scalar network analysis: an attractive and cosl … · 2018-04-15 · scalar network analysis: an...
TRANSCRIPT
SCALAR NETWORK ANALYSIS:AN ATTRACTIVE AND COSl-EFFECTIVE
TECHNIQUE FOR TESTINGMICROWAVE COMPONENTS
Hugo VitianNetwork Measurements Division
1400 Fountain Grove ParkwaySanta Rosa, CA 95401
RF ~ MicrowaveMeasurementSymposiumandExhibition
Flio- HEWLETT~~ PACKARD
www.HPARCHIVE.com
Scalar Network Analysis: an Attractive andCost-Effective Technique for Testing MicrowaveComponents
ABSTRACT
Scalar network characterization is an economic alternativeto vector network analysis; in particular, in a productiveenvironment where ease-of-use, throughput and low cost arekey considerations.
However, scalar measurements are not restricted to frequencyresponse measurements on linear networks. Scalar networkanalyzers are very powerful analysis tools to characterizeamplifiers, mixers, modulators, and other components as afunction of signal level or control parameter, as well asfor cable testing in frequency- and distance-domains.
This presentation describes the fundamental measurementconcepts and discusses some specific examples.
Author: Hugo Vifian, Engineering Section Manager foreconomy vector network analyzers, HP Network MeasurementsDivision, Santa Rosa, CA. Diplom-Ingenieur and PhD fromSwiss Federal Institute of Technology, (ETH). With HP since1969, projects include HP 8755 frequency response test set,HP 8505 RF vector network analyzer, HP 8756 and 8757 scalarnetwork analyzers.
2
www.HPARCHIVE.com
INTRODUCTION TO SCALARNETWORK MEASUREMENTS
2617
INTRODUCTION TO SCALARNETWORK MEASUREMENTS
Introduction: Even though scalarmicrowave measurements have beenmade ever since high frequencieswere discovered, they were notreally addressed, as such, butrather called power measurements orVSWR measurements, etc. It was notuntil modern network analysis withvector error correction was introduced that the rather obvious termmagnitude or scalar measurementsbecame more popular.
The objectives of this presentationare:
Discuss practical scalar measurement applications.
Introduce a variety of networkcharacterization techniques in themicrowave frequency range.
OBJECTIVES• Introduce Scalar Measurement
Concepts
• Explain Network CharacterizationTechnique in the MicrowaveFrequency Range
• Discuss Practical MeasurementApplications
2618
To explainScalar andconcepts.
the difference betweenVector Network Analyzer
SCALARMEASUREMENT CONCEPTS
• Scalar and Vector Network AnalyzerComparison
2622
1. SCALAR MEASUREMENT CONCEPTS
The fundamental difference betweenVector and Scalar Network Analyzersstems from different receiver concepts applied. All the other systems characteristics are a director indirect result of this.
3
www.HPARCHIVE.com
1.1 SCALAR AND VECTOR NETWORKANALYZER COMPARISON
2639
SCALAR AND VECTOR NETWORKANALYZER COMPARISON
Receiver • Homodyne • HeterodyneConcept Mostly Mostly
Single Converter Multi-Converter(MW-DC) (MW-IF1-IF2-DC)
Front End • Diode Detector • FundamentalFrequency or orConverter Thermo Couple Harmonic Mixer
The Scalar Analyzer Receiver measures the magni tude of the microwave test signal to be measured byconverting it into a low frequencyAC or DC signal by means of a thermo electrical effect or a diodehomodyne process. (1)
Most Vector Network Analyzers, onthe other hand, are heterodynereceivers with multiple frequencyconverters based on fundamental orharmonic mixing with a local oscillator signal. (2)
SCALAR VECTOR
VECTOR NETWORK ANALYZER BLOCK DIAGRAMS
SignalProcessingandDisplay
Test Set and Receiverr-------------,I Front End
T
Sourcer --.,I
: rv}-!----!..---l+!
I
:AIIIIIIIL __.J
SweptSource
Subsequent filtering allows reducing the processing bandwidth andtherefore improving the signal-tonoise ratio which ultimately leadsto improved sensitivity and dynamicrange. Furthermore, the narrowbandreceiver characteristic suppressesharmonic and spurious signals atother frequencies.
The Vector Analyzer then measuresthe down converted microwave signalcharacteristics in terms of amplitude and phase at a low intermediate frequency (IF).
2642
SCALAR NETWORK ANALYZER BLOCK DIAGRAM
2643
L ~
BroadbandHomodyneReceiver
Source Test Set and Receiver
r - -, r - -0;;';. ;'t.:-o,- - - - - - --,I 0 I Front End
I rv~-!-_----;+f ¥ ~I----+::J----~I I II L __ :J Signa'':A
L---------Kl-----+f ~;::,c.ss'ngDisplay
IIIIIIL __ .J
SweptSource
In contrast, Scalar Network Analyzers with diode detectors asfrequency (down) converters areinherently broadband receivers. (4)
This gives them the capability tomeasure any signal in the entirefrequency band instantaneously. Afeature which is important forcharacterizing frequency translating components (mixers), or whenhigh frequency agility is required.The trade offs are: a) less sensitivity mainly because of thebroadband noise and lower conversion efficiency of the diode converter and b) limited rejection ofundesirable signal components. (5)
4
www.HPARCHIVE.com
-10 dBm
2659
A simple way to model such a configuration is a mixer multipliercircuit with both RF and La inputsconnected together. At high RFpower levels, the signal at the Lainput is sufficient to convert theRF signal proportionally (with anominal conversion loss) to theoutput of the mixer.This operating area is calledlinear (conversion) for the diodedetector and an incremental changeat the input generates a proportional change in (detected) outputsignal.As the RF signal is reduced, the Lacomponent in the model drops andthe conversion efficiency is reduced until the RF and La only contribute in a mUltiplicative way tothe output signal. Therefore atlow signal levels, an incrementalRF input signal drop translatesinto twice the drop of detectedsignal at the output. This diodeoperating area is called square law(conversion) since the detectedsignal is proportional to the product of RF and La which is RFsquared.When used as a Scalar Analyzerfront end, the diode square lawcharacteristic is extended by appropriate signal processing overthe entire operating range (dynamicrange) in order to maintain a calibrated display.
VECTOR• Narrow Band• High
Selectivity
• High Sensitivity(~-100 dBm,Wide DynamicRange)
• SynchronouslyTuned LocalOscillator
• Higher Cost& Performance
SCALAR
• Self-ConvolvingMixer WithoutLocal Oscillator
• Lower Cost& Performance
DIODE DETECTOR AS FREQUENCY CONVERTER
Log Vo
Voltage !OutputPro ortionalto Power
SCALAR AND VECTOR NETWORKANALYZER COMPARISON
to Power !IL-+--+--+-+--+-+--+--+ Log
PIN = V,ZZo
Basic • Broad BandCharacteristics • Very Frequency
Agile
• LowerSensitivity(~-60 dBm)
2640
SCALAR AND VECTOR NETWORKANALYZER COMPARISON
SCALAR VECTOR
Having the signal vectors availablein complex form (amplitude andphase) allows the applicationsophisticated error correctiontechniques which dramatically reduces measurement ambiguities. (3)
FundamentalMeasurements
Accuracy
• SignalMagnitudeorMagnitudeSquared(Power)
• Scalar ErrorCorrectionOnly
• SignalAmplitudeandPhase
• ElaborateVector ErrorCorrection
However, the improved performanceof a Vector Network Analyzer doesnot come for free and the hardwarecost of the heterodyne structurewith the synchronously tuned localoscillator and the additional processing capability adds substantially to the cost of this measurement system.
2641
5
www.HPARCHIVE.com
1.2 FUNDAMENTAL SCALARMEASUREMENTS
Let's quickly review the two mostcommon scalar measurement concepts:1. Transmission2. Reflectionas a function of frequency. (6)(7)
FUNDAMENTAL SCALAR MEASUREMENTSTransmission (vs. Frequency)
TDUT
A
TRANSMISSION: 2647
The swept source provides a testsignal over the frequency range ofinterest which is applied to thedevice under test (DUT) and thetransmitted signal (T) is measuredby the Scalar Analyzer. Since weare usually interested in Transmission Gain or Loss, the incedentsignal (I) is measured and comparedwith the transmitted signal and thesignal ratio is computed as shownin the graph below.
Gain/Loss ; TransmittedIncident
Frequency
As an example, the transfer function or gain of a band-pass filteris displayed as a function offrequency.
2648
REFLECTION:
Reflection Measurements (vs. Frequency)
OUT
DevicePort
ITest
Directional PortDevice I
Slotted Line IncidentCoupler ~~~=":""--I'"Bridge
Source
In a reflection measurement, theincident and reflected signal areseparated by a directional device(bridge coupler, etc.) and measuredeither separately, or as a superposition in form of a standing wavepattern (slotted line).
2649
6
www.HPARCHIVE.com
r'M (~, V5WR, RL, 5,,)
Frequency
The reflection coefficient, gammais then computed by either formingthe ratio of reflected (R) overincident (I) signal or by converting the standing wave ratio asshown later.
r, * r'MMeasurement Errors are a Function of:Directivity (D), Test Port Match (rTP) andDevice Match (r,)
2658
ACCURACY CONSIDERATIONS:
Unfortunately, a perfect separationof incident and reflected signal isnot possible. A part of the incident signal, attenuated by the directivity (D) of the directionaldevice will be present also. Furthermore, since the directionaldevice does not have a perfect portmatch, some of the reflected signal(gamma-L) gets re-reflected(gamma-TP). Mainly, these two error terms cause the measuredreflection coefficient (gamma-LM)of the device under test to be different than the actual term(gamma-L).
40
0::o0::
t5 0
L:
~ +2 h..e-~·-~-~~H
~ +3L: +4
+50
INF-5
rTP
-4
III -3"U
- 2 h,,',5~~~i-------:7"
2650
The chart and the table show how wecan estimate the measurement ambiguity caused by the directivity (D)and the test port match (gamma-TP)as a function of the value ofreflection coefficient to be measured (gamma-L).
For high returnloss measurements(gamma-L >20dB), the directivity(D) dominates the ambiguityand for lower returnloss(gamma-L (10dB),the test port matchbecomes the major contributor.
10
*
*
~2.0
rTP
VSWR
30
40
40
DdB
40
4 dB
10 dB
30 dB
20 dB
Examples: ErrorAmbiguity
30 +3.5 dB-2.5
20 + 1.0 dB- .9+ 1.3 dB-1.2
4 +2.5 dB-2.0
CD 4 dB 40 ~1.1 4 ±.4 dB
For High (~20 dB) Return Loss MeasurementsDirectivity Dominates, for Low (~10 dB) R.L.the Test Port Match Dominates
CDCDCD
2651
7
www.HPARCHIVE.com
2. SCALAR NETWORKCHARACTERIZATION TECHNIQUES
2.1 OVERVIEW
A Scalar Analyzer is really ageneralized stimulus response testset capable of doing a wide varietyof network characterization tasks.
Frequency response measurements areprobably the most popular forlinear networks.
Other networks may exhibit signallevel dependent parameters likeamplifiers.
Another group of stimulus responsemeasurements allows us to characterize networks in a time/distanceor impedance domain.
ADVANCEDNETWORK CHARACTERIZATIONTECHNIQUES: AN OVERVIEW
2619
MEASUREMENT DOMAINS• Frequency Response (or Versus
Frequency) Measurements
• Network CharacterizationVersus Signal Level
• Frequency, Time and ImpedanceDomain Characterization
2619
STIMULUS -+ RESPONSECONCEPT
8
Stimulus
Response(Reflection)
Deviceor
NetworkUnderTest
Response(Transmission)
~
2620
www.HPARCHIVE.com
FREQUENCY RESPONSE OF LINEAR NETWORKS
2.2 FREQUENCY RESPONSE OF LINEARNETWORKS
Stimulus(Source)
OUT
TransmittedSignal
We already covered the most important scalar measurements (transmission and reflection) of this classof networks.
2621
Example: Filter
Frequency Response is Independentof Signal Level (Linear)
An additional selection of relatedmeasurements is listed below.
Transmission Reflection
2622
BASIC FREQUENCYRESPONSE MEASUREMENTSTRANSMISSION
• Gain/Loss
• Bandwidth
• Attenuation
• Coupling
• Directivity
• Etc.2623
REFLECTION
• Return Loss
• Impedance
• VSWR
• ReflectionCoefficient
• Etc.
9
These parameters are either directly derived from the basic transmission or reflection measurementslike bandwidth and directivity orare computed from the fundamentalmeasurement like VSWR fromreturnloss either internally to theAnalyzer, or externally with acontroller/computer.
www.HPARCHIVE.com
CHARACTERIZATIONOF PSEUDO-LINEAR NETWORKS
2.3 CHARACTERIZATION OF PSEUDOLINEAR NETWORKS
For example, an amplifier is a network where some of the keyparameters may depend on the signallevel. In that case, the frequencyresponse has to be measured forvarious power levels to fullycharacterize its behavior.
Amplifiers
A
InputSignal
l>Example: Amplifier
OutputSignal
2624
Furthermore, it is often importantto meC\sure the gain as a functionof the input or output signal toestablish the maximum power gain orthe compression points.
Gain
Frequency Response Depends onSignal Level (Non-Linear)
Gain
~-20dBm
v:::-
Frequency..
Signal Level
..-2625
Here is a set of additional measurements that might be required tobetter characterize this class ofnetworks.
10
ADDITIONAL FREQUENCYRESPONSE MEASUREMENTS• Isolation (Reverse Gain/Loss)
• Large Signal Gain
• Maximum Output Power
• Gain Compression
• Noise Figure
• Harmonic Distortion
• Etc.2626
www.HPARCHIVE.com
CHARACTERIZATION OF NON-LINEAR NETWORKS
Mixers and Frequency Translators
I--+-.... IF
LO
,1_J ...
2630
2.4 CHARACTERIZATION OFNON-LINEAR NETWORKS
Mixers and frequency translatorsare usually considered non-linearnetworks even though their frequency response is measured as if theywere linear components.
The flatness of the conversion fromRF to IF port (as a function of RFor LO frequency). as well as thefrequency response for various LOdrive levels has to be established.
IF Signal Level
...
IF Signal Level
LO +10 dBm
~, LO +8 dBm \
...
2631
RF Frequency....
LO Frequency
2632
OTHER PARAMETERS• Conversion (Loss) Efficiency
• Compression
• Feedthrough, Isolation, Balance
• Third Order Intercept (TOI)
• Etc.
11
A scalar network analyzer is theideal measurement tool to characterize these components because ofthe inherent broadband characteristic of the receiver.
www.HPARCHIVE.com
A different example of a non-linearnetwork is shown in the followingslides. A voltage controlled attenuator (or modulator) is characterized as a function of the control signal.
CHARACTERIZATION OF
NON-LINEAR NETWORKS
Voltage Controlled Attenuators
ControlSignal
OutputSignal
2627
The figure shows attenuationcharacteristic for two differentfrequencies as well as a deviationfrom the desired control functionplot.
Additional examples of componentswhich can be characterized in asimilar fashion are listed below.
Attenuation
...\\\
f, \\,'---
Control Signal
AttenuationControl Signal
Deviation fromDesired Control Function
Control Signal
2628
12
OTHER EXAMPLES• Modulator/Demodulator
• AGe Amplifier
• Log Amplifier
• Etc.
2629
www.HPARCHIVE.com
FREQUENCY & TIME DOMAINCHARACTERIZATION
r------iREF
I II I
OUT
..A2636
2.5 FREQUENCY AND TIHE DOHAINCHARACTERIZATION
The Scalar Analyzer concept (homodyne receiver) lends itself nicelyto do time and distance domaincharacterization via frequency domain measurements. An inverseFourier transform computation isused to give the time domaindisplay •
Equal Lengths Unequal Lengths
The basic operating principle isexplained below. The diode detectorconvolves the transmi tted (T) and thereference (REF.) signal resulting ina ripple pattern over the frequencyspan of interest due to the phasing ofthe two microwave vectors.
Frequency
.....Frequency
-...
If the reference and transmitted pathlengths are equal, the resultingswept response would be a flat traceas there is no time delay between thetwo signals. When one path length islonger, a ripple pattern will beexhibited.
2637
Frequency
Sj
Ref.
Trans.
MultipathII
~Time
Delay = l!..t.t
13
The number of envelop periods (N)divided by the frequency change(delta-f) will indicate the delay ofthe device under test with respect toa reference calibration.
Higher order ripples will occur ifthe signal is reflected and travelsthrough the network again or whenmulti path transmission occursthrough or around the device undertest.
www.HPARCHIVE.com
The major applications for thismeasurement principle are listedhere.
MEASUREMENT APPLICATIONS• Time Delay Measurements
• Fault Location
• Multipath Transmission
• Etc.
2638
3. PRACTICAL MEASUREMENTAPPLICATIONS
PRACTICAL MEASUREMENT
APPLICATIONS
14
www.HPARCHIVE.com
3.1 INCREASED DYNAMIC RANGEMEASUREMENTS
INCREASED DYNAMIC RANGE
Instead of only utilizing thedynamic range of our receiver channel (T), the incident signal (1) isalso varied as a function of theattenuation of the device undertest by means of the leveling loop.
As previously mentioned, scalaranalyzers have usually less dynamicrange than vector analyzers do.However, there are ways to significantly extend the measurement rangeby considering the following testset-up. (8)
+10 dBm
">--oW-TOUT
+10 dBm,r -30
:"11~I IL .....J
Leveling (+ 10 - 30 dBm)
2644
SIGNAL LEVELS AS A FUNCTIONOF OUT ATTENUATION:
For high attenuation, the amplifierin the test path brings the signallevel up by 40 dB at the same timethe source puts out its maximumpower (+10 dBm).
OUT Loss Incident Transmitted
CD o dB -40 dBm + 10 dBm20 -20 +10
® 40 0 +1060 0 -1080 0 -30
@ 100 0 -50
2645
devicetranssimulis reto exoutput
For low attenuation of theunder test, the (amplified)mit ted signal rises andtaneously the source powerduced via leveling loop notceed the maximum measureablesignal.
The ratio Til therefore varies from+50 to -50 dB thus displaying 100dBof dynamic range.
Frequency
A(w) of OUT
o dB
-100 dB
-50 dB
2646
15
www.HPARCHIVE.com
3.2 GAIN COMPRESSION MEASUREMENTSON AMPLIFIERS
The next example shows an application where the network has to becharacterized as a function of signal level to establish the maximumundistorted output power or at whatinput (or output) level the gaindrops by a given amount. For example. by 1 dB which is called the1dB compression point.
GAIN COMPRESSION MEASUREMENTA. Versus Signal Level (Fixed Frequency)
t----+.--~T ~ P~,
Sweep InputPower Leveland Measure Gain vs. p," or P••,
2669
Gain Compression (a) vs.Level (b) V5. Frequency
Compression
P~, (Set) t
Frequency
Gain
p... or P~
Signal Level
GAIN!I
The test is done in two steps:a) as a function of signal level
at a constant frequency. andb) as a function of frequency at
a certain output level.
(a) (b)
2670
l
OUT
t--~...-~ T ~ P~,
~ 11~ALC ------:-I------}) I ~ P~
I IL .J
B. Versus Frequency (Fixed Power)
The set-up for the latter measurement is shown here and includes theleveling loop to establish adesired power level at the output.The display indicates gain compression as a function of frequency forthis set output power.
Set Desired Output Powerand Measure Gain vs. Frequency
2671
16
www.HPARCHIVE.com
REFLECTION MEASUREMENT
3.3 REFLECTION MEASUREMENT WITHTHE LONG LINE OR RIPPLETECHNIQUE
Long line or Ripple Method
Example: Slotted Line
As shown in the first paragraph,scalar reflection measurements areusually taken by inserting a directional bridge or coupler as a signal separation device in front ofthe device under test. Comparingthe incident (I) and reflected (R)signal leads to the measuredreflection coefficient of the network, which includes the ambiguities due to directivity (D) andtest port match of the set up.
Device Under Test
R
Reflected Ir,le-.' + 0,
IIII
...~.------ .........~ IChange ProbePosition
Slotted Line e-w
Probe
Source
ChangeFrequency
2652
REFLECTION MEASUREMENTVSWR = Em.. =~ r = ~ ? = Irl
Em'" I-R I
In order to understand some of theimprovements and new developmentsin scalar measurements, it may bebeneficial to digress back to theslotted line reflection measurements as shown here.
Em,. = I + R The reflection coefficient in thiscase, is determined by measuringthe standing wave pattern along thetransmission line. Unlike in thecase of the reflection measurementwith a directional bridge (or coupler), this method is basicallycapable of measuring the complexreflection coefficient gamma (withthe appropriate calibration andsome computational effort).
2653
M = V12 + R2 - 2 IR cos 0
o = 2{3f - 0 L
_ 360 0 X Frequency{3 - Propagation Velocity
M [em] = ~ = 15 [em]f M [GHz]
M [GHz] = Y.. = 15 [em]f M [em]
Because of the duality betweenstanding wave patterns versusfrequency and length, this methodcan be readily expanded to measurediscontinuities (reflections) incables when the frequency is var-ied. By scaling the electricallength (1) with the appropriatepropagation velocity, the locationof the discontinuity can be calculated (e.g., open end of cable).
Llt [ns](Round Trip)
2M=v
= M [em] =15 [em]
2M [GHz]
2654
17
www.HPARCHIVE.com
3.4 MULTIPLE REFLECTIONS ANDFAULT LOCATION
REFLECTION MEASUREMENTThe measurement concept just discussed can be expanded to resolvemultiple reflections. Depending onthe location and size of discontinuities, the scalar analyzer willshow a superposition of ripple patterns which can be analyzed by extracting the indivivual mismatchripples.
Multiple Reflectionsand Fault Location '1 '2
The inverse Fourier transform (withthe appropropriate windowing) willconvert the ripple pattern in individual impluse functions spacedaccording to the electrical lengthof the position of the discontinuites along the propagationpath.
The resolution in the time or distance domain is obviously inverselyproportional to the higher frequency component of the test signal(@ 20 GHz A 1.5cm or .05ns).
In summary, scalar measurementswere presented as an economicalalternative for many microwave andRF measurements. We coveredseveral general and some specificapplications, and we will demonstrate some examples with the associated hardware and software atthe exhibits. (9)( 10)
18
......1------1'3-------1~
2655
r Ir,l
Separation of Components .... Fourier Transform
-- Time or Distance Domain Display
Ripple
2656 2657
SUMMARY ANDCONCLUSIONS
• Scalar Measurements Are anEconomical Alternative to MostMicrowave Measurements
• Several Advanced MeasurementApplications Were Discussed
• Hardware and SoftwareDemonstrations Will Be Givenat the Exhibits
2621
www.HPARCHIVE.com
REFERENCES
1. Hewlett-Packard Journal, November 1972
2. Hewlett Packard Journal, July 1976
3. Stephen F. Adam, MICROWAVE THEORY AND APPLICATIONS(Prentice-Hall, Inc., 1969)
4. Ray J. King, MICROWAVE HOMODYNE SYSTEMS (Peter Peregrinus,Ltd., 1978)
5. Kenneth K. Clarke and Donald T. Hess, COMMUNICATION CIRCUITS(Addison-Wesley, 1978)
6. HP MICROWAVE SCALAR NETWORK MEASUREMENT SEMINAR 1985
7. B. Eicher and C. Staeger, A SURVEY OF REFLECTION MEASUREMENTMETHODS (Technische Mitteilungen PTT, 1982)
8. HP Application Note #327-1
9. 8756A Operating and Service Manual (1983)
10. 85015A Systems Software Operating Manual (1983)
19
www.HPARCHIVE.com
March 1985
www.HPARCHIVE.com
Printed in U.S.A.