spectrum analysis basics agilent 5952-0292
Post on 13-Oct-2015
104 Views
Preview:
DESCRIPTION
TRANSCRIPT
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
1/89
Spectrum Analysis BasicsApplication Note 150
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
2/89
2
Agilent Technologies dedicates this application note to Blake Peterson.
Blakes outstanding service in technical support reached customers in all corners of the worldduring and after his 45-year career with Hewlett-Packard and Agilent Technologies. For manyyears, Blake trained new marketing and sales engineers in the ABCs of spectrum analyzer
technology, which provided the basis for understanding more advanced technology. He iswarmly regarded as a mentor and technical contributor in spectrum analysis.
Blakes many accomplishments include:
Authored the original edition of the Spectrum Analysis Basics application note and con-tributed to subsequent editions
Helped launch the 8566/68 spectrum analyzers, marking the beginning of modernspectrum analysis, and the PSA Series spectrum analyzers that set new performancebenchmarks in the industry when they were introduced
Inspired the creation of Blake Peterson Universityrequired training for all engineeringhires at Agilent
As a testament to his accomplishments and contributions, Blake was honored withMicrowaves & RF magazines first Living Legend Award in 2013.
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
3/89
3
Table of Contents
Chapter 1 Introduction .......................................................................................................5
Frequency domain versus time domain ...................................................................................5
What is a spectrum? ....................................................................................................................6
Why measure spectra? ................................................................................................................6Types of signal analyzers .............................................................................................................8
Chapter 2 Spectrum Analyzer Fundamentals ................................................................9
RF attenuator.................................................................................................................................10
Low-pass filter or preselector ....................................................................................................10
Tuning the analyzer .....................................................................................................................11
IF gain .............................................................................................................................................12
Resolving signals .........................................................................................................................13
Residual FM ...................................................................................................................................15
Phase noise ...................................................................................................................................16
Sweep time ....................................................................................................................................18Envelope detector .......................................................................................................................20
Displays ..........................................................................................................................................21
Detector types ...............................................................................................................................22
Sample detection .........................................................................................................................23
Peak (positive) detection ............................................................................................................24
Negative peak detection .............................................................................................................24
Normal detection .........................................................................................................................24
Average detection ........................................................................................................................27
EMI detectors: average and quasi-peak detection ...............................................................27
Averaging processes ...................................................................................................................28
Time gating ....................................................................................................................................31
Chapter 3 Digital IF Overview ........................................................................................36
Digital filters ..................................................................................................................................36
All-digital IF ....................................................................................................................................37
Custom digital signal processing ..............................................................................................38
Additional video processing features .....................................................................................38
Frequency counting ....................................................................................................................38
More advantages of all-digital IF ..............................................................................................39
Chapter 4 Amplitude and Frequency Accuracy ...........................................................40
Relative uncertainty ....................................................................................................................42Absolute amplitude accuracy ....................................................................................................42
Improving overall uncertainty ....................................................................................................43
Specifications, typical performance and nominal values ....................................................43
Digital IF architecture and uncertainties .................................................................................43
Amplitude uncertainty examples ..............................................................................................44
Frequency accuracy .....................................................................................................................44
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
4/89
4
Chapter 5 Sensitivity and Noise .....................................................................................46
Sensitivity.......................................................................................................................................46
Noise floor extension ..................................................................................................................48
Noise figure ...................................................................................................................................49
Preamplifiers .................................................................................................................................50
Noise as a signal ..........................................................................................................................53
Preamplifier for noise measurements .....................................................................................54
Chapter 6 Dynamic Range ...............................................................................................55
Dynamic range versus internal distortion ..............................................................................55
Attenuator test .............................................................................................................................56
Noise ...............................................................................................................................................57
Dynamic range versus measurement uncertainty ................................................................58
Gain compression ........................................................................................................................60
Display range and measurement range ..................................................................................60
Adjacent channel power measurements ................................................................................61
Chapter 7 Extending the Frequency Range...................................................................62
Internal harmonic mixing ............................................................................................................62
Preselection ...................................................................................................................................66
Amplitude calibration ..................................................................................................................68
Phase noise ..................................................................................................................................68
Improved dynamic range ............................................................................................................69
Pluses and minuses of preselection ........................................................................................70
External harmonic mixing ...........................................................................................................71
Signal identification .....................................................................................................................73
Chapter 8 Modern Signal Analyzers ..............................................................................76
Application-specific measurements.........................................................................................76
The need for phase information ............................................................................................... 77
Digital modulation analysis ........................................................................................................79
Real-time spectrum analysis .....................................................................................................80
Chapter 9 Control and Data Transfer .............................................................................81
Saving and printing data .............................................................................................................81
Data transfer and remote instrument control . ......................................................................81
Firmware updates ........................................................................................................................82Calibration, troubleshooting, diagnostics and repair ............................................................82
Summary ...............................................................................................................................83
Glossary of Terms .................................................................................................................83
Table of Contentscontinued
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
5/89
5
signal into its frequency-domain equivalent.Measurements in the frequency domaintell us how much energy is present at eachparticular frequency. With proper filtering, awaveform such as the one shown in Figure1-1 can be decomposed into separate sinu-
soidal waves, or spectral components, whichwe can then evaluate independently. Eachsine wave is characterized by its amplitudeand phase. If the signal we wish to analyzeis periodic, as in our case here, Fourier saysthat the constituent sine waves are sepa-rated in the frequency domain by 1/T, whereT is the period of the signal2.
Some measurements require that we pre-serve complete information about the signalfrequency, amplitude and phase. However,another large group of measurements can bemade without knowing the phase relation-ships among the sinusoidal components.This type of signal analysis is called spec-trum analysis. Because spectrum analysis issimpler to understand, yet extremely useful,
This application noteexplains the fundamentalsof swept-tuned, superhet-
erodyne spectrum analyzer s anddiscusses the latest advances in
spectrum analyzer capabilities.
At the most basic level, a spectrum analyzercan be described as a frequency-selective,peak-responding voltmeter calibrated todisplay the rms value of a sine wave. It isimportant to understand that the spectrumanalyzer is not a power meter, even thoughit can be used to display power directly. Aslong as we know some value of a sine wave(for example, peak or average) and knowthe resistance across which we measurethis value, we can calibrate our voltmeterto indicate power. With the advent of digitaltechnology, modern spectrum analyzershave been given many more capabilities. Inthis note, we describe the basic spectrumanalyzer as well as additional capabilitiesmade possible using digital technology anddigital signal processing.
Frequency domain versustime domain
Before we get into the details of describinga spectrum analyzer, we might first askourselves: Just what is a spectrum
and why would we want to analyze it?Our normal frame of reference is time.We note when certain events occur. Thisincludes electrical events. We can use anoscilloscope to view the instantaneousvalue of a particular electrical event(or some other event converted to voltsthrough an appropriate transducer) as afunction of time. In other words, we usethe oscilloscope to view the waveform of asignal in the time domain.
Fourier1theory tells us any time-domainelectrical phenomenon is made up ofone or more sine waves of appropriatefrequency, amplitude, and phase. In otherwords, we can transform a time-domain
Chapter 1. Introduction
we begin by looking first at how spectrumanalyzers perform spectrum analysis mea-surements, starting in Chapter 2.
Theoretically, to make the transformationfrom the time domain to the frequency
domain, the signal must be evaluated overall time, that is, over infinity. However, inpractice, we always use a finite time peri-od when making a measurement. You alsocan make Fourier transformations from thefrequency to the time domain. This casealso theoretically requires the evaluation ofall spectral components over frequenciesto infinity. In reality, making measurementsin a finite bandwidth that captures mostof the signal energy produces acceptableresults. When you perform a Fouriertransformation on frequency domain data,the phase of the individual componentsis indeed critical. For example, a squarewave transformed to the f requency domainand back again could turn into a sawtoothwave if you do not preserve phase.
1. Jean Baptiste Joseph Fourier, 1768-1830. A French mathematician and physicist who discovered that periodic functions can be expanded into a series of sines and cosines.
2. If the time signal occurs only once, then T is infinite, and the frequency representation is a continuum of sine waves.
Figure 1-1. Complex time-domain signal
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
6/89
6
Why measure spectra?
The frequency domain also has itsmeasurement strengths. We have alreadyseen in Figures 1-1 and 1-2 that thefrequency domain is better for determiningthe harmonic content of a signal. Peopleinvolved in wireless communications
are extremely interested in out-of-bandand spurious emissions. For example,cellular radio systems must be checked forharmonics of the carrier signal that mightinterfere with other systems operating atthe same frequencies as the harmonics.Engineers and technicians are also veryconcerned about distortion of the messagemodulated onto a carrier.
Third-order intermodulation (two tones of acomplex signal modulating each other) canbe particularly troublesome because thedistortion components can fall within the
band of interest, which means they cannotbe filtered away.
What is a spectrum?
So what is a spectrum in the context of thisdiscussion? A spectrum is a collection ofsine waves that, when combined properly,produce the time-domain signal underexamination. Figure 1-1 shows the waveformof a complex signal. Suppose that we were
hoping to see a sine wave. Although thewaveform certainly shows us that the signalis not a pure sinusoid, it does not give usa definitive indication of the reason why.Figure 1-2 shows our complex signal inboth the time and frequency domains. Thefrequency-domain display plots the ampli-tude versus the frequency of each sine wavein the spectrum. As shown, the spectrumin this case comprises just two sine waves.We now know why our original waveformwas not a pure sine wave. It contained asecond sine wave, the second harmonic inthis case. Does this mean we have no need
to perform time-domain measurements? Notat all. The time domain is better for manymeasurements, and some can be made onlyin the time domain. For example, pure time-domain measurements include pulse riseand fall times, overshoot and ringing.
Spectrum monitoring is another importantfrequency-domain measurement activity.Government regulatory agencies allocatedifferent frequencies for various radioservices, such as broadcast television andradio, mobile phone systems, police andemergency communications, and a host ofother applications. It is critical that eachof these services operates at the assignedfrequency and stays within the allocatedchannel bandwidth. Transmitters and otherintentional radiators often must operateat closely spaced adjacent frequencies. Akey performance measure for the poweramplifiers and other components usedin these systems is the amount of signalenergy that spills over into adjacentchannels and causes interference.
Electromagnetic interference (EMI) is aterm applied to unwanted emissions fromboth intentional and unintentional radiators.These unwanted emissions, either radiatedor conducted (through the power linesor other interconnecting wires), mightimpair the operation of other systems.Almost anyone designing or manufacturingelectrical or electronic products must test foremission levels versus frequency accordingto regulations set by various governmentagencies or industry-standard bodies.
Noise is often the signal you want tomeasure. Any active circuit or device willgenerate excess noise. Tests such as noisefigure and signal-to-noise ratio (SNR) are
important for characterizing the performanceof a device and its contribution to overallsystem performance.
Figures 1-3 through 1-6 show some ofthese measurements on an X-Seriessignal analyzer.
Time domain
measurements
Frequency domain
measurements
Figure 1-2. Relationship between time and frequency domain
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
7/89
7
Figure 1-3. Harmonic distortion test of a transmitter
Figure 1-6. Radiated emissions plotted against CISPR11 limits as part of an
EMI test
Figure 1- 5. Two-tone test on an RF power amplifier
Figure 1-4. GSM radio signal and spectral mask showing limits of
unwanted emissions
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
8/89
8
Advanced technology also has allowedcircuits to be miniaturized. As a result,rugged portable spectrum analyzers suchas the Agilent FieldFox simplify tasks suchas characterizing sites for transmittersor antenna farms. Zero warm-up timeeliminates delays in situations involvingbrief stops for quick measurements. Dueto advanced calibration techniques, fieldmeasurements made with these handheldanalyzers correlate with lab-grade bench-topspectrum analyzers within 10ths of a dB.
In this application note, we concentrateon swept amplitude measurements, onlybriefly touching on measurements involvingphasesee Chapter 8.
Note: When computers became Hewlett-Packards dominant business, it created andspun off Agilent Technologies in the late1990s to continue the test and measure-ment business. Many older spectrum analyz-ers carry the Hewlett-Packard name but aresupported by Agilent.
This application note will give you insightinto your particular spectrum or signalanalyzer and help you use this versatileinstrument to its maximum potential.
Types of signal analyzers
The first swept-tuned superheterodyneanalyzers measured only amplitude.However, as technology advanced and com-munication systems grew more complex,phase became a more important part of themeasurement. Spectrum analyzers, now
often labeled signal analyzers, have keptpace. By digitizing the signal, after one ormore stages of frequency conversion, phaseas well as amplitude is preserved and canbe included as part of the information dis-played. So todays signal analyzers such asthe Agilent X-Series combine the attributesof analog, vector and FFT (fast Fouriertransform) analyzers. To further improvecapabilities, Agilents X-Series signal analyz-ers incorporate a computer, complete witha removable disk drive that allows sensitivedata to remain in a controlled area shouldthe analyzer be removed.
More information
For additional information onvector measurements, seeVector Signal Analysis BasicsApplication Note, (literaturenumber 5989-1121EN). Forinformation on FFT analyzersthat tune to 0 Hz, see the Webpage for the Agilent 35670A atwww.agilent.com/find/35670A.
http://www.agilent.com/find/35670Ahttp://www.agilent.com/find/35670A -
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
9/89
9
This chapter focuses onthe fundamental theory ofhow a spectrum analyzer
works. While todays technol-ogy makes it possible to replace
many analog circuits withmodern digital implementations,it is useful to understandclassic spectrum analyzerarchitecture as a starting pointin our discussion.
In later chapters, we will look at the capabili-ties and advantages that digital circuitrybrings to spectrum analysis. Chapter 3discusses digital architectures used inspectrum analyzers available today.
Figure 2-1 is a simplified block diagramof a superheterodyne spectrum analyzer.Heterodyne means to mix; that is, to trans-late frequency. And super refers to super-audio frequencies, or frequencies above theaudio range. In the Figure 2-1 block diagram,
we see that an input signal passes throughan attenuator, then through a low-pass filter(later we will see why the filter is here) to amixer, where it mixes with a signal from thelocal oscillator (LO). Because the mixer isa non-linear device, its output includes not
only the two original signals, but also theirharmonics and the sums and differences ofthe original frequencies and their harmonics.If any of the mixed signals falls within thepass band of the intermediate-frequency (IF)filter, it is further processed (amplified andperhaps compressed on a logarithmic scale).It is essentially rectified by the envelopedetector, filtered through the low-pass filterand displayed. A ramp generator creates thehorizontal movement across the display fromleft to right. The ramp also tunes the LO soits frequency change is in proportion to theramp voltage.
If you are familiar with superheterodyne AMradios, the type that receive ordinary AMbroadcast signals, you will note a strongsimilarity between them and the block dia-gram shown in Figure 2-1. The differences
are that the output of a spectrum analyzer isa display instead of a speaker, and the localoscillator is tuned electronically rather thanby a front-panel knob.
The output of a spectrum analyzer is an X-Y
trace on a display, so lets see what informa-tion we get from it. The display is mappedon a grid (graticule) with 10 major horizontaldivisions and generally 10 major verticaldivisions. The horizontal axis is linearlycalibrated in frequency that increases fromleft to right. Setting the frequency is a two-step process. First we adjust the frequencyat the centerline of the graticule with thecenter frequency control. Then we adjustthe frequency range (span) across the full10 divisions with the frequency span control.These controls are independent, so if wechange the center frequency, we do not alter
the frequency span. Alternatively, we canset the start and stop frequencies insteadof setting center frequency and span. Ineither case, we can determine the absolutefrequency of any signal displayed and therelative frequency difference between anytwo signals.
Input
signal
RF inputattenuator
Pre-selector, orlow-pass filter
Mixer IF gain
Localoscillator
Referenceoscillator
Sweepgenerator Display
IF filterLogamp
Envelopedetector
Videofilter
Figure 2-1. Block diagram of a classic superheterodyne spectrum analyzer
Chapter 2. Spectrum Analyzer Fundamentals
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
10/89
10
of an attenuator circuit with a maximumattenuation of 70 dB in increments of 2 dB.The blocking capacitor is used to preventthe analyzer from being damaged by a DCsignal or a DC offset of the signal beingviewed. Unfortunately, it also attenuateslow-frequency signals and increases theminimum useable start frequency of the ana-lyzer to 9 kHz, 100 kHz or 10 MHz, dependingon the analyzer.
In some analyzers, an amplitude referencesignal can be connected as shown inFigure 2-3. It provides a precise frequencyand amplitude signal, used by the analyzerto periodically self-calibrate.
The vertical axis is calibrated in amplitude.You can choose a linear scale calibratedin volts or a logarithmic scale calibratedin dB. The log scale is used far more oftenthan the linear scale because it has amuch wider usable range. The log scaleallows signals as far apart in amplitudeas 70 to 100 dB (voltage ratios of 3200 to100,000 and power ratios of 10,000,000 to10,000,000,000) to be displayed simultane-ously. On the other hand, the linear scale isusable for signals differing by no more than20 to 30 dB (voltage ratios of 10 to 32). Ineither case, we give the top line of the grati-cule, the reference level, an absolute valuethrough calibration techniques1and use thescaling per division to assign values to otherlocations on the graticule. Therefore, wecan measure either the absolute value of asignal or the relative amplitude differencebetween any two signals.
Scale calibration, both frequency and ampli-tude, is shown by annotations written ontothe display. Figure 2-2 shows the display of atypical analyzer.
Now, lets turn our attention back to thespectrum analyzer components diagramed inFigure 2-1.
RF attenuator
The first part of our analyzer is the RF inputattenuator. Its purpose is to ensure thesignal enters the mixer at the optimum level
to prevent overload, gain compression anddistortion. Because attenuation is a protec-tive circuit for the analyzer, it is usually setautomatically, based on the reference level.However, manual selection of attenuation isalso available in steps of 10, 5, 2, or even 1dB. The diagram in Figure 2-3 is an example
1. See Chapter 4, Amplitude and Frequency Accuracy.
Figure 2-2. Typical spectrum analyzer display with control settings
Figure 2-3. RF input attenuator circuitry
Low-pass filter or preselector
The low-pass filter blocks high-frequencysignals from reaching the mixer. Thisfiltering prevents out-of-band signals frommixing with the local oscillator and creat-ing unwanted responses on the display.Microwave spectrum analyzers replace the
low-pass filter with a preselector, which isa tunable filter that rejects all frequenciesexcept those we currently wish to view. InChapter 7, we go into more detail about theoperation and purpose of the preselector.
RF input
Amplitudereferencesignal
0 to 70 dB, 2 dB steps
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
11/89
11
mixing product will fall in the IF pass bandat some point on the ramp (sweep), and wewill see a response on the display.
The ramp generator controls both the hori-zontal position of the trace on the displayand the LO frequency, so we can nowcalibrate the horizontal axis of the display interms of the input signal frequency.
We are not quite through with the tuningyet. What happens if the frequency of theinput signal is 9.0 GHz? As the LO tunesthrough its 3.8- to 8.7-GHz range, it reachesa frequency (3.9 GHz) at which it is the IFaway from the 9.0-GHz input signal. At thisfrequency we have a mixing product that isequal to the IF, creating a response on thedisplay. In other words, the tuning equationcould just as easily have been:
fsig= fLO+ fIF
This equation says that the architecture ofFigure 2-1 could also result in a tuning rangefrom 8.9 to 13.8 GHz, but only if we allowsignals in that range to reach the mixer. Thejob of the input low-pass filter in Figure 2-1is to prevent these higher frequencies fromgetting to the mixer. We also want to keepsignals at the intermediate frequency itselffrom reaching the mixer, as previouslydescribed, so the low-pass filter must do agood job of attenuating signals at 5.1 GHz aswell as in the range from 8.9 to 13.8 GHz.
Tuning the analyzer
We need to know how to tune our spectrumanalyzer to the desired frequency range.Tuning is a function of the center frequencyof the IF filter, the frequency range of theLO and the range of frequencies allowedto reach the mixer from the outside world
(allowed to pass through the low-pass filter).Of all the mixing products emerging from themixer, the two with the greatest amplitudes,and therefore the most desirable, are thosecreated from the sum of the LO and inputsignal and from the difference between theLO and input signal. If we can arrange thingsso that the signal we wish to examine iseither above or below the LO frequency bythe IF, then only one of the desired mixingproducts will fall within the pass-band ofthe IF filter and be detected to create anamplitude response on the display.
We need to pick an LO frequency and anIF that will create an analyzer with thedesired tuning range. Lets assume thatwe want a tuning range from 0 to 3.6 GHz.We then need to choose the IF. Lets try a1-GHz IF. Because this frequency is withinour desired tuning range, we could have aninput signal at 1 GHz. The output of a mixeralso includes the original input signals, soan input signal at 1 GHz would give us aconstant output from the mixer at the IF. The1-GHz signal would thus pass through thesystem and give us a constant amplituderesponse on the display regardless of the
tuning of the LO. The result would be a holein the frequency range at which we couldnot properly examine signals because theamplitude response would be independentof the LO frequency. Therefore, a 1-GHz IFwill not work.
Instead, we choose an IF that is above thehighest frequency to which we wish to tune.In the Agilent X-Series signal analyzers thatcan tune to 3.6 GHz, the first LO frequencyrange is 3.8 to 8.7 GHz, and the IF chosenis about 5.1 GHz. Remember that we wantto tune from 0 Hz to 3.6 GHz (actually from
some low frequency because we cannotview a 0-Hz signal with this architecture).
If we start the LO at the IF (LO minus IF= 0 Hz) and tune it upward from there to3.6 GHz above the IF, we can cover thetuning range with the LO minus IF mixingproduct. Using this information, we cangenerate a tuning equation:
fsig= fLO- fIF
where fsig= signal frequency fLO = local oscillator frequency, and fIF = intermediate frequency (IF)
If we wanted to determine the LO frequen-cy needed to tune the analyzer to a low-,mid-, or high-frequency signal (say, 1 kHz,1.5 GHz, or 3 GHz), we would first restatethe tuning equation in terms of fLO:
fLO= fsig+ fIF
Then we would apply the numbers for thesignal and IF in the tuning equation2:
fLO= 1 kHz + 5.1 GHz = 5.100001 GHz
fLO= 1.5 GHz + 5.1 GHz = 6.6 GHz or
fLO= 3 GHz + 5.1 GHz = 8.1 GHz.
Figure 2-4 illustrates analyzer tuning. Inthis figure, fLO is not quite high enough tocause the fLO fsig mixing product to fall inthe IF pass band, so there is no responseon the display. If we adjust the ramp gen-erator to tune the LO higher, however, this
2. In the text, we round off some of the frequency values for simplicity, although the exact values are shown in the figures.
Freq rangeof analyzer
Freq range of LO
A
A
fsig
fLO
fLO fsig fLO+ fsigfLO
IF
f
ff
Freq rangeof analyzer
Figure 2-4. The LO must be tuned to fIF+ f
sigto produce a response on the display
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
12/89
12
In summary, we can say that for a single-band RF spectrum analyzer, we wouldchoose an IF above the highest frequencyof the tuning range. We would make the LOtunable from the IF to the IF plus the upperlimit of the tuning range and include a low-pass filter in front of the mixer that cuts offbelow the IF.
To separate closely spaced signals (seeResolving signals later in this chapter),some spectrum analyzers have IF band-widths as narrow as 1 kHz; others, 10 Hz;still others, 1 Hz. Such narrow filters aredifficult to achieve at a center frequency of5.1 GHz, so we must add additional mixingstages, typically two to four stages, todown-convert from the first to the final IF.Figure 2-5 shows a possible IF chain basedon the architecture of a typical spectrumanalyzer. The full tuning equation for thisanalyzer is:
fsig= fLO1 (fLO2+ fLO3+ ffinal IF)
However,
fLO2+ fLO3+ ffinal IF
= 4.8 GHz + 300 MHz + 22.5 MHz
= 5.1225 GHz, the first IF.
Simplifying the tuning equation by usingjust the first IF leads us to the sameanswers. Although only passive filtersare shown in figure 2-5, the actualimplementation includes amplification inthe narrower IF stages. The final IF sectioncontains additional components, such aslogarithmic amplifiers or analog -to-digitalconverters, depending on the design of theparticular analyzer.
Most RF spectrum analyzers allow an LOfrequency as low as, and even below, thefirst IF. Because there is finite isolationbetween the LO and IF ports of the mixer,the LO appears at the mixer output. Whenthe LO equals the IF, the LO signal itself isprocessed by the system and appears asa response on the display, as if it were aninput signal at 0 Hz. This response, the LOfeedthrough, can mask very low-frequencysignals, so not all analyzers allow the displayrange to include 0 Hz.
IF gain
Referring back to Figure 2-1, we see thenext component of the block diagram is avariable gain amplifier. It is used to adjustthe vertical position of signals on the dis-play without affecting the signal level at theinput mixer. When the IF gain is changed,
the value of the reference level is changedaccordingly to retain the correct indicatedvalue for the displayed signals. Generally,we do not want the reference level tochange when we change the input attenu-ator, so the settings of the input attenuatorand the IF gain are coupled together.
A change in input attenuation will auto-matically change the IF gain to offset theeffect of the change in input attenuation,thereby keeping the signal at a constantposition on the display.
3.6 GHz
3.8 to 8.7 GHz
Sweepgenerator
Display
5.1225 GHz Envelopedetector
322.5 MHz
300 MHz
22.5 MHz
4.8 GHz
Figure 2-5. Most spectrum analyzers use two to four mixing steps to reach the final IF.
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
13/89
13
Resolving signals
After the IF gain amplifier, we find the IFsection, which consists of the analog ordigital resolution bandwidth (RBW) filters,or both.
Analog filters
Frequency resolution is the ability of aspectrum analyzer to separate two inputsinusoids into distinct responses. Fouriertells us that a sine-wave signal only hasenergy at one frequency, so we should nothave any resolution problems. Two signals,no matter how close in frequency, shouldappear as two lines on the display. But acloser look at our superheterodyne receivershows why signal responses have a definite
width on the display. The output of a mixerincludes the sum and difference productsplus the two original signals (input and LO).A bandpass filter determines the intermedi-ate frequency, and this filter selects thedesired mixing product and rejects allother signals. Because the input signal isfixed and the local oscillator is swept, theproducts from the mixer are also swept. If amixing product happens to sweep past theIF, the characteristic shape of the bandpassfilter is traced on the display. See Figure 2-6.The narrowest filter in the chain determinesthe overall displayed bandwidth, and in thearchitecture of Figure 2-5, this filter is in the22.5- MHz IF.
Two signals must be far enough apart orthe traces they make will fall on top ofeach other and look like only one response.
Fortunately, spectrum analyzers have select-able resolution (IF) filters, so it is usuallypossible to select one narrow enough toresolve closely spaced signals.
Agilent data sheets describe the abilityto resolve signals by listing the 3-dBbandwidths of the available IF filters. Thisnumber tells us how close together equal-amplitude sinusoids can be and still beresolved. In this case, there will be about a3-dB dip between the two peaks traced outby these signals. See Figure 2-7. The signalscan be closer together before their tracesmerge completely, but the 3-dB bandwidth isa good rule of thumb for resolution of equal-amplitude signals3.
Figure 2-6. As a mixing product sweeps past the IF filter, the filter shape is traced on the display
Figure 2-7. Two equal-amplitude sinusoids separated by the 3-dB BW of the selected IF filter can
be resolved.
3. If you experiment with resolution on a spectrum
analyzer using the normal (rosenfell) detector mode (See
Detector types later in this chapter) use enough video
filtering to create a smooth trace. Otherwise, you will see
smearing as the two signals interact. While the smeared
trace certainly indicates the presence of more than one
signal, it is difficult to determine the amplitudes of the
individual signals. Spectrum analyzers with positive peak
as their default detector mode may not show the smear-
ing effect. You can observe the smearing by selecting the
sample detector mode.
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
14/89
14
More often than not, we are dealing withsinusoids that are not equal in amplitude.The smaller sinusoid can actually be lostunder the skirt of the response traced outby the larger. This effect is illustrated inFigure 2-8. The top trace looks like a singlesignal, but in fact represents two signals:one at 300 MHz (0 dBm) and another at300.005 MHz (30 dBm). The lower traceshows the display after the 300-MHz signalis removed.
Another specification is listed for theresolution filters: bandwidth selectivity(or selectivity or shape factor). Bandwidthselectivity helps determine the resolvingpower for unequal sinusoids. For Agilentanalyzers, bandwidth selectivity is generallyspecified as the ratio of the 60-dB bandwidthto the 3-dB bandwidth, as shown in Figure2-9. The analog filters in Agilent analyzersare a four-pole, synchronously tuned design,with a nearly Gaussian shape4. This typeof filter exhibits a bandwidth selectivity ofabout 12.7:1.
Figure 2-8. A low-level signal can be lost under the skirt of the response to a larger signal
4. Some older spectrum analyzer models used five-pole filters for the narrowest resolution bandwidths to provide improved selectivity of about 10:1. Modern designs achieve even better bandwidth
selectivity using digital IF filters.
3 dB
60 dB
3 dB
60 d
Figure 2-9. Bandwidth selectivity, ratio of 60-dB to 3-dB bandwidths
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
15/89
15
Residual FM
The instability and residual FM of the LOsin an analyzer, particularly the first LO, oftendetermine the minimum usable resolutionbandwidth. The unstable YIG (yttrium irongarnet) oscillator used in early analyzerstypically had a residual FM of about 1 kHz.Because this instability was transferredto any mixing product involving the LO,
there was no point in having resolutionbandwidths narrower than 1 kHz because itwas impossible to determine the cause ofany instability on the display.
However, modern analyzers have dramati-cally improved residual FM. For example,residual FM in Agilent PXA Series analyzersis nominally 0.25 Hz; in PSA Series analyz-ers, 1 to 4 Hz; and in ESA Series analyzers,2 to 8 Hz. This allows bandwidths as low as1 Hz in many analyzers, and any instabilitywe see on a spectrum analyzer today is dueto the incoming signal.
For example, what resolution bandwidthmust we choose to resolve signals thatdiffer by 4 kHz and 30 dB, assuming 12.7:1bandwidth selectivity? Because we areconcerned with rejection of the larger signalwhen the analyzer is tuned to the smallersignal, we need to consider not the fullbandwidth, but the frequency differencefrom the filter center frequency to the skirt.To determine how far down the filter skirtis at a given offset, we use the followingequation:
H(f) = 10(N) log10 [(f/f0)2 + 1]
WhereH(f) is the filter skirt rejection in dB,N is the number of filter poles,f is the frequency offset from the centerin Hz, and
f0is given by
For our example, N=4 and f = 4000. Letsbegin by trying the 3-kHz RBW filter.First, we compute f0:
f0=3000
2 21= 3448.44
Now we can determine the filter rejectionat a 4-kHz offset:
H(4000) = 10(4) log10 [(4000/3448.44)2 + 1]
= 14.8 dB
This is not enough to allow us to see thesmaller signal. Lets determine H(f)again using a 1-kHz filter:
f0=1000
2 21 = 1149.48
This allows us to calculate the filter rejection:
H(4000) = 10(4) log10[(4000/1149.48)2 + 1]= 44.7 dB
Thus, the 1-kHz resolution bandwidthfilter does resolve the smaller signal, asillustrated in Figure 2-10.
Digital filters
Some spectrum analyzers use digital tech-niques to realize their resolution bandwidthfilters. Digital filters can provide importantbenefits, such as dramatically improvedbandwidth selectivity. The Agilent PSA andX-Series signal analyzers implement allresolution bandwidths digitally. Otheranalyzers, such as the Agilent ESA-E Series,take a hybrid approach, using analog filtersfor the wider bandwidths and digital filtersfor bandwidths of 300 Hz and below.Refer to Chapter 3 for more information ondigital filters.
RBW
2 21/N1Figure 2-10. The 3-kHz filter (top trace) does not resolve the smaller signal; reducing the resolutionbandwidth to 1 kHz (bottom trace) does
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
16/89
16
Phase noise
No oscillator is perfectly stable. Eventhough we may not be able to see theactual frequency jitter of a spectrum analyz-er LO system, there is still a manifestationof the LO frequency or phase instability thatcan be observed. This is known as phase
noise (sometimes called sideband noise).All are frequency or phase modulated byrandom noise to some extent. As previouslynoted, any instability in the LO is transferredto any mixing products resulting from theLO and input signals. So the LO phasenoise modulation sidebands appear aroundany spectral component on the displaythat is far enough above the broadbandnoise floor of the system (Figure 2-11). Theamplitude difference between a displayedspectral component and the phase noiseis a function of the stability of the LO. The
more stable the LO, the lower the phasenoise. The amplitude difference is also afunction of the resolution bandwidth. If wereduce the resolution bandwidth by a factorof 10, the level of the displayed phase noisedecreases by 10 dB5.
The shape of the phase noise spectrum isa function of analyzer design, in particular,the sophistication of the phase-lock loopsemployed to stabilize the LO. In someanalyzers, the phase noise is a relatively flatpedestal out to the bandwidth of the stabiliz-ing loop. In others, the phase noise may fall
away as a function of frequency offset fromthe signal. Phase noise is specified in termsof dBc (dB relative to a carrier) and normal-ized to a 1-Hz noise power bandwidth. It issometimes specified at specific frequencyoffsets. At other times, a curve is given toshow the phase noise characteristics over arange of offsets.
Generally, we can see the inherent phasenoise of a spectrum analyzer only in thenarrower resolution filters, when it obscuresthe lower skirts of these filters. The use ofthe digital filters previously described does
not change this effect. For wider filters, thephase noise is hidden under the filter skirt,just as in the case of two unequal sinusoidsdiscussed earlier.
Todays spectrum or signal analyzers, suchas Agilents X-Series, allow you to selectdifferent LO stabilization modes to optimizethe phase noise for different measurementconditions. For example, the PXA signalanalyzer offers three different modes:
Optimize phase noise for frequency
offsets < 140 kHz from the carrier
In this mode, the LO phase noise is opti-
mized for the area close in to the carrierat the expense of phase noise beyond140-kHz offset.
Optimize phase noise for frequency
offsets > 160 kHz from the carrier
This mode optimizes phase noise foroffsets above 160 kHz away fromthe carrier.
Optimize LO for fast tuning
When this mode is selected, LO behaviorcompromises phase noise at all offsetsfrom the carrier below approximately
2 MHz. This mode minimizes measure-ment time and allows the maximummeasurement throughput when changingthe center frequency or span.
The PXA signal analyzers phase noiseoptimization can also be set to auto mode,which automatically sets the instrumentsbehavior to optimize speed or dynamicrange for various operating conditions.When the span is > 44.44 MHz or the RBWis > 1.9 MHz, or the source mode is set toTracking, the PXA selects Fast Tuningmode. Otherwise, the PXA automaticallychooses Best Close-In Phase Noise when
center frequency < 195 kHz, or when centerfrequency 1 MHz and span 1.3 MHz andRBW 75 kHz. If these conditions are notmet, the PXA automatically chooses BestWide-Offset Phase Noise. These rules applywhen you use swept spans, zero span orFFT spans.
In any case, phase noise becomes theultimate limitation in an analyzers abilityto resolve signals of unequal amplitude.As shown in Figure 2-13, we may havedetermined that we can resolve two signalsbased on the 3-dB bandwidth and selectivity,
only to find that the phase noise covers upthe smaller signal.
5. The effect is the same for the broadband noise floor (or any broadband noise signal). See Chapter 5, Sensitivity and Noise.
Figure 2-11. Phase noise is displayed only when a signal is displayed far enough above the system noise floor
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
17/89
17
Figure 2-12a. Phase noise performance can be optimized for different
measurement conditions
Figure 2-12b. Detail of the 140-kHz carrier offset region
Figure 2-13. Phase noise can prevent resolution of unequal signals
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
18/89
18
Sweep time
Analog resolution filters
If resolution were the only criterion on whichwe judged a spectrum analyzer, we mightdesign our analyzer with the narrowest pos-sible resolution (IF) filter and let it go at that.But resolution affects sweep time, and we
care very much about sweep time. Sweeptime directly affects how long it takes tocomplete a measurement.
Resolution comes into play because the IFfilters are band-limited circuits that requirefinite times to charge and discharge. If themixing products are swept through themtoo quickly, there will be a loss of displayedamplitude, as shown in Figure 2-14. (SeeEnvelope detector, later in this chapter, foranother approach to IF response time.) If wethink about how long a mixing product staysin the pass band of the IF filter, that timeis directly proportional to bandwidth andinversely proportional to the sweep in Hz perunit time, or:
Time in pass band = RBWSpan/ST
= (RBW)(ST)Span
WhereRBW = resolution bandwidth andST = sweep time.
On the other hand, the rise time of a filteris inversely proportional to its bandwidth,and if we include a constant of proportion-
ality, k, then:Rise time = k
RBW
If we make the terms equal and solve forsweep time, we have:
k = (RBW)(ST)Span
or ST = k (Span)RBW2
For the synchronously-tuned, near-Gaussianfilters used in many analog analyzers, thevalue of k is in the 2 to 3 range .
The important message here is that achange in resolution has a dramatic effecton sweep time. Older analog analyzers typi-cally provided values in a 1, 3, 10 sequenceor in ratios roughly equaling the square rootof 10. So sweep time was affected by afactor of about 10 with each step in resolu-tion. Agilent X-Series signal analyzers offerbandwidth steps of just 10% for an evenbetter compromise among span, resolution
and sweep time.
Spectrum analyzers automatically couplesweep time to the span and resolutionbandwidth settings. Sweep time is adjustedto maintain a calibrated display. If the needarises, we can override the automaticsetting and set sweep time manually.If you set a sweep time longer than themaximum available, the analyzer indicatesthat the display is uncalibrated with aMeas Uncal message in the upper-rightpart of the graticule.
Figure 2-14. Sweeping an analyzer too fast causes a drop in displayed amplitude and a shift inindicated frequency
RBW
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
19/89
19
Digital resolution filters
The digital resolution filters used in Agilentspectrum analyzers have an effect on sweeptime that is different from the effects wevejust discussed for analog filters. For sweptanalysis, the speed of digitally implementedfilters, with no further processing, can showa two to four times improvement.
However, the X-Series signal analyzers withOption FS1 are programmed to correct forthe effect of sweeping too fast for resolu-tion bandwidths between about 3 kHz and100 kHz. As a result, sweep times thatwould otherwise be many seconds may bereduced to milliseconds, depending uponthe particular settings. See Figure 2-14a. Thesweep time without the correction would be79.8 seconds. Figure 2-14b shows a sweeptime of 1.506 s with Option FS1 installed.For the widest resolution bandwidths, sweeptimes are already very short. For example,
using the formula with k = 2 on a span of1 GHz and a RBW of 1 MHz, the sweep timecalculates to just 2 msec.
For narrower resolution bandwidths, analyz-ers such as the Agilent X-Series use fastFourier transforms (FFTs) to process thedata, also producing shorter sweep timesthan the formula predicts. The differenceoccurs because the signal being analyzed isprocessed in frequency blocks, dependingupon the particular analyzer. For example, ifthe frequency block was 1 kHz, then whenwe select a 10-Hz resolution bandwidth, the
analyzer is in effect simultaneously process-ing the data in each 1-kHz block through100 contiguous 10-Hz filters. If the digitalprocessing were instantaneous, we wouldexpect sweep time to be reduced by a factorof 100. In practice, the reduction factor isless, but is still significant. For more informa-tion on the advantages of digital processing,refer to Chapter 3.
Figure 2-14a. Full span sweep speed, RBW of 20 kHz, without Option FS1
Figure 2-14b. Full span sweep speed, RBW of 20 kHz, with Option FS1
More information
A more detailed discussionabout fast sweep measure-ments can be found in UsingFast-Sweep Techniques to
Accelerate Spur Searches
Application Note, literaturenumber 5991-3739EN
http://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdfhttp://cp.literature.agilent.com/litweb/pdf/5991-3739EN.pdf -
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
20/89
20
However, there are times when we deliber-ately choose a resolution bandwidth wideenough to include two or more spectralcomponents. At other times, we have nochoice. The spectral components are closerin frequency than our narrowest bandwidth.Assuming only two spectral componentswithin the pass band, we have two sinewaves interacting to create a beat note,and the envelope of the IF signal varies, asshown in Figure 2-16, as the phase betweenthe two sine waves varies.
Envelope detector6
Older analyzers typically converted the IFsignal to video with an envelope detector7.In its simplest form, an envelope detectorconsists of a diode, resistive load andlow-pass filter, as shown in Figure 2-15. Theoutput of the IF chain in this example, an
amplitude modulated sine wave, is appliedto the detector. The response of the detectorfollows the changes in the envelope of theIF signal, but not the instantaneous value ofthe IF sine wave itself.
For most measurements, we choose aresolution bandwidth narrow enough toresolve the individual spectral componentsof the input signal. If we fix the frequency ofthe LO so that our analyzer is tuned to oneof the spectral components of the signal, theoutput of the IF is a steady sine wave witha constant peak value. The output of the
envelope detector will then be a constant(DC) voltage, and there is no variation for thedetector to follow.
The width of the resolution (IF) filterdetermines the maximum rate at whichthe envelope of the IF signal can change.This bandwidth determines how far aparttwo input sinusoids can be so that afterthe mixing process they will both be withinthe filter at the same time. Lets assume a
22.5-MHz final IF and a 100-kHz bandwidth.Two input signals separated by 100 kHzwould produce mixing products of 22.45 and22.55 MHz and would meet the criterion.See Figure 2-16. The detector must beable to follow the changes in the envelopecreated by these two signals but not the22.5-MHz IF signal itself.
The envelope detector is what makesthe spectrum analyzer a voltmeter. Letsduplicate the situation above and have twoequal-amplitude signals in the pass bandof the IF at the same time. A power meter
would indicate a power level 3 dB aboveeither signal, that is, the total power of thetwo. Assume that the two signals are closeenough so that, with the analyzer tunedhalf-way between them, there is negligible
attenuation due to the roll-off of the filter 8.The analyzer display will vary between avalue that is twice the voltage of either (6 dBgreater) and zero (minus infinity on the logscale). We must remember that the twosignals are sine waves (vectors) at differentfrequencies, and so they continually change
in phase with respect to each other. At sometime they add exactly in phase; at another,exactly out of phase.
So the envelope detector follows thechanging amplitude values of the peaksof the signal from the IF chain but not theinstantaneous values, resulting in the lossof phase information. This gives the analyzerits voltmeter characteristics.
Digitally implemented resolution bandwidthsdo not have an analog envelope detector.Instead, the digital processing computes the
root sum of the squares of the I and Q data,which is mathematically equivalent to anenvelope detector. For more information ondigital architecture, refer to Chapter 3.
More information
Additional information on enve-lope detectors can be found inSpectrum and Signal Analyzer
Measurements and NoiseApplication Note, literaturenumber 5966-4008E.
Figure 2-15. Envelope detector
IF signal
t t
6. The envelope detector should not be confused with the display detectors. See Detector types later in this chapter.
7. A signal whose frequency range extends from zero (DC) to some upper frequency determined by the circuit elements. Historically, spectrum analyzers with analog displays used this signal to drive the
vertical deflection plates of the CRT directly. Hence it was known as the video signal.
8. For this discussion, we assume the filter is perfectly rectangular.
Figure 2-16. Output of the envelope detector follows the peaks of the IF signal
http://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdf -
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
21/89
21
Displays
Up until the mid-1970s, spectrum analyzerswere purely analog. The displayed trace pre-sented a continuous indication of the signalenvelope, and no information was lost.However, analog displays had drawbacks.The major problem was in handling the
long sweep times required for narrow reso-lution bandwidths. In the extreme case, thedisplay became a spot that moved slowlyacross the cathode ray tube (CRT), with noreal trace on the display. So a meaningfuldisplay was not possible with the longersweep times.
Figure 2-17. When digitizing an analog signal, what value should be displayed at each point?
Agilent Technologies (part of Hewlett-Packard at the time) pioneered a variable-persistence storage CRT in which we couldadjust the fade rate of the display. Whenproperly adjusted, the old trace would justfade out at the point where the new tracewas updating the display. This displaywas continuous, had no f licker and avoidedconfusing overwrites. It worked quite well,but the intensity and the fade rate had tobe readjusted for each new measurementsituation. When digital circuitry becameaffordable in the mid-1970s, it was quickly
put to use in spectrum analyzers. Oncea trace had been digitized and put intomemory, it was permanently available fordisplay. It became an easy matter to updatethe display at a flicker-free rate withoutblooming or fading. The data in memorywas updated at the sweep rate, and sincethe contents of memory were written to thedisplay at a flicker-free rate, we could followthe updating as the analyzer swept throughits selected frequency span just as we couldwith analog systems.
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
22/89
22
The first three detectors, sample, peak,
and negative peak are easy to understandand are visually represented in Figure 2-19.Normal, average, and quasipeak are morecomplex and will be discussed later.
Lets return to the question of how todisplay an analog system as faithfully aspossible using digital techniques. Letsimagine the situation illustrated inFigure 2-17. We have a display that containsonly noise and a single CW signal.
Detector types
With digital displays, we had to decide whatvalue should be displayed for each displaydata point. No matter how many data pointswe use across the display, each pointmust represent what has occurred oversome frequency range and, although we
usually do not think in terms of time whendealing with a spectrum analyzer, over sometime interval.
It is as if the data for each interval is throwninto a bucket and we apply whatever math isnecessary to extract the desired bit of infor-mation from our input signal. This datum isput into memory and written to the display.This process provides great flexibility.
Here we will discuss six differentdetector types.
In Figure 2-18, each bucket contains datafrom a span and timeframe that is deter-mined by these equations:
Frequency:bucket width = span/(trace points 1)Time:bucket width = sweep time/(trace points 1)
The sampling rates are different for vari-ous instruments, but greater accuracy isobtained from decreasing the span orincreasing the sweep time because thenumber of samples per bucket will increase
in either case. Even in analyzers withdigital IFs, sample rates and interpolationbehaviors are designed to be the equivalentof continuous-time processing.
The bucket concept is important, as it willhelp us differentiate the six detector types:
Sample
Positive peak (also simply called peak)
Negative peak
Normal
Average
Quasipeak
Positive peak
Sample
Negative peak
One bucket
Figure 2-19. The trace point saved in memory is based on the detector type algorithm
Figure 2-18. Each of the 1001 trace points (buckets) covers a 100-kHz frequency span and a
0.01-millisecond time span
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
23/89
23
Sample detection
As a first method, let us simply select thedata point as the instantaneous level at thecenter of each bucket (see Figure 2-19).This is the sample detection mode. To givethe trace a continuous look, we design asystem that draws vectors between the
points. Comparing Figure 2-17 with 2-20,it appears that we get a fairly reasonabledisplay. Of course, the more points thereare in the trace, the better the replicationof the analog signal will be. The number ofavailable display points can vary for differentanalyzers. On X-Series signal analyzers,the number of display points for frequencydomain traces can be set from a minimumof 1 point to a maximum of 40,001 points. Asshown in Figure 2-21, more points do indeedget us closer to the analog signal.
While the sample detection mode does a
good job of indicating the randomness ofnoise, it is not a good mode for analyzingsinusoidal signals. If we were to look at a100-MHz comb on an Agilent PXA, we mightset it to span from 0 to 26.5 GHz. Even with1,001 display points, each display pointrepresents a span (bucket) of 26.5 MHz.This is far wider than the maximum 8-MHzresolution bandwidth.
As a result, the true amplitude of a combtooth is shown only if its mixing producthappens to fall at the center of the IF whenthe sample is taken. Figure 2-22a shows a
10-MHz span with a 750-Hz bandwidth usingsample detection. The comb teeth shouldbe relatively equal in amplitude, as shownin Figure 2-22b (usingpeak detection).Therefore, sample detection does not catchall the signals, nor does it necessarily reflectthe true peak values of the displayed signals.When resolution bandwidth is narrowerthan the sample interval (the bucket width),sample mode can give erroneous results.
Figure 2-21. More points produce a display closerto an analog display
Figure 2-20. Sample display mode using 10 pointsto display the signal shown in Figure 2-17
Figure 2-22a. A 10-MHz span of a 250-kHz comb in the sample display mode
Figure 2-22b. The actual comb over a 10-MHz span using peak (positive) detection
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
24/89
24
Peak (positive) detection
One way to insure that all sinusoids arereported at their true amplitudes is to displaythe maximum value encountered in eachbucket. This is the positive peak detectionmode, or peak. This mode is illustratedin Figure 2-22b. Peak is the default mode
offered on many spectrum analyzersbecause it ensures that no sinusoid ismissed, regardless of the ratio betweenresolution bandwidth and bucket width.However, unlike sample mode, peak doesnot give a good representation of randomnoise because it only displays the maximumvalue in each bucket and ignores the truerandomness of the noise. So spectrumanalyzers that use peak detection as theirprimary mode generally also offer samplemode as an alternative.
Negative peak detection
Negative peak detection displays theminimum value encountered in each bucket.It is generally available in most spectrumanalyzers, though it is not used as often asother types of detection. Differentiating CWfrom impulsive signals in EMC testing is one
application where negative peak detectionis valuable. Later in this application note,we will see how negative peak detection isalso used in signal identification routineswhen you use external mixers for high-frequency measurements.
Normal detection
To provide a better visual display of randomnoise than offered bypeak mode and yetavoid the missed-signal problem of thesample mode, the normal detection mode(informally known as rosenfell9mode) isoffered on many spectrum analyzers. Should
the signal both rise and fall, as determinedby the positive peak and negative peakdetectors, the algorithm classifies the signalas noise.
In that case, an odd-numbered data pointdisplays the maximum value encounteredduring its bucket. And an even-numbereddata point displays the minimum valueencountered during its bucket. See Figure2-25. Normal and sample modes are com-pared in Figures 2-23a and 2-23b.10
Figure 2-23a. Normal mode
9. Rosenfell is not a persons name but rather a description of the algorithm that tests to see if the signal rose and fell within the bucket represented by a given data point. It is also sometimes written as
rosenfell.
10. Because of its usefulness in measuring noise, the sample detector is usually used in noise marker applications. Similarly, the measurement of channel power and adjacent-channel power requires a
detector type that gives results unbiased bypeak detection. For analyzers without averaging detectors, sample detection is the best choice.
Figure 2-23b. Sample mode
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
25/89
25
What happens when a sinusoidal signalis encountered? We know that as a mix-ing product is swept past the IF filter, ananalyzer traces out the shape of the filteron the display. If the filter shape is spreadover many display points, we encounter asituation in which the displayed signal onlyrises as the mixing product approaches thecenter frequency of the filter and only fallsas the mixing product moves away from thefilter center frequency. In either of thesecases, the positive-peak and negative-peakdetectors sense an amplitude change inonly one direction, and, according to thenormal detection algorithm, the maximumvalue in each bucket is displayed. SeeFigure 2-24.
What happens when the resolutionbandwidth is narrow, relative to a bucket?The signal will both rise and fall during thebucket. If the bucket happens to be an odd-numbered one, all is well. The maximumvalue encountered in the bucket is simplyplotted as the next data point. However,if the bucket is even-numbered, then theminimum value in the bucket is plotted.Depending on the ratio of resolution band-width to bucket width, the minimum valuecan differ from the true peak value (the onewe want displayed) by a little or a lot. Inthe extreme, when the bucket is much widerthan the resolution bandwidth, the differencebetween the maximum and minimum valuesencountered in the bucket is the full dif-ference between the peak signal value and
the noise. This is true for the example inFigure 2-25. See bucket 6. The peak valueof the previous bucket is always comparedto that of the current bucket. The greaterof the two values is displayed if the bucketnumber is odd, as depicted in bucket 7. Thesignal peak actually occurs in bucket 6 butis not displayed until bucket 7.
Figure 2-24. Normal detection displays maximum values in buckets where the signal only rises or only falls
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
26/89
26
Figure 2-25. Trace points selected by the normal detection algorithm
Figure 2-26. Normal detection can show two peaks when only one peak actually exists
The normaldetection algorithm:
If the signal rises and falls within a bucket:Even-numbered buckets display theminimum (negative peak) value in thebucket. The maximum is remembered. Odd-numbered buckets display the maxi(positivepeak) value determined by comparing thecurrent bucket peak with the previous
(remembered) bucket peak. If the signal onlyrises or only falls within a bucket, the peakis displayed. See Figure 2-25.
This process may cause a maximum valueto be displayed one data point too far to theright, but the offset is usually only a smallpercentage of the span. Some spectrumanalyzers, such as the Agilent PXA signalanalyzer, compensate for this potential effectby moving the LO start and stop frequencies.
Another type of error occurs when twopeaks are displayed when only one actually
exists. Figure 2-26 shows this error. Theoutline of the two peaks is displayed usingpeak detection with a wider RBW.
Sopeak detection is best for locating CWsignals well out of the noise. Sample is bestfor looking at noise, and normal is best forviewing signals and noise.
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
27/89
27
instruments takes an envelope-detectedsignal and passes it through a low-passfilter with a bandwidth much less thanthe RBW. The filter integrates (averages)the higher-frequency components such asnoise. To perform this type of detection in anolder spectrum analyzer that doesnt have abuilt-in voltage averaging detector function,set the analyzer in linear mode and select avideo filter with a cut-off frequency belowthe lowest PRF of the measured signal.
Quasipeak detectors (QPD) are also usedin EMI testing. QPD is a weighted formof peak detection. The measured valueof the QPD drops as the repetition rateof the measured signal decreases. Thus,an impulsive signal with a given peakamplitude and a 10-Hz pulse repetition ratewill have a lower quasipeak value than asignal with the same peak amplitude buthaving a 1-kHz repetition rate. This signalweighting is accomplished by circuitry withspecific charge, discharge and display timeconstants defined by CISPR13.
QPD is a way of measuring and quantifyingthe annoyance factor of a signal. Imaginelistening to a radio station suffering frominterference. If you hear an occasional popcaused by noise once every few seconds,you can still listen to the program withouttoo much trouble. However, if that sameamplitude pop occurs 60 times per second,it becomes extremely annoying, making theradio program intolerable to listen to.
Log-power (video)averagingaverages thelogarithmic amplitude values (dB) of theenvelope signal measured during the bucketinterval. Log power averaging is best forobserving sinusoidal signals, especiallythose near noise.11
Thus, using the average detector with theaveraging type set to power provides trueaverage power based upon rms voltage,while the average detector with the averag-ing type set to voltage acts as a general-purpose average detector. The averagedetector with the averaging type set to loghas no other equivalent.
Average detection is an improvement overusing sample detection for the determinationof power. Sample detection requires multiplesweeps to collect enough data points togive us accurate average power information.Average detection changes channel powermeasurements from being a summationover a range of buckets into integrationover the time interval representing a rangeof frequencies in a swept analyzer. In a fastFourier transfer (FFT) analyzer12, the summa-tion used for channel power measurementschanges from being a summation overdisplay buckets to being a summation overFFT bins. In both swept and FFT cases, theintegration captures all the power informa-tion available, rather than just that whichis sampled by the sample detector. As aresult, the average detector has a lowervariance result for the same measurement
time. In swept analysis, it also allows theconvenience of reducing variance simply byextending the sweep time.
EMI detectors: average andquasipeak detection
An important application of averagedetection is for characterizing devices forelectromagnetic interference (EMI). In thiscase, voltage averaging, as described inthe previous section, is used for measuringnarrowband signals that might be maskedby the presence of broadband impulsivenoise. The average detection used in EMI
Average detection
Although modern digital modulationschemes have noise-like characteristics,sampledetection does not always provide uswith the information we need. For instance,when taking a channel power measurementon a W-CDMA signal, integration of the
rms values is required. This measurementinvolves summing power across a range ofanalyzer frequency buckets. Sample detec-tion does not provide this capability.
While spectrum analyzers typically collectamplitude data many times in each bucket,sample detection keeps only one of thosevalues and throws away the rest. On theother hand, an averaging detector uses allthe data values collected within the time(and frequency) interval of a bucket. Once wehave digitized the data, and knowing the cir-cumstances under which they were digitized,
we can manipulate the data in a variety ofways to achieve the desired results.
Some spectrum analyzers refer to theaveraging detector as an rms detector whenit averages power (based on the root meansquare of voltage). Agilent X-Series signalanalyzers have an average detector thatcan average the power, voltage or log ofthe signal by including a separate control toselect the averaging type:
Power (rms) averagingaverages rms levelsby taking the square root of the sum of the
squares of the voltage data measured duringthe bucket interval, divided by the charac-teristic input impedance of the spectrumanalyzer, normally 50 ohms. Power averagingcalculates the true average power and is bestfor measuring the power of complex signals.
Voltage averagingaverages the linear volt-age data of the envelope signal measuredduring the bucket interval. It is often used inEMI testing for measuring narrowband sig-nals (this topic will be discussed further inthe next section). Voltage averaging is alsouseful for observing rise and fall behavior
of AM or pulse-modulated signals such asradar and TDMA transmitters.
11. See Chapter 5, Sensitivity and Noise.
12. Refer to Chapter 3 for more information on the FFT analyzers. They perform math computations on many buckets simultaneously, which improves measurement speed.
13. CISPR, the International Special Committee on Radio Interference, was established in 1934 by a group of international organizations to address radio interference. CISPR is a non-governmental group
composed of National Committees of the International Electrotechnical Commission (IEC), as well as numerous international organizations. CISPRs recommended standards generally form the basisfor statutory EMC requirements adopted by governmental regulatory agencies around the world.
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
28/89
28
Averaging processes
There are several processes in a spectrumanalyzer that smooth the variations in enve-lope-detected amplitude. The first method,average detection, was discussed previously.Two other methods, video filtering and traceaveraging, are discussed next.14
Video filtering
Discerning signals close to the noise is notjust a problem when performing EMC tests.Spectrum analyzers display signals plus theirown internal noise, as shown in Figure 2-27.To reduce the effect of noise on the displayedsignal amplitude, we often smooth or aver-age the display, as shown in Figure 2-28.Spectrum analyzers include a variable videofilter for this purpose. The video filter is alow-pass filter that comes after the envelopedetector and determines the bandwidth ofthe video signal that will later be digitized to
yield amplitude data. The cutoff frequency ofthe video filter can be reduced to the pointwhere it becomes smaller than the band-width of the selected resolution bandwidth(IF) filter. When this occurs, the video systemcan no longer follow the more rapid varia-tions of the envelope of the signal(s) passingthrough the IF chain. The result is an averag-ing or smoothing of the displayed signal.
The effect is most noticeable in measuringnoise, particularly when you use a wide-resolution bandwidth. As we reduce thevideo bandwidth, the peak-to-peak variationsof the noise are reduced. As Figure 2-29shows, the degree of reduction (degree ofaveraging or smoothing) is a function of theratio of the video to resolution bandwidths.At ratios of 0.01 or less, the smoothing isvery good. At higher ratios, the smoothing isnot as good. The video filter does not affectany part of the trace that is already smooth(for example, a sinusoid displayed well outof the noise).
Figure 2-27. Spectrum analyzers display signal plus noise
Figure 2-28. Display of Figure 2-27 after full smoothing
Figure 2-29. Smoothing effect of VBW-to-RBW ratios of 3:1, 1:10, and 1:100
14. A fourth method, called a noise marker, is discussed in
Chapter 5, Sensitivity and Noise.
More information
A more detailed discussionabout noise markers can befound in Spectrum and SignalAnalyzer Measurements and
Noise Application Note,literature number 5966-4008E
http://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdfhttp://cp.literature.agilent.com/litweb/pdf/5966-4008E.pdf -
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
29/89
29
If we set the analyzer topositive peakdetection mode, we notice two things:First, if VBW > RBW, then changing theresolution bandwidth does not make muchdifference in the peak-to-peak fluctuationsof the noise. Second, if VBW < RBW,changing the video bandwidth seems toaffect the noise level. The fluctuations donot change much because the analyzeris displaying only the peak values of thenoise. However, the noise level appears tochange with video bandwidth because theaveraging (smoothing) changes, therebychanging the peak values of the smoothednoise envelope. See Figure 2-30a. Whenwe select average detection, we see theaverage noise level remains constant. SeeFigure 2-30b.
Because the video filter has its ownresponse time, the sweep time increasesapproximately inversely with video band-width when the VBW is less than the reso-lution bandwidth. The sweep time (ST) cantherefore be described by this equation:
ST k(Span)
(RBW)(VBW)
The analyzer sets the sweep time automati-cally to account for video bandwidth as wellas span and resolution bandwidth.
Figure 2-30a. Positive peak detection mode: reducing video bandwidth lowers peak noise but not
average noise
Figure 2-30b. Average detection mode: noise level remains constant, regardless of VBW-to-RBW ratios
(3:1, 1:10 and 1:100)
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
30/89
30
spectrum that changes with time can yielda different average on each sweep when weuse video filtering. However, if we choosetrace averaging over many sweeps, we willget a value much closer to the true average.See Figures 2-32a and 2-32b.
Figures 2-32a and 2-32b show how videofiltering and trace averaging yield different
results on an FM broadcast signal.
full effect of the averaging or smoothing ateach point on the display as the sweep pro-gresses. Each point is averaged only once,for a time of about 1/VBW on each sweep.Trace averaging, on the other hand, requiresmultiple sweeps to achieve the full degree ofaveraging, and the averaging at each pointtakes place over the full time period neededto complete the multiple sweeps.
As a result, we can get significantly differentresults from the two averaging methods oncertain signals. For example, a signal with a
Trace averaging
Digital displays offer another choice forsmoothing the display: trace averaging.Trace averaging uses a completely differentprocess from the smoothing performed usingthe average detector. In this case, averagingis accomplished over two or more sweepson a point-by-point basis. At each display
point, the new value is averaged in with thepreviously averaged data:
Aavg = AnAprior avg +( () )n 1n 1nwhereAavg = new average valueAprior avg = average from prior sweepAn= measured value on current sweepn = number of current sweepThus, the display gradually converges to anaverage over a number of sweeps. As withvideo filtering, we can select the degree of
averaging or smoothing. We do this by set-ting the number of sweeps over which theaveraging occurs. Figure 2-31 shows traceaveraging for different numbers of sweeps.While trace averaging has no effect onsweep time, the time to reach a givendegree of averaging is about the same aswith video filtering because of the numberof sweeps required.
In many cases, it does not matter whichform of display smoothing we pick. If thesignal is noise or a low-level sinusoid veryclose to the noise, we get the same results
with either video filtering or trace averaging.However, there is a distinct differencebetween the two. Video filtering performsaveraging in real time. That is, we see the
Figure 2-31. Trace averaging for 1, 5, 20 and 100 sweeps, top to bottom (trace position offset for each set
of sweeps
Figure 2-32a. Video filtering Figure 2-32b. Trace averaging
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
31/89
31
Time gating can be achieved using threedifferent methods we will discuss below.However, there are certain basic concepts oftime gating that apply to any implementa-tion. In particular, you must have, or beable to set, the following four items:
An externally supplied gate trigger signal
The gate control or trigger mode (edge orlevel) (The X-Series signal analyzers canbe set to gate-trigger holdoff to ignorepotential false triggers.)
The gate delay setting, which determineshow long after the trigger signal the gateactually becomes active and the signalis observed
The gate length setting, which deter-mines how long the gate is on and thesignal is observed
Measuring time divisionduplex signals
To illustrate using time-gating capability toperform difficult measurements, considerFigure 2-33a, which shows a simplifieddigital mobile-radio signal in which tworadios, #1 and #2, are time-sharing a singlefrequency channel. Each radio transmits a
single 1-ms burst, then shuts off while theother radio transmits for 1 ms. The chal-lenge is to measure the unique frequencyspectrum of each transmitter.
Unfortunately, a traditional spectrumanalyzer cannot do that. It simply showsthe combined spectrum, as seen in Figure2-33b. Using the time-gating capability andan external trigger signal, you can see thespectrum of just radio #1 (or radio #2 if youwish) and identify it as the source of thespurious signal shown, as in Figure 2-33c.
Time gating
Time-gated spectrum analysis allows youto obtain spectral information about signalsoccupying the same part of the frequencyspectrum that are separated in the timedomain. Using an external trigger signal tocoordinate the separation of these signals,
you can perform the following operations: Measure any one of several signalsseparated in time (For example, you canseparate the spectra of two radios time-sharing a single frequency.)
Measure the spectrum of a signal in onetime slot of a TDMA system
Exclude the spectrum of interferingsignals, such as periodic pulse edge tran-sients that exist for only a limited time
Why time gating is needed
Traditional frequency-domain spectrumanalysis provides only limited informationfor certain difficult-to-analyze signals.Examples include the following signal types:
Pulsed RF
Time multiplexed
Time domain multiple access (TDMA)
Interleaved or intermittent
Burst modulated
In some cases, time-gating capabilityenables you to perform measurements thatwould otherwise be very difficult, if notimpossible to make.
Figure 2-33a. Simplified digital mobile-radio signal in the time domain
Figure 2-33b. Frequency spectrum of combined
signals. Which radio produces the spurious
emissions?
Figure 2-33c. The time-gated spectrum of
signal #1 identifies it as the source of spurious
emission
Figure 2-33d. The time-gated spectrum of signal
#2 shows it is free of spurious emissions
-
5/22/2018 Spectrum Analysis Basics Agilent 5952-0292
32/89
32
Controlling these parameters will allowus to look at the spectrum of the signalduring a desired portion of the time. Ifyou are fortunate enough to have a gatingsignal that is only true during the period ofinterest, you can use level gating, as shownin Figure 2-34. However, in many cases thegating signal will not perfectly coincide withthe time we want to measure the spectrum.Therefore, a more flexible approach is to useedge triggering in conjunction with a speci-fied gate delay and gate length to preciselydefine the time period in which to measurethe signal.
Consider the GSM signal with eight timeslots in Figure 2-35. Each burst is 0.577 msand the full frame is 4.615 ms. We may beinterested in the spectrum of the signalduring a specific time slot. For the purposesof this example, lets assume we are usingonly two of the eight available time slots(time slots 1 and 3), as shown in Figure 2-36.When we look at this signal in the frequencydomain in Figure 2-37, we observe anunwanted spurious signal present in thespectrum. In order to troubleshoot the prob-lem and find the source of this interferingsignal, we need to determine the time slotin which it is occurring. If we wish to lookat time slot 3, we set up the gate to triggeron the rising edge of the burst in time slot 3,and, then specify a gate delay of 1.4577 msand a gate length of 461.60 s, as shown inFigure 2-38. The gate delay assures that weonly measure the spectrum of time slot 3
while the burst is fully on. Note that the gatestart and stop value is carefully selectedto avoid the rising and falling edge of theburst, as we want to allow time for the RBWfiltered signal to settle out before we make ameasurement. Figure 2-39 shows the spec-trum of time slot 3, which reveals that thespurious signal is not caused
top related