phase linearity and
response reference. It provides the information that other imperfections of the signal path. Some corrections are
allows DSP to compensate for the amplitude, phase applied in real-time, others are applied further downstream in
and delay variations across the acquisition bandwidth. the signal processing path.
Digital Signal Processing (DSP) Concepts A digital downconversion and decimation process converts
the A/D samples into streams of in-phase (I) and quadrature
This section contains several architectural diagrams of the
(Q) baseband signals as shown in Figure 2-4 (on the next
main acquisition and analysis blocks typical of Tektronix
page). This IQ representation of the desired signal is the basic
RSAs. Specific implementations vary by model number
form for representing signals in all RSAs. DSP is then used to
and by specific measurement function. Some ancillary
functions have been omitted to clarify the discussion. perform all further signal conditioning and measurements.
Both real-time DSP and batch mode DSP are used in RSAs.
Primer
Decimate I
by N
Analog
IF Corrections IQ Representation of
ADC
(if used) Baseband Time
Decimation
90° Domain Data
Numeric Filters
Oscillator
Decimate
Q
by N
Digital Downconverter
Figure 2-4. IF to IQ conversion in a Real-Time Spectrum Analyzer.
IF Digitizer Internal Alignment

The response across the acquisition bandwidth is affected
Tektronix RSAs typically digitize a band of frequencies
by the combination of mixers, filters and amplifiers that
centered on an intermediate frequency (IF). This band of
comprise the IF processing path. These components can
frequencies is the widest frequency for which real-time
have fine-grain amplitude and phase ripple over the wide
analysis can be performed. Digitizing at a high IF rather than bandwidths acquired by RSAs. An internal alignment
at DC or baseband has several signal processing advantages process measures the amplitude and phase response as a
(spurious performance, DC rejection, dynamic range, etc.). function of offset from the center frequency. The alignment
The sampling rate is chosen so that the desired IF bandwidth is done at the time and place where the instrument is in
fits within a Nyquist zone as shown in Figure 2-5 (on the use and can be triggered either manually or as a function
next page). The sampling rate must be at least twice the IF of temperature. This response is stored in memory.
bandwidth. Sampling without aliasing is possible if the entire Correction Process
if bandwidth fits between zero and one-half the sampling The RSA correction process combines the RF response
frequency, one-half and one, three-halves and twice, etc. measured at the factory with the IF response measured
The practical implementations of IF filters require typical during internal alignments to generate FIR coefficients for
sampling rates at least two-and-a-half times the IF bandwidth. a set of correction filters that compensate for amplitude
flatness and phase response of the entire path between
Corrections the input connector and the ADC. These correction filters
The RSA specifications for amplitude flatness, phase linearity are implemented either in real-time digital hardware or in
and level accuracy far exceed what is practical with the software-based DSP depending on RSAs model and are
components that comprise the analog RF and IF signal applied to the digitized IQ stream.
conditioning portions of the signal path. Tektronix RSAs use
a combination of factory calibration and internal self-alignment Digital Downconverter (DDC)
to compensate for analog component variations (temperature, A common and computationally efficient way to represent
tolerance, aging, etc.) in the signal path. bandpass signals is to use the complex baseband
representation of the waveform.
Factory Calibration
The RF frequency response of the RSA over its range RSAs use the Cartesian complex form, representing the time
of input frequencies is measured at the factory. The RF sampled data as I (in-phase) and Q (quadrature) baseband
behavior at the center of an acquisition bandwidth is components of the signal. This is achieved using a digital
predictable over temperature and does not vary appreciably downconverter (DDC) as shown in Figure 2-4.
as the instrument ages. Once measured in the factory, the
RF response is stored in a correction table that resides in
non-volatile memory.
Primer
Digitized IF Signal IF Signal
Fs Fs 3Fs Fs
4 2 2
I baseband Q baseband
-Fs Fs -Fs Fs
4 4 4 4
Figure 2-5. Passband information is maintained in I and Q even at half the sample rate.
In general, a DDC contains a numeric oscillator that between one half of the sampling frequency and the sampling
generates a sine and a cosine at the center frequency frequency. Sampling produces an image of this signal
of the band of interest. The sine and cosine are numerically between zero and one-half the sampling frequency. The
multiplied with the digitized IF signal; generating streams of signal is then multiplied with coherent sine and cosine
I and Q baseband samples that contain all of the information signals at the center of the passband of interest and followed
present in the original IF. DDCs are used not only to convert by an anti-aliasing filter, generating I and Q baseband signals.
digitized IF signals to baseband but also in fine frequency The baseband signals are real-valued and symmetric about
tuning in RSAs. the origin. The same information is contained in the positive
and negative frequencies. All of the modulation contained in
IQ Baseband Signals the original passband is also contained in these two signals.
Figure 2-5 illustrates the process of taking a frequency band The minimum required sampling frequency for each is now
and converting it to baseband using digital downconversion. half of the original. It is then possible to decimate by two.
The original IF signal in this case is contained in the space
Primer
Decimation longer time records for an equivalent number of samples.

The Nyquist theorem states that for baseband signals The disadvantage of the lower effective sampling rate is a
one need only sample at a rate equal to twice the highest reduced time resolution. The advantages of the lower
frequency of interest. For bandpass signals one needs to effective sampling rate are fewer computations for analysis
sample at a rate at least twice the bandwidth. The sample and less memory usage for a given time record.
rate can be reduced when the needed bandwidth is less than Decimation Filtering
the maximum. Sample rate reduction, or decimation, can be
The Nyquist requirements must also be observed when
used to balance bandwidth, processing time, record length
decimating. If the data rate is reduced by a factor of two,
and memory usage. The Tektronix RSA6000 Series, for
then the bandwidth of the digital signal also must be reduced
example, uses a 100 MS/s sampling rate at the A/D converter
by a factor of two. This must be done with a digital filter
to digitize a 40 MHz acquisition bandwidth, or span. The I
prior to the reduction in sample rate to prevent aliasing.
and Q records that result after DDC, filtering and decimation
Many levels of decimation are used in Tektronix RSAs.
for this 40 MHz acquisition bandwidth are at an effective
Each level contains a digital filter followed by a reduction
sampling rate of half the original, that is, 50 MS/s. The total
in the number of samples. An additional advantage of
number of samples is unchanged: we are left with two sets
decimation and filtering is a reduction in noise with the
of samples, each at an effective rate of 50 MS/s instead of
reduced bandwidth. This reduction in noise is often called
a single set at 100 MS/s. Further decimation is made for
processing gain.
narrower acquisition bandwidths or spans, resulting in
Primer
DFT-Based Spectrum Analysis*

Memory Contents
Input Signal
ADC DFT Engine
Time
Time Samples
Equivalent Bank of Filters

Bank of N Bandpass Complex
filters with centers
separated by one FFT
Envelope
frequency bin width Detection
M/
Input Signal Θ M/
Θ M/
Θ
Time M/
Θ
* The Fast Fourier Transform (FFT) is a common
implementation of a Discrete Fourier Transform (DFT).
Figure 2-6. A DFT-based Spectrum Analyzer and an equivalent implementation using a bank of bandpass filters.
Transforming Time Domain Waveforms to the repetitive Discrete Fourier Transforms (DFTs) are performed
Frequency Domain as shown in Figure 2-6 is such a way that signal processing
keeps up with the input signal. Repetitive Fourier transforms
Spectrum analysis, also called Fourier analysis, separates the
can also be used to discover, capture and analyze infrequent
various frequency components of an input signal. The typical
transient events in the frequency domain even when the
spectrum analyzer display plots the level of the individual
requirements for real-time are not strictly met.
frequency components versus frequency. The difference
between the start and stop frequencies of the plot is the
span. Spectrum analysis is said to be real-time when
Primer
Real-Time Spectrum Analysis Discovering and Capturing Transient Events

For spectrum analysis to be classified real-time, all information Another application of fast and repetitive Fourier transforms
contained within the span of interest must be processed is the discovery, capture and observation of rare events in
indefinitely without gaps. An RSA must take all information the frequency domain. A useful specification is the minimum
contained in time domain waveform and transform it into event duration for 100% probability of capturing a single
frequency domain signals. To do this in real-time requires non-repetitive event. A minimum event is defined as the
several important signal processing requirements: narrowest rectangular pulse that can be captured with 100%
Enough capture bandwidth to support analysis of the certainty at the specified accuracy. Narrower events can be
signal of interest detected, but the accuracy and probability may degrade.
Discovering, capturing and analyzing transients requires:
A high enough ADC clock rate to exceed the Nyquist
criteria for the capture bandwidth Enough capture bandwidth to support analysis of the
signal of interest
A long enough analysis interval to support the narrowest
resolution bandwidth (RBW) of interest A high enough ADC clock rate to exceed the Nyquist
criteria for the capture bandwidth
A fast enough DFT transform rate to exceed the Nyquist
criteria for the RBW of interest A long enough analysis interval to support the narrowest
resolution bandwidth (RBW) of interest
DFT rates exceeding the Nyquist criteria for RBW require
overlapping DFT frames: A fast enough DFT transform rate to support the minimum
event duration
The amount of overlap depends on the window function
At over 292,000 spectrums per second, the DPX Spectrum
The window function is determined by the RBW
mode in the RSA6000 Series can detect RF pulses as short
Today’s RSAs meet the real-time requirements listed above as 10.3 microseconds with the full accuracy specifications
for Frequency Mask Trigger (FMT) for spans up to their with 100% probability. A Swept Spectrum Analyzer (SA)
maximum real-time acquisition bandwidth. Triggering on with 50 sweeps per second requires pulses longer than
frequency domain events, therefore, considers all the 20 milliseconds for 100% probability of detection with
information contained in the selected acquisition bandwidth. full accuracy.
Primer
is represented as a color. The real-time DFT effectively

Spectrum
samples the entire spectrum of the incoming signal at the
rate with which new spectrums are computed. Events
Color Scale
occurring between the time segments on which the FFTs

are performed are lost. RSAs minimize or eliminate the
“dead time” by performing hardware-based DFTs, often
Frequency performing transforms on overlapping time segments at
the fastest sample rate.
An SA, in contrast, is tuned to a single frequency at any
given time. The frequency changes as the sweep advances
tracing the diagonal line in Figure 2-7. The slope of the line
Time
Sw
becomes steeper as the sweep slows so that the function of
ee
p a spectrum analyzer in zero-span can be represented as a
vertical line indicating that the instrument is tuned to a single
frequency as time advances. Figure 2-7 also shows how a
sweep can miss transient events such as the single frequency
Spectrogram hop depicted.
Figure 2-7. Spectrum, Spectrogram and Sweep.
RBW on the Real-Time Spectrum Analyzer
Relating RSA to Swept Spectrum Analyzer Frequency resolution is an important spectrum analyzer
specification. When we try to measure signals that are close
Consider a RSA system as described on the previous page. A
in frequency, frequency resolution determines the capability
passband of interest is downconverted to an IF and digitized.
of the spectrum analyzer to distinguish between them. On
The time domain samples are digitally converted to a base-
traditional SAs, the IF filter bandwidth determines the ability
band record composed of a sequence of I (in-phase) and Q
to resolve adjacent signals and is also called the resolution
(quadrature) samples. DFTs are sequentially performed on
bandwidth (RBW). For example, in order to resolve two
segments of the IQ record generating a mathematical
signals of equal amplitude and 100 kHz apart in frequency,
representation of frequency occupancy over time, as shown
RBW needs to be less than 100 kHz.
in Figure 2-6 (on page 16).
For spectrum analyzers based on the DFT technique, the
Taking sequential equally spaced DFTs over time is
RBW is inversely proportional to the acquisition time. Given
mathematically equivalent to passing the input signal
the same sampling frequency, more samples are required to
through a bank of bandpass filters and then sampling
achieve a smaller RBW. In addition, windowing also affects
the magnitude and phase at the output of each filter. The
the RBW.
frequency domain behavior over time can be visualized as
a spectrogram as shown in Figure 2-7, where frequency is
plotted horizontally, time is plotted vertically and the amplitude
Primer
Frame 1 Frame 2 Frame 3 Frame 2 Frame 2 Frame 2
Figure 2-8. Three frames of a sampled time domain signal. Figure 2-9. Discontinuities caused by periodic extension of samples in a single frame.
Windowing The choice of window function depends on its frequency

response characteristics such as side-lobe level, equivalent
There is an assumption inherent in the mathematics of
noise bandwidth, and amplitude error. The window shape
Discrete Fourier Transform (DFT) analysis that the data to be
also determines the effective RBW filter.
processed is a single period of a periodically repeating signal.
Figure 2-8 depicts a series of time domain samples. When Like other spectrum analyzers, the RSAs allow the user to
DFT processing is applied to Frame 2 in Figure 2-8, for select the RBW filter. The RSAs also allow the user to select
example, the periodic extension is made to the signal. The among many common window types. The added flexibility
discontinuities between successive frames will generally to directly specify the window shape enables the users to
occur as shown in Figure 2-9. optimize for specific measurements. Special attention, for
example, should be paid to the spectrum analysis of pulse
These artificial discontinuities generate spectral artifacts
signals. If the pulse duration is shorter than the window
not present in the original signal. This effect produces an
length, uniform window (no windowing) should be used to
inaccurate representation of the signal and is called spectral
avoid de-emphasizing effects on both ends of the DFT frame.
leakage. Spectral leakage not only creates signals in the
For further information on the importance of overlap processing
output that were not present in the input, but also reduces
as it relates to acquisition and signal processing, windowing
the ability to observe small signals in the presence of nearby
effects, and the requirements for defining real-time processing,
large ones.
please refer to the Tektronix Primer, “Understanding FFT
Tektronix Real-Time Spectrum Analyzers apply a windowing Overlap Processing on the Real-Time Spectrum Analyzer”.
technique to reduce the effects of spectral leakage. Before
performing the DFT, the DFT frame is multiplied by a window
function with the same length sample by sample. The window
functions usually have a bell shape, reducing or eliminating
the discontinuities at the ends of the DFT frame.
Primer
1
Vertical Linear Scale
0.9
0
0.8 -20
0.7 -40
0.6 -60
-80
0.5
-100
0.4
-120
0.3
-140
0.2 -160
0.1 -180
0 -200
50 100 150 200 250 -8 -6 -4 -2 0 2 4 6 8
Figure 2-10. Kaiser Window (beta 16.7) in Time Domain (Horizontal is time sample, Figure 2-11. The spectrum of a Kaiser window (beta 16.7). The horizontal scale unit is
Vertical is linear scale). the frequency bin (Fs/N). The vertical scale is in dB.
The magnitude of the frequency response of the window Discrete Fourier Transforms (DFT) in the Real-Time
function determines the RBW shape. For example, the RBW Spectrum Analyzer
on RSA6000 is defined as the 3 dB bandwidth and is related to
The DFT is defined below:
the sampling frequency and samples in the DFT as follows:
k * Fs Equation 1
RBW =
N
or
k * Fs Equation 2 This is the basis of a real-time spectrum analyzer and it is
N= used to estimate the individual frequency components x(k)
RBW
Where k is a window-related coefficient, N is the number of from the input sequence x(n). The DFT is block-based and N
time-domain samples used in the DFT calculation, and Fs is the total sample number of each DFT block (or Frame). The
is the sampling frequency. For the Kaiser window with beta1 input sequence x(n) is a sampled version of the input signal
16.7, k is about 2.23. The RBW shape factor, defined as x(t). Although the input sequence is only defined for integer
the frequency ratio between the spectrum amplitude at values of n, the output is a continuous function of k, where
60 dB and 3 dB, is about 4:1. On the RSA6000, the k=(NW)/(2P) and W is the radian frequency. The magnitude of
spectrum analysis measurement uses Equation 2 to X[k] represents the magnitude of the frequency component at
calculate the required number of samples for the DFT frequency W that is present in the input sequence x(n).
based on the input span and RBW settings. There are various efficient methods to compute the DFT.
The time domain and the spectrum of the Kaiser window Examples include the Fast Fourier Transform (FFT) and the
used for RSA6000 spectrum analysis is shown in Figure Chirp-Z Transform (CZT). The choice of implementation
2-10 and Figure 2-11. This is the default window used in the method depends on the particular needs of the application.
RSA6000 for spectrum analysis. Other windows (such as The CZT, for example, has greater flexibility in choosing the
Blackman-Harris, Uniform, Hanning) may be user-selected frequency range and the number of output points than the
to meet special measurement requirements, and may be FFT. The FFT is less flexible but requires fewer computations.
used by the instrument when performing some of the Both the CZT and the FFT are used in RSAs.
measurements available in the instrument.
[Reference 1] Oppenheim, A.V. and R.W Schafer, Discrete-time Signal Processing, Prentice-Hall, 1989, p. 453.
Primer
Input Signal: Frequency = 4.7222 Hz : Sample Frequency = 20 Hz DFT (X[k])

1 8
0.8 7
0.6
6
0.4
5
0.2
Magnitude
Amplitude
0 4
-0.2 3
-0.4
2
-0.6
1
x(t) -0.8
x(n) DFT (X[k])
-1 0
0 4 8 12 0 5 10 15 20
Sample Number (n) Frequency (Hz)
Figure 2-12. Input Signal. Figure 2-13. DFT of x(n) evaluated continuously.
FFT CZT: Start Frequency = 2.5 Hz : Stop Frequency = 7.4 Hz

8 8
7 7
6 6
5 5
Magnitude
Magnitude
4 4
3 3
2 2
1 1
DFT (X[k]) DFT (X[k])
FFT 0 CZT 0
0 5 10 15 20 0 5 10 15 20
Frequency (Hz) Frequency (Hz)
Figure 2-14. FFT of x(n), length of FFT = N = length of x(n). Figure 2-15. CZT of x(n).
The ability to resolve frequency components is not dependent A FFT returns N-equally spaced frequency domain samples
on the particular implementation of the DFT and is determined of X[k]. The magnitude of X[k] is shown in Figure 2-14. Note
by the time length of the input sequence or the RBW. that the samples returned by the FFT might miss the peaks
of magnitude of X[k].
To illustrate the relationship of the DFT to the FFT and the
CZT, a sampled Continuous Waveform (CW) signal will be A CZT can return M frequency domain samples with an
analyzed. For illustration purposes a real-valued sine wave arbitrary start and stop frequency (Figure 2-15). Notice that
x(t) will be used as the input signal (Figure 2-12). The sample the CZT does not change the underlying frequency domain
version of x(t) is x(n). In this case N = 16 and the sample output of the DFT. It only takes a different set of frequency
rate is 20 Hz. domain samples than the FFT.
Figure 2-13 shows the result of evaluating the DFT for An advantage of using the CZT is that the frequency of
0 ≤ k < N. Note that the magnitude of X[k] for W > P the first and last sample in the frequency domain can be
(f > 10 Hz) is a mirror image of the first half. This is the result arbitrarily selected and does not depend on the input sample
for a real-valued input sequence x(n). In practice, the results rate. The same result can also be achieved by arbitrarily
from P < W < 2 P are discarded (or not computed) when a controlling the input sample rate so that the output of the
real input signal is analyzed. For a complex input, a unique FFT produces the same output samples as the CZT. The
result can be obtained for 0 ≤ W < 2 P (0 ≤ f < 20 Hz). end result is the same in both cases. The choice is purely
an implementation issue, and depending on the requirements
and available HW, one or the other will be a more
optimal solution.
Primer
Frequency Response
Digital Filtering
1.2 Finite Impulse Response (FIR)
1
Filters
Frequency filters are used in many
0.8
applications to select some frequencies
Amplitude
0.6 and reject others. While traditional

filters are implemented using analog
0.4 circuit elements (RLC), DSP selects
the frequencies to be enhanced or
0.2
attenuated mathematically. A common
0 mathematical implementation is the
0 10 20 30 40 50 60 70
Frequency
Finite Impulse Response (FIR) filter.
RSAs make extensive use of FIR
Figure 2-16. Frequency response of a lowpass filter. filters. In addition to the usual signal
conditioning applications requiring the
Impulse Response passage or rejection of specific bands,
FIR filters are also used to adjust for
1.2
analog signal path imperfections.
1 Internally generated alignment data
0.8
is combined with stored factory
calibration data to create FIR filters
0.6 with a response that compensates
for the analog signal path frequency
Value
0.4
response, making the cascade of the
0.2
analog and digital paths have flat
0
amplitude response and linear phase.
-100 -80 -60 -40 -20 0 20 40 60 80 100
-0.2 Frequency Response vs.

-0.4
Impulse Response
Time The theory of Fourier transforms shows
an equivalency between the frequency
Figure 2-17. Impulse response for the lowpass filter in Figure 2-15.
domain and the time domain. It further tells us that the transfer
function of a device, usually expressed as its amplitude and
phase response over frequency, is equivalent to the impulse
response over time. A FIR filter emulates the impulse response
of the desired filter transfer function with a discrete-time
approximation that has finite timeduration. Signal filtering is
then performed by convolving the input signal with the
impulse response of the filter.
Figure 2-16 shows the magnitude of the transfer function of
a lowpass filter. Figure 2-17 shows its impulse response.
Primer
Frequency Frequency
Domain Input Domain Output Input Time
Samples
Output Time
Samples
Multiplication
Frequency Response
1.2
0.8
Impulse Response
Convolution
Amplitude
1.2
0.6
1
0.4
0.8
0.2
0.6
0
0 10 20 30 40 50 60 70
Value
Frequency 0.4
0.2
Filter
-100 -80 -60 -40 -20 0 20 40 60 80 100
-0.2
-0.4
Time
Frequency Sampled Filter

Response
Frequency Domain Impulse Response Discrete Time
Figure 2-18. Multiplying a filter by its frequency response. Figure 2-19. Convolution in the time domain is equivalent to multiplication in the
frequency domain.
Numerical Convolution
Fourier theory states that a multiplication in the frequency
The frequency domain is often used to analyze the responses domain is the equivalent of a convolution in the time
of linear systems such as filters. Signals are expressed in domain. The frequency domain multiplication shown above
terms of their frequency content. The spectrum of the signal is equivalent to convolving the time domain representation
at the output of a filter is computed by multiplying the input of the input signal with the impulse response of the filter as
signal spectrum by the frequency response of the filter. Figure shown in Figure 2-19.
2-18 illustrates this frequency domain operation.
Primer
Input time
samples Time samples shifted each clock cycle
Output time
samples
M-1 M-2 M-3 M-4 M-5 1 0
Impulse Response
1.2
0.8
0.6
Value
Sampled filter
0.4
0.2
0
-100 -80 -60 -40 -20 0 20 40 60 80 100
impulse response -0.2
-0.4
Time
stored in registers Numerical Convolution

Figure 2-20. Discrete Time Numerical Convolution.
All frequency filters involve the use of memory elements. The lower registers are used to store values of the filter’s
Capacitors and inductors, the common reactive elements impulse response with the earlier samples on the right and
used in analog filters, have memory since their output the later samples on the left. The upper registers are used
in a circuit depends on the current input as well as the to shift the input signal from left to right with one shift each
input at previous points in time. A discrete time filter can clock cycle. The contents of each corresponding register
be constructed using actual memory elements as shown are multiplied together and all of the resulting products are
in Figure 2-20. summed each clock cycle. The result of the sum is the
filtered signal.
Primer
• Offset:
0.00 dBm
• dB/div:
10.00 dB
• RBW
91KHz
-100.0 dBm
• CF: 2.445 GHz • Span: 10.00 MHz
a) Swept Spectrum Analyzer b) Tektronix Real-Time Spectrum Analyzer
Figure 2-21 a, b. Comparison (a), Swept Spectrum Analyzer MaxHold trace after 120 seconds and (b), Tektronix Real-Time Spectrum Analyzer with DPX bitmap MaxHold trace
after 20 seconds.
In summary, the RSA relies heavily on digital signal processing In DPX mode, CZTs are used for resolution
for spectrum analysis. Key points of DSP as applied to the bandwidth flexibility.
RSAs are: In Real-time mode, windowed FFTs are used, specified in
The RSA6000 uses a combination of FFTs and CZTs to the classical method for FFT analysis, that of the Noise
achieve spectrum displays. Bandwidth. The Noise Bandwidth is approximately 6%
(0.25 dB) greater than the RBW.
FFTs are more computationally efficient, allowing faster
transform rates, but CZTs are more flexible, allowing variable As we have seen in this section, digitally implemented
resolution bandwidths for a fixed set of input samples. corrections and filtering are a key factor in implanting the
The resolution bandwidth (RBW) shape is achieved by high transform rate required of a RSA. The next section
applying an optimized window function to the time domain looks at the practical use of these filters in one of the
signals before performing a Fourier transform. RBWs are unique displays available in the RSA, the Digital Phosphor
specified by their 3 dB bandwidth and 60 dB:3 dB shape Spectrum Display.
factor, in the same fashion as an analog implementation. In
DPX Technology: a Revolutionary Tool for
general, the shape factor of the digitally implemented filter
is lower (sharper) than an analog implementation, yielding
Signal Discovery
easier resolution of closely spaced signals of widely Tektronix’ patented Digital Phosphor technology or
different amplitudes. DPX reveals signal details that are completely missed by
conventional spectrum analyzers and VSAs (Figure 2-21).
Other shape factors can be used for special applications by
The DPX Spectrum’s Live RF display shows signals never
applying optimized window functions.
seen before, giving users instant insight and greatly accelerating
The RSA3000 Series RSA uses a combination of methods problem discovery and diagnosis. DPX is a standard feature in
when performing spectrum analysis. all Tektronix RSAs.
In Spectrum mode, the result of a windowed FFT is
convolved with a RBW shape to yield Spectrum traces
with a specified RBW, similar to an analog spectrum
analyzer. This process yields a slightly wider shape factor
of approximately 5:1, compared to 4.1:1 for the RSA6000.
Primer
Digital Phosphor Display Proportionality: The slower the electron beam passes
through a point on the phosphor-coated screen, the brighter
The name “Digital Phosphor” derives from the phosphor
the resulting light. Brightness of a spot also increases as the
coating on the inside of cathode ray tubes (CRTs) used as
beam hits it more frequently. Users intuitively know how to
displays in older televisions, computer monitors and test
interpret this z-axis information: a bright section of the trace
equipment where the electron beam is directly controlled
indicates a frequent event or slow beam motion, and a dim
by the input waveform. When the phosphor is excited by
trace results from infrequent events or fast-moving beams.
an electron beam, it fluoresces, lighting up the path drawn
In the DPX display, both color and brightness provide
by the stream of electrons.
z-axis emphasis.
Liquid Crystal Displays (LCDs) replaced CRTs in many
Persistence and proportionality do not come naturally to
applications due to their smaller depth and lower power
instruments with LCDs and a digital signal path. Tektronix
requirements, among other advantages. However, the
developed Digital Phosphor technology so the analog
combination of phosphor coatings and vector drawing in
benefits of a variable persistence CRT could be achieved,
CRTs provided several valuable benefits.
and even improved upon, in our industry-leading digital
Persistence: Phosphor continues to glow even after the oscilloscopes and now in our RSAs. Digital enhancements
electron beam has passed by. Generally, the fluorescence such as intensity grading, selectable color schemes and
fades quickly enough that viewers don’t perceive it lingering, statistical traces communicate more information in less time.
but even a small amount of persistence will allow the human
eye to detect events that would otherwise be too short
to see.
Primer
1
1
1 2
Amplitude
Amplitude
1 4
2 1
1 1 1 5 7
1 1 7 8
1 1 1 1 1 1 9 9 9 1 1 1 1 1 9 9 9
Frequency Frequency
Figure 2-22. Example 3-D Bitmap Database after 1 (left) and 9 (right) updates. Note that each column contains the same total number of “hits.”
The DPX Display Engine Number of Occurrences Color

Performing thousands of spectral measurements per second 0 black
and updating the screen at a live rate is an oversimplified 1 blue
description of the role DPX technology performs in an RSA. 2 light blue
Thousands of acquisitions are taken and transformed into 3 cyan
4 green blue
spectrums every second. This high transform rate is the key
5 green
to detecting infrequent events, but it is far too fast for the
6 yellow
LCD to keep up with, and it is well beyond what human eyes 7 orange
can perceive. So the incoming spectrums are written into a 8 red orange
bitmap database at full speed then transferred to the screen 9 red
at a viewable rate. Picture the bitmap database as a dense
Figure 2-23. Example Color-Mapping Algorithm.
grid created by dividing a spectrum graph into rows
representing trace amplitude values and columns for
points on the frequency axis. Each cell in this grid contains The grid on the right shows values that our simplified
the count of how many times it was hit by an incoming database might contain after an additional eight spectral
spectrum. Tracking these counts is how Digital Phosphor transforms have been performed and their results stored
implements proportionality, so you can visually distinguish in the cells. One of the nine spectrums happened to be
rare transients from normal signals and background noise. computed at a time during which the signal was absent,
as you can see by the string of “1” values at the noise floor.
The actual 3-D database in an RSA contains hundreds
of columns and rows, but we will use an 11X10 matrix to When we map the Number of Occurrences values to a color
illustrate the concept. The picture on the left in Figure 2-22 scale, data turns into information. The table found in Figure
shows what the database cells might contain after a single 2-23 shows the color-mapping algorithm that will be used
spectrum is mapped into it. Blank cells contain the value for this example. Warmer colors (red, orange, yellow) indicate
zero, meaning that no points from a spectrum have fallen more occurrences. Other intensity-grading schemes can also
into them yet. be used.
Primer
Figure 2-24. Color-coded low-resolution example with Temperature Bitmap (left), and a real DPX display (right) shown with Spectrum Bitmap.
In Figure 2-24, the left image is the result of coloring the “variable persistence.” Adjusting the fraction changes the
database cells according to how many times they were length of time it takes for a signal event to decay from the
written into by the nine spectrums. Displaying these colored database, and thus fade from the display.
cells, one per pixel on the screen, creates the spectacular
Imagine a signal that popped up only once during the time
DPX displays, as seen in the right image.
DPX was running. Further, assume that it was present for
Persistence all 1465 of the spectrum updates in a frame and that the
In the RSA6000 Series, for example, over 292,000 spectrums Variable Persistence Factor causes 25% attenuation after
enter the database each second. At the end of each frame each frame. The cells it affected would start out with a value
of 14,600 input spectrums (about 20 times per second), the of 1465 and be displayed at full force. One frame later, the
bitmap database is transferred out for additional processing Number of Occurrences values become 1099. After the next
before being displayed, and data from a new frame starts frame, they are 824, then smaller and smaller until they are so
filling the bitmap. dim as to be invisible. On the screen, you would initially see a
bright trace with a spike at the signal frequency. The part of
To implement persistence, the DPX engine can keep the
the trace where the signal occurred fades away. During this
existing counts and add to them as new spectrums arrive,
time, the pixels start to brighten at the noise level below the
rather than clearing the bitmap database counts to zero at
fading signal. In the end, there is only a baseline trace in the
the start of each new frame. Maintaining the full count values
display (Figure 2-25, on the next page).
across frames is “infinite persistence.” If only a fraction of
each count is carried over to the next frame, it is called
Primer
Figure 2-25. With variable persistence, a brief CW signal captured by DPX remains in the display for an adjustable period of time before fading away.
Amplitude
Amplitude
Amplitude
Frequency Frequency Frequency
+Peak Detection -Peak Detection +Average Detection
Figure 2-26. Detected traces example: +Peak Detection (left); - Peak Detection (middle); and Average Detection (right).
Persistence is an extremely valuable troubleshooting aid, DPX +Peak trace is almost exactly the same as the
delivering all the benefits of MaxHold and more. To find MaxHold trace on a typical spectrum analyzer, with the
out if there is an intermittent signal or occasional shift in important difference that the DPX trace’s update rate is
frequency or amplitude, you can turn on Infinite Persistence orders of magnitude faster.
and let the RSA baby-sit. When you return, you will see not
only the highest level for each frequency point, but also the The DPX Transform Engine
lowest levels and any points in between. Once the presence So how do all those spectrums get generated? In parallel with
of transient behavior or intruding signals has been revealed, the software batch processing used for most measurements,
you can characterize the problem in detail with Variable and using the same stream of incoming IQ data, there is a
Persistence. hardware-based computation engine devoted to continuous,
Statistical Line Traces real-time signal processing. This subsystem supports time-
A colorful bitmap is DPX Spectrum’s signature trace, but DPX critical functions like power-level triggering, frequency mask
also produces statistical line traces. The database contents triggering and others. It also performs DFTs fast enough to
are queried for the highest, lowest and average amplitude produce the spectrum rate used by the DPX display system.
values recorded in each frequency column. The three resulting
DPX Density Measurements
trace detections are +Peak, -Peak and Average (Figure 2-26).
“Density” is a measure of the amount of time during a defined
The +Peak and -Peak traces show signal maxima and minima measurement period during which signals are present within
instantly and clearly. Average detection finds the mean level a particular area of the DPX Spectrum bitmap. A clean CW
for the signal at each frequency point. All these traces can tone gives a 100% reading, while a pulse that is on for one
be saved and restored for use as reference traces. microsecond out of every millisecond reads 0.1%.
Just like regular spectrum traces, DPX line traces can be
accumulated over ongoing acquisitions to yield MaxHold,
MinHold and Average trace functions. Using Hold on the
Primer
Figure 2-27. Example of Density Trigger Function. Left: A free-run DPX spectrum display showing pulses with varying frequency. Occasionally, a short pulse in the middle
appears for a splut instant, but it is hard to capture it with just a Run/Stop button. Right: The triggered DPX displays shows the low-amplitude pulse that was not apparent
in the untriggered display. The analyzer was set to trigger whenever the average density in the user-drawn box measured 50% or higher.
that the area of interest in the waveform

be selected for analysis. This section
illustrates the triggering, acquisition and
analysis period controls of the RSA,
seen in Figure 2-28 (RSA6000 Series).
The timing controls, when used in
conjunction with triggers, offer a powerful
combination for analyzing transient or
other timing related parameters.The
acquisition length specifies the length
of time for which samples will be stored
in memory in response to a trigger. The
acquisition history determines how
many previous acquisitions will be kept
after each new trigger. Tektronix RSAs
show the entire acquisition length in the
Figure 2-28. Time overview with spectrum length and multi-domain analysis length indicators as shown in time domain overview window.
the RSA6000.
The spectrum length determines
the length of time for which spectrum
Timing and Triggers displays are calculated. The spectrum offset determines the
delay or advance from the instant of the trigger event until the
Real-time processing enables the DPX display that makes
beginning of the FFT frame that is displayed. Both spectrum
the RSA a powerful discovery tool. DPX Density trigger can
length and spectrum offset have a time resolution of one FFT
discriminate signals within a precise amplitude-frequency
frame. While the RSA3000 Series uses a fixed FFT length, the
range without the operator having to know any characteristics
added flexibility of the high performance RSA6000 Series
of the target signals. More information on the advanced
analyzers allow one to vary the FFT length for spectrum
function of the Tektronix patented DPX technologies can be
viewing. Tektronix RSAs indicate the spectrum offset and
found in the technology primer: "DPX Acquisition Technology
spectrum length using a colored bar at the bottom of the
for Spectrum Analyzers Fundamentals Primer".
time domain overview window. The bar color is keyed to the
The DPX display does not keep a time domain record that pertinent display.
can re-analyzed in multiple domains. Capture and additional
analysis requires that the signal be written into memory, and
Primer
The analysis length determines the length of time for which

modulation analysis and other time-based measurements are
made. The analysis offset determines the delay or advance
from the instant of the trigger until the beginning of the analysis. Input
Tektronix RSAs indicate the analysis offset and length using Signal
a colored bar at the bottom of the time domain overview
window. The bar color is keyed to the pertinent display.
The output trigger indicator allows the user to selectively
enable a TTL rear-panel output at the instant of a trigger. Trigger
This can be used to synchronize RSA measurements with
other instruments such as oscilloscopes or logic analyzers.
Real-Time Triggering and Acquisition

The RSA is capable of performing time, spectrum, and
modulation analysis. Triggering is critical to capturing time Oscilloscope
domain information. The RSA offers unique trigger functionality, Display
providing power and FMTs as well as the usual power, external
and level-based triggers.
The most common trigger system is the one used in most
oscilloscopes. In traditional analog oscilloscopes, the signal
to be observed is fed to one input while the trigger is fed to
another. The trigger event causes the start of a horizontal
Figure 2-29. Traditional oscilloscope triggering.
sweep while the amplitude of the signal is shown as a vertical
displacement superimposed on a calibrated graticule. In its
simplest form, analog triggering allows events that happen
after the trigger to be observed, as shown in Figure 2-29.
Primer
First In/First Out (FIFO) Memory

Most recent sample Older samples
ADC 0 1 2 3 4 5 6 7 8 9 10 11 N N N
N
-3 -2 -1
Memory Control
Fs Freeze
TD = 0 Pre-trigger samples only
TD = Memory length Post-trigger samples only
Trigger Variable delay
= 50% Memory length Half pre-trigger, half post-trigger
0 to N clocks
Figure 2-30. Triggering in digital acquisition systems.
Triggering in Systems with Digital Acquisition those that preceded the trigger and half the stored samples
followed it. This concept is similar to a trigger delay used in
The ability to represent and process signals digitally, coupled
zero span mode of a conventional SA. The RSA can
with large memory capacity, and allows the capture of events
capture much longer time records, however, and this
that happen before the trigger as well as after it.
signal data can subsequently be analyzed in the frequency,
Digital acquisition systems of the type used in Tektronix time, and modulation domains. This is a powerful tool for
RSAs use an Analog-to-Digital Converter (ADC) to fill a applications such as signal monitoring and device
deep memory with time samples of the received signal. troubleshooting.
Conceptually, new samples are continuously fed to the
memory while the oldest samples fall off. The example Trigger Modes and Features
shown in Figure 2-30 shows a memory configured to store The free-run mode acquires samples of the received IF
“N” samples. The arrival of a trigger stops the acquisition, signal without the consideration of any trigger conditions.
freezing the contents of the memory. The addition of a Spectrum, modulation or other measurements are displayed
variable delay in the path of the trigger signal allows events as they are acquired and processed.
that happen before a trigger as well as those that come
The triggered mode requires a trigger source as well as the
after it to be captured.
setting of various parameters that define the conditions for
Consider a case in which there is no delay. The trigger event triggering as well as the instrument behavior in response to
causes the memory to freeze immediately after a sample a trigger.
concurrent with the trigger is stored. The memory then
A selection of continuous or single trigger determines
contains the sample at the time of the trigger as well as
whether acquisitions repeat each time a trigger occurs or
“N” samples that occurred before the trigger. Only pre-trigger
are taken only once each time a measurement is armed. The
events are stored.
trigger position, adjustable from 0 to 100%, selects which
Consider now the case in which the delay is set to match portion of an acquisition block is pre-trigger. A selection of
exactly the length of the memory. “N” samples are then 10% captures pre-trigger data for one tenth of the selected
allowed to come into the memory after the trigger occurrence block and post-trigger data for nine tenths. Trigger slope
before the memory is frozen. The memory then contains “N” allows the selection of rising edges, falling edges or their
samples of signal activity after the trigger. Only post-trigger combination for triggering. Rise and fall allows the capture
events are stored. of complete bursts. Fall and rise allows the capture of gaps
Both post- and pre-trigger events can be captured if the delay in an otherwise continuous signal.
is set to a fraction of the memory length. If the delay is set
to half of the memory depth, half of the stored samples are
Primer
Trigger Source Trigger Signal Setting Units Time Uncertainty Notes

External External trigger Volts (variable) RSA3300: ± 40 ns RSA3000: 1 ext. trig Input
connector or TTL RSA3400: ± 20 ns RSA6000: 2 ext. trig Inputs
RSA6000: ± 12 ns (front and rear)
Power Level Power calculated Power, or dB relative ±1 time domain RSA6000 adds
Trigger within current to full scale points (based on user-settable filter
acquisition BW effective sampling rate) bandwidth to trigger
Frequency Mask Point-by-point dB and Hz, based ±1 frame length Frame length
Trigger comparison at the output on the graphical mask (based on effective
of a FFT processor drawn on screen sampling rate)
DPX Density User-defined amplitude- dB and Hz based Approximately RSA6000 Series with
Trigger frequency area and rectangular area drawn 50ms Option 200
signal density on screen and Density
Measurement percentage
Time-Qualified Power calculated and Power, dB above and ±1 time domain RSA6000 Series with
Triggers coupled with logic below, time offset points (based on Option 200
conditions and offset timing effective sampling rate)
Table 2-1. Comparison of Real-Time Spectrum Analyzers trigger sources.
Real-Time Spectrum Analyzer Trigger Sources spans that require further filtering and decimation. Level
triggering uses the full digitization rate and can detect
Tektronix RSAs provide several methods of internal and
events with durations as brief as one sample at the full
external triggering. Table 2-1 summarizes the various real-
sampling rate. The time resolution of the downstream
time trigger sources, their settings, and the time resolution
analysis, however, is limited to the decimated effective
that is associated with each one.
sampling rate. Level trigger is available on some
External triggering allows an external signal to control the RSA models.
acquisition. This is typically a control signal such as a
Power triggering calculates the power of the signal after
frequency switching command from the system under test.
filtering and decimation. The power of each filtered pair
This external signal prompts the acquisition of an event in
of IQ samples (I2+Q2) is compared with a user-selected
the system under test.
power setting. Some RSA models offer a filtered power
Internal triggering depends on the characteristics of the trigger where a filter with a selectable bandwidth to be
signal being tested. The various RSA models have the ability applied to the signal in the trigger path.
to trigger on the level of the digitized signal, on the power of
DPX Density triggering uses the same screen-based
the signal after filtering and decimation, or on the occurrence
measurement box as the DPX Density measurement.
of specific spectral components using the FMT. Each of the
The trigger system monitors the density measurement
trigger sources and modes offers specific advantages in
and actuates a trigger whenever the density value exceeds
terms of frequency selectivity, time resolution and dynamic
the adjustable density threshold.
range. The functional elements that support these features
are shown in Figure 2-31 (on the next page). Time-Qualified triggering can be combined with all other
triggering methods and utilizes calculated power, logic
Level triggering compares the digitized signal at the output
conditions, and offset timing to these triggers with a fast
of the ADC with a user-selected setting. The full bandwidth
sample rate timing uncertainty.
of the digitized signal is used, even when observing narrow
Primer
External Trigger Processing Block

Trigger
Level
Fs
ADC
Power Trigger
F s effective Timing and
Power
Control
I 2 + Q2
Fs Density/Mask
DPX Density
and Frequency
Mask HW
FFT engine
DDC/
Decimator Memory
Freeze Memory Control
Figure 2-31. Real-Time Spectrum Analyzer trigger processing.
violations and much more. A full DFT is required to compare

a signal to a mask, requiring a complete frame. The time
resolution for FMT is roughly one DFT frame. Trigger events
are determined in the frequency domain using a dedicated
hardware DFT processor, as shown in the block diagram in
Figure 2-31.
Constructing a Frequency Mask

Like other forms of mask testing, the FMT (also known
as frequency domain trigger) starts with a definition of
an on-screen mask. This definition is done with a set of
frequency points and their amplitudes. The mask can be
defined point-by-point or graphically by drawing it with a
mouse or other pointing device. Triggers can be set to occur
when a signal outside the mask boundary “breaks in,” or
Figure 2-32. Frequency mask definition. when a signal inside the mask boundary “breaks out.”
Figure 2-32 shows a frequency mask defined to allow
Frequency mask triggering compares the spectrum shape the passage of the normal spectrum of a signal but not
to a user-defined mask. This powerful technique allows momentary aberrations. On the next page, Figure 2-33 shows
changes in a spectrum shape to trigger an acquisition. FMTs a spectrogram display for an acquisition that was triggered
can reliably detect signals far below full-scale even in the when the signal momentarily exceeded the mask. Figure 2-34
presence of other signals at much higher levels. This ability shows the spectrum for the first frame where the mask was
to trigger on weak signals in the presence of strong ones exceeded. Note that pre-trigger and post-trigger data were
is critical for detecting intermittent signals, the presence of collected and are both shown in the spectrogram.
inter-modulation products, transient spectrum containment
Primer
Figure 2-33. Spectrogram showing a transient signal adjacent to the carrier. The Figure 2-34. One frame of the spectrogram showing the trigger event where the
cursor is set to the trigger point, so pre-trigger data is displayed above the cursor line, transient signal breaks the boundary of the frequency mask.
and post-trigger data is displayed below the cursor line. The narrow white line at the
left of the blue area denotes post-trigger data.
Demodulation is therefore the time derivative of Phase Modulation (PM).

Quadrature Phase Shift Keying (QPSK) is a digital modulation
Modulation is the means through which RF signals carry
format in which the symbol decision points occur at multiples
information. Modulation analysis using the Tektronix RSA
of 90 degrees of phase. Quadrature Amplitude Modulation
not only extracts the data being transmitted but also
(QAM) is a high-order modulation format in which both
measures the accuracy with which signals are modulated.
amplitude and phase are varied simultaneously to provide
Moreover, it quantifies many of the errors and impairments
multiple states. Even highly complex modulation formats
that degrade modulation quality.
such as Orthogonal Frequency Division Multiplexing (OFDM)
Modern communications systems have dramatically increased can be decomposed into magnitude and phase components.
the number of modulation formats in use. The RSA is
Magnitude and phase can be thought of as the length and
capable of analyzing the most common formats and has
the angle of a vector in a polar coordinate system. The
an architecture that allows for the analysis of new formats
same point can be expressed in Cartesian or rectangular
as they emerge.
coordinates (X,Y). The IQ format of the time samples stored
Amplitude, Frequency and Phase Modulation in memory by the RSA are mathematically equivalent to
Cartesian coordinates with I representing the horizontal or
RF carriers can transport information in many ways based on
X component and Q the vertical or Y component.
variations in the amplitude or phase of the carrier. Frequency
is the time derivative of phase. Frequency Modulation (FM)
Primer
TX Signal
I Mod Filter
e
nc
re
rfe
Data IQ
te
ift
In
Encoder Demux Up-Conversion pler
Sh
Clock Mod Filter Dop
Noise
Q Oth
er
Err
ors
Local
Transmitter Oscillator
Rx Filter I Symbol
IQ Frequency Data
Down- Phase & Demod Decoder
conversion Rx Filter Clock Recovery Clock
Q Symbol
Clock
Local
Oscillator Receiver
Figure 2-36. Typical digital communications system.
Q Digital Modulation
Figure 2-36 shows the signal processing in a typical digital
communications system. The transmit process starts with
de
itu
the data to be sent and a clock. The data and clock are
gn
Ma
passed through an encoder that re-arranges the data, adds

Phase
synchronization bits, and does encoding and scrambling
I
which might be used in the receiver to recover errors that
Magnitude = I 2 + Q2 inevitable occur when signals are transmitted over the air.
Q
Phase = tan –1
I( ) The data is then split into I and Q paths and filtered, changing
it from bits to analog waveforms which are then up-converted
Figure 2-35. Vector representation of magnitude and phase. to the appropriate channel and transmitted over the air.
The process of reception is the reverse of transmission with
Figure 2-35 illustrates the magnitude and phase of a vector some additional steps. The RF signal is downconverted to I
along with its I and Q components. AM demodulation and Q baseband signals which are passed through Rx filters
consists of computing the instantaneous magnitude for often designed to remove inter-symbol interference. The
each IQ sample and plotting the result over time. PM signal is then passed through an algorithm that recovers the
demodulation consists of computing the phase angle of the exact frequency, phase and data clock. This is necessary to
IQ samples and plotting them over time after accounting correct for multi-path delay and Doppler shift in the path
for the discontinuities of the arctangent function at ±P/2. and for the fact that the Rx and Tx local oscillators are not
FM is computed using the property that frequency is the usually synchronized. Once frequency, phase and clock are
time-derivative of phase. FM is the time-derivative of PM. recovered, the signal is demodulated and decoded, errors
are corrected and bits lost due to signal degradation in the
transmission path are recovered.
Primer
Rx Filter I Modulation
Analysis Views
Frequency,
IQ Phase &
Up-Conversion Demod Constellation
Rx Filter Clock Recovery Symbol table
Q EVM
Rho
Ideal Signal Frequency Frror
Local Reconstruction Magnitude Error
Oscillator Phase Error
Ideal I Ideal Q Origin Offset
Code Domain
RTSA Modulation Analysis Actual Q
Compare Power
Actual I More...
Figure 2-37. RSA modulation analysis block diagram.
The varieties of digital modulation are numerous and include Power Measurements and Statistics
the familiar FSK, BPSK, QPSK, GMSK, QAM, OFDM and
Tektronix RSAs can perform power measurements both in
others. Digital modulation is often combined with channel
the frequency domain and in the time domain. Time domain
assignments, filtering, power control, error correction and
measurements are made by integrating the power in the I
communications protocols to encompass a particular digital
and Q baseband signals stored in memory over a specified
communication standard whose purpose is to transmit
time interval. Frequency domain measurements are made
error-free bits of information between radios at opposite
by integrating the power in the spectrum over a specified
ends of a link. Much of the complexity incurred in a digital
frequency interval. Channel filters, required for many stan-
communication format is necessary to compensate for the
dards-based measurements, may be applied to yield the
errors and impairments that enter the system as the signal
resultant channel power. Calibration and normalization
travels over the air or other media.
parameters are also applied to maintain accuracy under
Figure 2-37 illustrates the signal processing steps required all specified conditions.
for a digital modulation analysis. The basic process is the
Communications standards often specify statistical
same as that of a receiver except that modulation accuracy
measurements for components and end-user devices.
measurements require a comparison between the received
RSAs have measurement routines to calculate statistics such
and ideal modulation waveforms. The method used in
as the Complementary Cumulative Distribution Function
most RSA modulation quality measurements is to use
(CCDF) of a signal which is often used to characterize the
the recovered symbols to reconstruct the mathematically
statistics including the Peak-Average Ratio (PAR) behavior
ideal I and Q signals. These ideal signals are compared
of complex modulated signals.
with the actual or degraded I and Q signals to generate
the required modulation analysis views and measurements.
Primer
pective
’s Pers
Viewer
Ti
m
e
ency
Frequ
Amplitude
Figure 3-1. DPX Spectrum showing multiple signals occupying the same frequencies
es
at different times. Fram
Figure 3-2. Spectrogram display.

Chapter 3: Real-Time Spectrum
Analyzer Measurements
This chapter describes the operational modes and 10.3 µsec in the RSA6000. DPX spectrum also provides a
measurements of Real-Time Spectrum Analyzers (RSA). true representation of multiple signals occupying the same
Several pertinent details such as the sampling rates, frequencies at different times as shown in Figure 3-1.
the number of Discrete Fourier Transform (DFT) points, DPX spectrum provides a Live RF display, allowing the user
measurement analysis limits and sample limits are product to see fleeting signals dramatically reducing the time required
dependent. Like the other measurement examples in this to discover and resolve problems. On some RSA models,
document, the information in this section applies specifically DPX Spectrum is not limited in span by its real-time band-
to the Tektronix RSA3000 Series and RSA6000 Series width. Swept DPX operates similar to the regular spectrum
of RSAs. display as it steps through multiple real-time frequency
segments, building a wide-span display with line traces
RSA Measurement Types
and the bitmap.
Tektronix RSAs can operate across the frequency, time,
Spectrum and Spectrogram Measurements
modulation and statistical domains. This section provides
Both the spectrum and spectrogram measurements are
an overview of the measurements available in the various
based on DFT analysis of data that is stored in memory.
RSA models.
These measurements can run either triggered or in free
Frequency Domain Measurements run mode. Spectrum and spectrogram measurements are
displayed separately in all RSAs. They are treated together
The basic frequency domain measurements described
here because both measurements operate on the same
in this chapter are the Live RF Digital Phosphor Display (DPX)
acquired data and are linked in both time and frequency
Spectrum Display measurements, the Spectrum Display
by markers.
measurements, and the Spectrogram Display measurement
capability. As shown in Figure 3-2, the spectrogram display is an
important measurement that provides an intuitive display
DPX Spectrum of how frequency and amplitude behavior change over
The DPX Spectrum measurement is the key to the RSA’s time. Like the spectrum display, the horizontal axis shows
ability to discover elusive signals that are missed by other the frequency range of the span. In the spectrogram, though
analyzers. The key performance parameter for DPX the vertical axis represents time, and amplitude is represented
technology in all Tektronix RSAs is minimum duration for by the color of the trace. Each “slice” of the spectrogram
a 100% Probability of Intercept (POI) of a single event. This corresponds to a single frequency spectrum calculated
basic figure of merit for signal discovery can be as brief as from one spectrum length of data.
Primer
Figure 3-3. Time-correlated views: Power vs. Frequency display (left) and spectrogram Figure 3-4. Standard SA mode showing an off-the-air measurement over a 1 GHz
display (right). frequency span using maxhold.
Figure 3-3 shows a screen shot displaying the spectrum and columns needed for displaying the frequency segment.
spectrogram displays for the signal illustrated in Figure 3-2. Compression is done by averaging pixel densities of the
On the spectrogram, the oldest frame is shown at the top points being combined together. The final swept bitmap
of the display. This measurement shows an RF signal whose contains a representation of the same pixel bitmap resolution,
frequency is changing over time, and it also reveals a low- just like the non-swept bitmaps. Line traces are also created
level transient signal that appears and disappears near the in full for each segment, and then horizontally compressed to
end of the time block. Since the data is stored in memory, the user-selected number of trace points for the full span.
a marker can be used to scroll “back in time” through the
spectrogram. In Figure 3-3, a marker has been placed on Time Domain Measurements
the transient event on the spectrogram display, which causes Frequency vs. Time
the corresponding spectrum display to be shown at that The Frequency vs. Time measurement displays frequency
particular point in time. on the vertical axis and time on the horizontal axis. While
Spectrum Measurements Beyond the Real-Time Bandwidth the spectrogram display can show the same information,
As shown in Figure 3-4, RSAs provides frequency domain Frequency vs. Time has two important differences. First,
measurements that emulate a traditional Swept Spectrum the Frequency vs. Time view has much better time domain
Analyzer (SA). For frequency spans that exceed the real-time resolution than the spectrogram, as described in detail below.
bandwidth of the instrument, this is achieved by tuning the Second, this measurement calculates a single frequency
RSA across the span of interest much like a traditional value for every point in time, which means that it cannot
spectrum analyzer and successive spectrum acquisitions display multiple RF signals like the spectrogram can.
are taken before tuning to the next step. The spectrogram is a compilation of DFT results and has a
When configured with the optional hardware, the RSA6000 line-by-line time resolution equal to the length of one DFT
Series can implement the Swept DPX capability. The analyzer frame. The Frequency vs. Time view has a time resolution of
“dwells” in each frequency segment for one ormore DPX one sample interval. If 1024 time domain samples were used
frames, each containing the results of up to 14,600 spectral in a calculating a spectrum, the resolution in this mode is
transforms. Dwell time is adjustable, so you can monitor each 1024 times finer than that of the spectrogram. This makes it
segment of the sweep for up to 100 seconds before moving easy to see small, brief frequency shifts in great detail. The
to the next step. While dwelling in a segment, the probability view acts similar to a frequency discriminator or a frequency
of intercept for signals within that frequency band is the same counter. Each of the sample points represents a frequency
as in normal, real-time spans: 100% capture of events as value, whether the span is a few hundred hertz or many
short as 10.3 µsec. A full pixel bitmap is created for every megahertz. Constant-frequency signals such as CW or AM
segment and compressed horizontally to the number of produce a flat, level display.
Primer
F1 F2
Time (Frames)
F2
Frequency
m Frames
F1
Frequency Time (Samples)
Spectrogram View Frequency vs. Time View

n Frames m x 1024 Samples
Figure 3-5. Comparison of spectrogram and Frequency vs. Time.
Figure 3-6. Spectrogram view of frequency settling over 5 MHz of frequency and
35 ms of time.
Figures 3-6, 3-7, and 3-8 show three different analysis

views of the same acquisition. As shown in Figure 3-6,
the frequency mask trigger (FMT) was used to capture a
transient signal coming from a transmitter having occasional
problems with frequency stability during turn-on. Since the
oscillator was not tuned to the frequency at the center of the
screen, the RF signal broke the frequency mask shown on the
left and caused a trigger. The spectrogram plot on the right
shows the frequency settling behavior of the device.
The next two figures show Frequency vs. Time displays of the
Figure 3-7. Frequency vs. Time view of frequency settling over 5 MHz of frequency
and 25 ms of time. same signal. Figure 3-7 shows the same frequency settling
behavior as the spectrogram using a 25 ms analysis length.
The Frequency vs. Time view provides the best results when Figure 3-8 shows the ability to zoom in to an analysis length
there is a relatively strong signal at one unique frequency. of 1 ms, showing the changes in frequency over time with
Figure 3-5 is a simplified illustration contrasting the Frequency much finer time domain resolution. This reveals a residual
vs. Time display with a spectrogram. The Frequency vs. Time oscillation on the signal even after it has settled to the
display is in some ways a zoomed-in view that magnifies a correct frequency.
portion of the spectrogram. This is very useful for examining
transient events such as frequency overshoot or ringing.
When there are multiple signals in the measured environment,
or one signal with an elevated noise level or intermittent
spurs, the spectrogram remains the preferred view. It provides
visualization of all the frequency and amplitude activity across
the chosen span.
Primer
Figure 3-8. Zooming in to view frequency settling over 50 kHz of frequency and 1ms Figure 3-9. Power vs. Time display.
of time.
For each time domain sample point, power is calculated as

follows:
The Power vs. Time display is also available in the Time

Overview display. The Time Overview display shows the
RF amplitude in the full acquisition bandwidth for the entire
acquisition while the Power vs. Time display shows the
time segment defined by the analysis offset and analysis
length displays.
Phase vs. Time
The Phase vs. Time is similar in concept to Power vs. Time.
The phase value for each IQ pair is computed and the result
is displayed as a function of time. The phase at each IQ
Figure 3-10. IQ vs. Time measurement of settling transient. sample is computed as follows:
Amplitude or Power vs. Time

The Power vs. Time display (Figure 3-9) shows how the
power of a signal changes on a sample by sample basis. IQ vs. Time
The amplitude of the signal is plotted in dBm on a logarithmic Transient IQ vs. Time (Figure 3-10) is another time domain
scale. This display is similar to an oscilloscope’s time domain view that displays the amplitudes of I and Q as a function
view in that the horizontal axis represents time. In contrast, of time. This measurement shows the raw I and Q output
the vertical axis shows power on a log scale instead of signals coming from the digital downconverter. As a result,
voltage on a linear scale. On the RSA3000 Series Spectrum this display is not synchronized to any modulation that might
analyzers, the trace represents the total power detected be present on the signal being analyzed. Some of the digital
within the span. The RSA6000 Series Spectrum analyzers demodulation measurements can also show IQ vs. Time after
offer the choice of a user-selectable filter in addition to the full synchronization.
span bandwidth. A constant power signal will yield a flat trace
display since there is no power change as time advances. This measurement can be another useful troubleshooting tool
for expert users, especially in terms of lending insight into
frequency and phase errors and instabilities.
Primer
Figure 3-11. AM demodulation analysis of a pulsed signal using amplitude shift keying Figure 3-12. FM demodulation analysis of a signal being modulated by a sine wave.
to encode data.
Modulation Domain Measurements

Analog Modulation Analysis
The analog demodulation mode provides measurements to
demodulate and analyze amplitude modulation (Figure 3-11),
frequency modulation (Figure 3-12), and phase modulation
(Figure 3-13). Just like the time domain measurements,
these tools are based on the concept of multi-domain
analysis, and the spectrum and analysis windows can
be positioned anywhere within the block shown in the
Time Overview window.
Figure 3-13. PM demodulation analysis showing phase instability over a long burst.
Primer
Figure 3-14. EVM analysis over time of 16 QAM signal reveals sinusoidal amplitude Figure 3-15. Constellation display showing phase instability in a PDC signal. Lower
distortion displayed in EVM vs. time result in lower right quadrant. right hand quadrant shows slewed symbol timing at constant amplitude representing
phase instability.
Digital Modulation Analysis

The digital modulation analysis function can demodulate and
analyze many common digital signals based on phase shift
keying (PSK), frequency shift keying (FSK), and Quadrature
Amplitude Modulation (QAM). The RSA provides a wide
range of measurements including: constellation, error-vector
magnitude (EVM), magnitude error, phase error, demodulated
IQ vs. time, symbol table, and eye diagram. To make these
measurements, it is necessary to properly configure variables
such as the modulation type, symbol rate, measurement
(receive) filter type and parameter such as roll-off factor of
a root-raised cosine filter, and reference filter type.
The RSA provides a powerful solution for characterizing
dynamic modulated signals by combining the digital
demodulation measurements of a Vector Signal Analyzer
Figure 3-16. Eye diagram display showing a PDC signal.
(VSA) with real-time triggering and time-correlated
multi-domain analysis, as illustrated in Figures 3-14,
3-15, and 3-16.
Primer
Figure 3-17. Modulation analysis of a W-CDMA handset under closed-loop power Figure 3-18. Spectrogram, constellation, EVM, and phase error vs. time of frequency-
control. The constellation display (lower right) shows the error associated with large hopping GSM signal.
glitches that occur during level transitions, which can be seen in the Power vs.
Time display.
tive
erspec
wer’s P
en Vie
y Scre
Displa
Code
ain
Dom r
P e
o w
Ti
m
e
nel
/Chan
Code
Figure 3-19. Illustration of the codogram display. Figure 3-20. Codogram measurement of W-CDMA compressed mode.
Standards-based Modulation Analysis Code Domain Measurements

The RSA also provides solutions for modulation analysis
Codogram Display
of many communications standards such as W-CDMA,
The codogram display (Figure 3-19) of the RSA adds a time
HSDPA/HSUPA, LTE, GSM/EDGE, CDMA2000, 1xEV-DO,
axis to code domain power measurements for CDMA-based
Zigbee, WiMax, WLAN (IEEE 802.11 a/b/g/n) and more.
communications standards. Like the spectrogram, the
Figures 3-17 and 3-18 show examples of standards-
codogram intuitively shows changes over time.
based modulation analysis.
Figure 3-20 is a W-CDMA codogram display from an RSA.
This particular codogram shows a simulated W-CDMA
compressed mode handoff in which the data rate is
momentarily increased to make room for brief, temporary
gaps in the transmission. These gaps allow dual-mode
W-CDMA/GSM user equipment to search for an available
GSM base station while remaining connected to a W-CDMA
Node B.
Primer
Figure 3-22. The marker on each display above represents the same point in time in
the acquisition memory.
Figure 3-21. CCDF measurement. Time-Correlated Multi-Domain Analysis

Once a signal has been acquired and stored in memory, it
can be analyzed using the wide variety of time-correlated
Statistical Domain Measurements
views available in the RSA, as illustrated in Figure 3-21.
Complementary Cumulative Distribution Function
The Complementary Cumulative Distribution Function (CCDF) This is especially useful for device troubleshooting and signal
view displays the probability that the signal power above characterization applications. All of these measurements
average power on the measured signal exceeds the amplitude are based on the same underlying set of time domain
displayed on the horizontal scale. The horizontal axis is sample data, which underscores two significant
scaled in dB above average signal power. Probability is architectural advantages:
displayed as a percent on the vertical scale. The vertical Comprehensive signal analysis in the frequency, time,
axis is logarithmic. and modulation domains based on a single acquisition.
CCDF analysis can be used to measure the time-varying crest Domain correlation to understand how specific events in
factor, which is important for many digital signals, the frequency, time, and modulation domains are related
especially those that use CDMA and OFDM. The crest factor based on a common time reference.
is the ratio of a signal’s peak power divided by its average Spectrum and Spectrogram
power, with the result expressed in dB: The RSA provides two views that are correlated in both time
and frequency, the spectrum display and the spectrogram
display. These two views can be seen in Figure 3-22, together
with the time overview and Amplitude vs. Time analysis. In
each display, the marker on screen corresponds to the same
The crest factor of the signal determines how linear a transmitter point within the acquisition memory.
or receiver must be in order to avoid unacceptable levels of
signal distortion. In Figure 3-21, the CCDF curves show the
measured signal in yellow and a Gaussian reference trace in
blue. The CCDF and crest factor are especially interesting
to designers who must balance power consumption and
distortion performance of devices such as amplifiers. When
CCDF is used over a selected period of a waveform and
compared to measures of distortion such as ACP, the user
can determine how statistical variations in the signal
create distortions in an amplifier output.
Primer
Figure 3-23. Time Overview, EVM vs. Time, Constellation and Spectrum with time-
correlated markers.
Correlating Modulation and Time Domain Results Figure 3-24. Multi-domain view showing Spectrogram, power vs. frequency, and
In the other real-time measurement modes for time domain power vs. time.
analysis and modulation domain analysis, the RSAs shows

The window in the bottom half of the screen (outlined in
multiple views of the captured signal as illustrated in
green) displays the results of the selected time or modulation
Figure 3-23.
analysis measurement.
The window in the lower right in Figure 3-23 displays Time
Figure 3-23 shows an example of frequency modulation
Overview (Power vs. Time) for the entire acquisition. This
analysis, and Figure 3-24 shows an example of transient
overview shows all of the data that was acquired in the block,
Power vs. Time analysis. The green analysis window can be
and it serves as the index for the other analysis windows.
positioned anywhere within the time record shown in the Time
The window in the upper right in Figure 3-24 shows the Overview, which has corresponding green bars to indicate its
spectrum or Power vs. Frequency for one frame of the position. In addition, the width and position of the analysis
spectrogram shown on the upper left. Just like the display window can be adjusted using the analysis length and
in Figure 3-23, this is the spectrum of one frame of data, analysis offset controls.
and it is possible to scroll through the entire time record
to see the spectrum at any point in time. This is done by
adjusting the spectrum offset. Also note that there is a
purple bar in the spectrogram window that indicates
position in time that corresponds to the frequency domain
display in the spectrum window.
Primer
Chapter 4: Applications Using the Radar

Real-Time Spectrum Analyzer Integration of pulse measurements into a single tool for time
and frequency domain simplifies radar testing and improves
Today's RF/Microwave world is merging the digital computing
test reproducibility. The RSA has the combined test capability
and traditional analog RF technologies. This integration of
that replaces several common tools used in radar transmitter
digital and analog RF is presenting engineers with a new
test analysis (oscilloscopes, spectrum analyzers, vector
highly complex environment, necessitating a new generation
analyzers, power meters, and peak-power meters). In
of RF/Microwave test tools. Tektronix delivers the signal
addition, the Digital Phosphor Display (DPX) can display
generation and analysis capabilities required to overcome
interference components and the products non-linearities
the most challenging microwave and RF design challenges
that go undetected with other tools.
with confidence. For the latest information on applications
using the real-time spectrum analyzers, visit the web: RFID
www.tektronix.com/rf
RFID systems are used for a variety of applications: Near-
Pulse Measurements Field Communications (NFC) for transaction information
on cellular telephones; recently mandated Tire Pressure
Tektronix Real-Time Spectrum Analyzers (RSAs) are ideally
Monitoring Systems (TPMS) for vehicle safety; Real-Time
suited for making pulse measurements. Automatic pulse
Location Systems (RTLS) for the tracking of goods, and
measurement software can be included on all models of RSA.
even the asset tracking systems for commercial goods,
Extensive analysis on individual and pulse trend information
with the ePC Global initiatives. RFID systems operate over
can be selected. Unlike traditional spectrum analyzers, the
a wide range of frequencies from HF frequencies (135 kHz)
various models of RSA are specified for system risetime/fall-
to microwave (2.4 GHz) and at various power levels for
time (to 10 ns), minimum pulse duration (as short as 50 ns),
both passive and active modes. However, as many RFID
and modulation bandwidths to 110 MHz. For faster risetimes,
systems also operate in the unlicensed Industrial Scientific
pulses, and wider modulation bandwidths, the SignalVu™
and Medical (ISM) bands, these systems become susceptible
software can be used directly on the Tektronix DPO7000/70000
to collisions of other RFID systems and communications
and DSA70000 Series oscilloscopes for automatic pulse
operating in the same bands.
measurements in the 10s of picoseconds and bandwidths
up to 20 GHz. The ability to automate pulse measurements is important for
designing to the function of all RFID systems. The RSAs have
SignalVu is embedded into the acquisition engine of the
the most comprehensive support of industry, international,
oscilloscope and provides access to all the oscilloscope
and proprietary RFID standards. Tag and reader/interrogator
functionality and memory for measurements with a
performance and interaction can be completely analyzed with
similar user interface to the RSA6000 Series Real-time
quick, easy, and repeatable measurement setups and deep
spectrum analyzer.
memory capture
Individual pulse information for carrier frequency, risetime,
falltime, droop, ripple, occupied spectrum, and pulsewidth
are among the many pulse measurements that can be
selected and characterized on the pulse measurement
analysis software.
Pulse trend statistics information includes PRF, duty cycle,
and pulse-to-pulse phase information. In addition, on the
RSA6000 Series, statistical information can be displayed
as a trend, or Fast Fourier Transforms (FFTs) can be
performed over the sampled data to reveal important
diagnostic information.
Primer
Spectrum Management/Surveillance By eliminating the need to switch modes between time and
frequency analysis, RSAs have the ability to time-correlate
Whether you are securing a facility or protecting your
capture data and analyze in the multi-domains of time,
spectrum investment, interference must be mitigated.
frequency, modulation, code, and statistics. Capture your
The detection of low-level and intermittent interference
data once, and analyze completely.
in the presence of larger signals and the characterization
of impulsive noise are not possible with any certainty The combination of DPX Spectrum and FMT provide a
using most spectrum analyzer tools. powerful and unique capability for Radio Debug. DPX can
discover unusual transient events such as impulse noise
With DPX displaying Live RF, you will see signals you never
caused by software faults or other low-level spectrum noise
knew existed. RSAs will change the way you search and
while the DPX Density and FMT can be set up to isolate
discover elusive signals. With industry-leading performance
those events. Using the external trigger output in conjunction
for combined bandwidth and dynamic range, RSAs are the
with the FMT allows you to trigger other instruments.
only spectrum analyzers specified with a 100% probability
Triggering oscilloscopes and logic analyzers enables you
of detecting signals within a band of interest.
to isolate events deep into your block diagram so that the
Unique triggering functionality such as DPX Density and root cause can be isolated. Tektronix iView capability and
Frequency Mask Trigger (FMT) enable the RSA to trigger Mixed-Signal Oscilloscopes allow for the accurate time
on low-level signals or impulses with a 100% probability for alignment and display of analog and digital signals to with
capture and analysis. picosecond accuracy. Subroutines and lines of code can be
time correlated to the actual spectrum event of interest.
Radio Communications
More and more of the control and operation of modern radio
communications are controlled by software. Digital circuits
are replacing analog functions for dynamic operation of
modulation, power, channel control, and channel loading.
These modern radios can change power, frequency,
modulation, coding, and statistics rapidly over time.
Primer
Power Amplifier Testing become very sensitive to these unintentional, internally

generated sources of modulation. Compounding the
Voice communications in mobile communications tend to
problem, the intermittent nature of these disturbances
operate at some level of steady traffic state. As high rate
has traditionally made the detection and isolation of these
data services become integrated into modern radio
events difficult and time consuming.
communication standards, (HSDPA/HSUPA, LTE, 1xEV,
WiMax), the output power needs to be rapidly adapt to the The Real-Time Spectrum Analyzers have unique ability to
changing load. Modulation quality, spectral regrowth, and display the presence of these events with certainty using DPX
power efficiency are a constant trade-off in amplifier design. spectrum display at >292,000 spectrum updates per second.
Isolation, characterization, measurement, and correlation to
Whether advances in linearization are drive by the need for
root cause can be simplified using a combination of the
better efficiency, reduced cost, or performance, various digital
unique Frequency Mask Trigger and External Trigger output
techniques using some for of adaptive digital predistortion
on the Real-Time Spectrum Analyzers.
(DPD) have become common in cellular amplifier design.
These designs are susceptible to memory effects, bias feed EMI/EMC
through, and other transient spectral regrowth abnormalities.
In the world of Electromagnetic Compatibility (EMC), different
With the ability to reveal the existence of memory effects equipment and techniques are used at different stages of
using DPX Spectrum, the RSA can provide unique insight design, qualification, and commissioning. While governed
for amplifier designers. Add the capability of correlated by industry, national, and international conformance
multi-domain analysis for amplitude statistics (CCDF), requirements, many sensitivity electronic devices need
modulation quality (EVM), and Adjacent Channel Power to test more than compliance to ensure functionality. The
(ACP) analysis, and the RSAs provides invaluable insight performance impact of low-level noise, susceptance to
for amplifier designers. internal impulse noise, and the stability of clocks and Phase
Locked Loops must be understood to ensure proper device
Microphonics and Phase-Hits Analysis
functionality and RF manufacturing integrity. Elusive events
Microphonics and phase-hit problems have a reputation for are difficult to discover, trigger, capture, and analyze without
disrupting development programs and stopping production the unique insight and confidence provided by Tektronix
lines. As more of today’s modern communications systems real-time spectrum analyzers.
and electronics have become portable, these systems have
Primer
Chapter 5: Terminology
Glossary Distortion - Degradation of a signal, often a result of
Acquisition - An integer number of time-contiguous samples. nonlinear operations, resulting in unwanted frequency
components. Harmonic and intermodulation distortions
Acquisition Time - The length of time represented by are common types.
one acquisition.
DPX - Digital Phosphor analysis - A signal analysis and
Amplitude - The magnitude of an electrical signal. compression methodology that allows the live view of
Amplitude Modulation (AM) - The process in which the time-changing signals allowing the discovery of rare
amplitude of a sine wave (the carrier) is varied in accordance transient events.
with the instantaneous voltage of a second electrical signal DPX Spectrum - DPX technology applied to spectrum
(the modulating signal). analysis. DPX Spectrum provides a Live RF view as well
Analysis Time - A subset of time-contiguous samples from as the observation frequency domain transients.
one block, used as input to an analysis view. Dynamic Range - The maximum ratio of the levels of two
Analysis View - The flexible window used to display real-time signals simultaneously present at the input which can be
measurement results. measured to a specified accuracy.
Carrier - The RF signal upon which modulation resides. FFT - Fast Fourier Transform – A computationally efficient
method of computing a Discrete Fourier Transform (DFT). A
Carrier Frequency - The frequency of the CW component common FFT algorithm requires that the number of input and
of the carrier signal. output samples are equal and a power of 2 (2,4,8,16,…).
Center Frequency - The frequency corresponding to Frequency - The rate at which a signal oscillates, expressed
the center of a frequency span of a spectrum the as hertz or number of cycles per second.
analyzer display.
Frequency Domain View - The representation of the power
CZT-Chirp-Z transform - A computationally efficient method of the spectral components of a signal as a function of
of computing a Discrete Fourier Transform (DFT). CZTs offer frequency; the spectrum of the signal.
more flexibility for example in selecting the number of output
frequency points than the conventional FFT at the expense of Frequency Drift - Gradual shift or change a signal frequency
additional computations. over the specified time, where other conditions remain
constant. Expressed in hertz per second.
Codogram - Code Channel vs. Time vs. Power display where
the CDMA code channel is represented on x-axis and time Frequency Mask Trigger - A flexible real-time trigger based
respectively on the y-axis. The power level is expressed by on specific events that occur in the frequency domain. The
the color. triggering parameters are defined by a graphical mask.
CW Signal - Continuous wave signal. A sine wave. Frequency Modulation (FM) - The process in which the
frequency of an electrical signal (the carrier) is varied
dBfs - A unit to express power level in decibels referenced according to the instantaneous voltage of a second
to full scale. Depending on the context, this is either the full electrical signal (the modulating signal).
scale of the display screen or the full scale of the ADC.
Frequency Range - The range of frequencies over which
dBm - A unit to express power level in decibels referenced a device operates, with lower and upper bounds.
to 1 milliwatt.
Frequency Span - A continuous range of frequencies
dBmV - A unit to express voltage levels in decibels extending between two frequency limits.
referenced to 1 millivolt.
Marker - A visually identifiable point on a waveform trace,
Decibel (dB) - Ten times the logarithm of the ratio of one used to extract a readout of domain and range values
electrical power to another. represented by that point.
DFT-Discrete Fourier transform - A mathematical process Modulate - To vary a characteristic of a signal, typically in
to calculate the frequency spectrum of a sampled time order to transmit information.
domain signal.
Noise - Unwanted random disturbances superimposed on
Display Line - A horizontal or vertical line on a waveform a signal which tend to obscure that signal.
display, used as a reference for visual (or automatic)
comparison with a given level, time, or frequency. Noise Floor - The level of noise intrinsic to a system that
represents the minimum limit at which input signals can be
observed; ultimately limited by thermal noise (kTB).
Primer
Noise Bandwidth (NBW) - The exact bandwidth of a filter Acronym Reference

that is used to calculate the absolute power of noise or
ACP: Adjacent Channel Power
noise-like signals in dBm/Hz.
ADC: Analog-to-Digital Converter
Probability of Intercept - The certainty to which a signal
can be detected within defined parameters. AM: Amplitude Modulation
Real-Time Bandwidth - The frequency span over which BH4B: Blackman-Harris 4B Window
real-time seamless capture can be performed, which is a BW: Bandwidth
function of the digitizer and the IF bandwidth of a Real-Time
Spectrum Analyzer. CCDF: Complementary Cumulative Distribution Function
Real-Time Seamless Capture - The ability to acquire and CDMA: Code Division Multiple Access
store an uninterrupted series of time domain samples that CW: Continuous Wave
represent the behavior of an RF signal over a long period dB: Decibel
of time.
dBfs: dB Full Scale
Real-Time Spectrum Analysis - A spectrum analysis
technique based on Discrete Fourier Transforms (DFT) that DDC: Digital Downconverter
is capable of continuously analyzing a bandwidth of interest DFT: Discrete Fourier Transform
without time gaps. Real-Time Spectrum Analysis provides
DPX: Digital Phosphor Display, Spectrum, etc.
100% probability of display and trigger of transient signal
fluctuations within the specified span, resolution bandwidth DSP: Digital Signal Processing
and time parameters. EVM: Error Vector Magnitude
Real-Time Spectrum Analyzer - Instrument capable FFT: Fast Fourier Transform
of measuring elusive RF events in RF signals, triggering
on those events, seamlessly capturing them into memory, FM: Frequency Modulation
and analyzing them in the frequency, time, and FSK: Frequency Shift Keying
modulation domains.
IF: Intermediate Frequency
Reference Level - The signal level represented by the
IQ: In-Phase Quadrature
uppermost graticule line of the analyzer display.
LO: Local Oscillator
Resolution Bandwidth (RBW) - The width of the narrowest
measurable band of frequencies in a spectrum analyzer NBW: Noise Bandwidth
display. The RBW determines the analyzer’s ability to OFDM: Orthogonal Frequency Division Multiplexing
resolve closely spaced signal components.
PAR: Peak-Average Ratio
Sensitivity - Measure of a spectrum analyzer’s ability to
PM: Phase Modulation
display minimum level signals, usually expressed as
Displayed Average Noise Level (DANL). POI: Probability of Intercept
Spectrogram - Frequency vs. Time vs. amplitude display PRBS: Pseudorandom Binary Sequence
where the frequency is represented on x-axis and time on PSK: Phase Shift Keying
the y-axis. The power is expressed by the color.
QAM: Quadrature Amplitude Modulation
Spectrum - The frequency domain representation of a
signal showing the power distribution of its spectral QPSK: Quadrature Phase Shift Keying
component versus frequency. RBW: Resolution Bandwidth
Spectrum Analysis - Measurement technique for RF: Radio Frequency
determining the frequency content of an RF signal. RMS: Root Mean Square
Vector Signal Analysis - Measurement technique for RSA: Real-Time Spectrum Analyzer
charactering the modulation of an RF signal. Vector
analysis takes both magnitude and phase into account. SA: Spectrum Analyzer
VSA: Vector Signal Analyzer
Contact Tektronix:
ASEAN / Australasia (65) 6356 3900
Austria +41 52 675 3777
Balkans, Israel, South Africa and other ISE Countries +41 52 675 3777
Belgium 07 81 60166
Brazil +55 (11) 3759-7627
Canada 1 (800) 661-5625
Central East Europe, Ukraine and the Baltics +41 52 675 3777
Central Europe & Greece +41 52 675 3777
Denmark +45 80 88 1401
Finland +41 52 675 3777
France +33 (0) 1 69 86 81 81
Germany +49 (221) 94 77 400
Hong Kong (852) 2585-6688
India (91) 80-42922600
Italy +39 (02) 25086 1
Japan 81 (3) 6714-3010
Luxembourg +44 (0) 1344 392400
Mexico, Central/South America & Caribbean 52 (55) 54247900
Middle East, Asia and North Africa +41 52 675 3777
The Netherlands 090 02 021797
Norway 800 16098
People’s Republic of China 86 (10) 6235 1230
Poland +41 52 675 3777
Portugal 80 08 12370
Republic of Korea 82 (2) 6917-5000
Russia & CIS +7 (495) 7484900
South Africa +27 11 206 8360
Spain (+34) 901 988 054
Sweden 020 08 80371
Switzerland +41 52 675 3777
Taiwan 886 (2) 2722-9622
United Kingdom & Ireland +44 (0) 1344 392400
USA 1 (800) 426-2200
For other areas contact Tektronix, Inc. at: 1 (503) 627-7111
Contact information updated 4 August 2009
For Further Information

Tektronix maintains a comprehensive, constantly expanding
collection of application notes, technical briefs and other
resources to help engineers working on the cutting edge of
technology. Please visit www.tektronix.com
Copyright © 2009, Tektronix. All rights reserved. Tektronix products are covered
by U.S. and foreign patents, issued and pending. Information in this publication
supersedes that in all previously published material. Specification and price
change privileges reserved. TEKTRONIX and TEK are registered trademarks
of Tektronix, Inc. All other trade names referenced are the service marks,
trademarks or registered trademarks of their respective companies.
08/09 EA/WWW 37W-17249-4

Fundamentals of Real Time Spectrum Analysis

Uploaded by

Copyright:

Available Formats

Fundamentals of Real Time Spectrum Analysis

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

Fundamentals of Real Time Spectrum Analysis

Uploaded by

Copyright:

Available Formats

Primer

Chapter 1: Introduction and Overview

Modern RF Measurement Challenges Capturing burst transmissions, glitches,

a) Swept Tuned Spectrum Analyzer (SA)

Band-Pass Local Acquisition Bandwidth Post Capture Processing

b) Vector Signal Analyzer (VSA)

Modern FFT-Based Analyzers

Real Time Engine

Real-Time Display X-Y

Real-Time Bandwidth Display Processing

c) Real-Time Spectrum Analyzer (RSA6000 Series + Option 110)

Frame Frame Frame Frame

Frame Frame Frame Frame

FFT FFT FFT FFT

Processing Time < Acquisition time!

Frame Frame Frame Frame

LF or IQ section (if present)

Figure 2-1. Real-Time Spectrum Analyzer RF/IF block diagram.

Frequency Conversion/IF Section

Real-Time Digital Signal Processing

Figure 2-4. IF to IQ conversion in a Real-Time Spectrum Analyzer.

IF Digitizer Internal Alignment

Digitized IF Signal IF Signal

Decimation longer time records for an equivalent number of samples.

DFT-Based Spectrum Analysis*

ADC DFT Engine

Equivalent Bank of Filters

Real-Time Spectrum Analysis Discovering and Capturing Transient Events

is represented as a color. The real-time DFT effectively

occurring between the time segments on which the FFTs

Frame 1 Frame 2 Frame 3 Frame 2 Frame 2 Frame 2

Windowing The choice of window function depends on its frequency

Input Signal: Frequency = 4.7222 Hz : Sample Frequency = 20 Hz DFT (X[k])

FFT CZT: Start Frequency = 2.5 Hz : Stop Frequency = 7.4 Hz

0.6 and reject others. While traditional

-0.2 Frequency Response vs.

Frequency Sampled Filter

M-1 M-2 M-3 M-4 M-5 1 0

impulse response -0.2

stored in registers Numerical Convolution

The DPX Display Engine Number of Occurrences Color

that the area of interest in the waveform

The analysis length determines the length of time for which

Real-Time Triggering and Acquisition

First In/First Out (FIFO) Memory

Figure 2-30. Triggering in digital acquisition systems.

Trigger Source Trigger Signal Setting Units Time Uncertainty Notes

Table 2-1. Comparison of Real-Time Spectrum Analyzers trigger sources.

External Trigger Processing Block

Figure 2-31. Real-Time Spectrum Analyzer trigger processing.

violations and much more. A full DFT is required to compare

Constructing a Frequency Mask

Demodulation is therefore the time derivative of Phase Modulation (PM).

Figure 2-36. Typical digital communications system.

passed through an encoder that re-arranges the data, adds

Figure 2-37. RSA modulation analysis block diagram.

Figure 3-2. Spectrogram display.

Frequency Time (Samples)

Spectrogram View Frequency vs. Time View

Figure 3-5. Comparison of spectrogram and Frequency vs. Time.