2006

Nov1,2006

RFMeasurementBasicsforNonRFTestEngineers
RobertGreen,KeithleyInstruments

WirelesscommunicationsproductsthatdependonRFprinciplesare
everywherethesedays,andthegrowthrateisastonishing.Fromcellphones
andwirelessPDAstoWiFienabledlaptops,Bluetoothheadsets,RFIDtags,
wirelessmedicaldevices,andZigBeesensors,theRFdevicemarketis
booming.Thisyearalone,morethan850millioncellularphoneswillbe
manufacturedandsoldaroundtheglobe.
Forthoroughproducttestingandhightestthroughput,testengineersneedto
understandthebasicsofRFtechnology,whattotestfor,andwhich
instrumentsarebestsuitedforthejob.Mostengineerswithexperiencein
lowfrequencyapplicationsbelow1MHzwillnotlikelybefamiliarwithhigh
frequencyapplications.
RFTerminology
ThinkinTermsofPower
RFsignalstrengthcanvarybyvastamounts.Asthesignalpropagates
throughspace,thepowerperunitareadecreasesinproportiontothe
distancesquared.Changesinpoweraremeasuredindecibels(dB).
Usingdecibelsforpowermeasurementsgreatlysimplifiescalculations.
GainsorlossesindBaddorsubtract.Forinstance,themultiplication

operationreducestosimpleaddition.TheformaldefinitionofdBis:
dB=10log(Pout/Pin)
AdBvalueisarelativequantity.ArelatedunitisdBm,whichistheabsolute
powermeasuredrelativeto1mW.Figure1showsdBmvaluesandtheir
correspondingvaluesinwatts.Thepowertransmissionrangeofamobile
phoneisshownforreferenceaswellashowlowasignalasensitivereceiver
candetect.

Figure1.RelationofdBmtoWattsWithTypicalSensitivityand
TransmitterRangesforReference

Figure2showsanequationdefiningthetheoreticalnoisefloorforRFsignals
atroomtemperature.DuetoanRFsignalslossypropagationthroughair
aswellasatmosphericinterferenceandinterferencefromothersignals,the
signallevelthatreachesthereceivercanbequitelow.Itisnotunusualfora
receivertodetectsignallevelsbelow0.1pW.

Figure2.EquationforTheoreticalNoiseFloor

MismatchonTransmissionLines
Atlowfrequency,thegoalistotransfervoltagesthroughcircuitswithminimal
lossinmagnitude.Themosteffectivecircuitshavehighinputimpedanceand
lowoutputimpedance.
WithRFapplicationswhereacablelengthcanbeaquarterwavelength,
signalsmustbetreatedaswaves.Anytimeawavehitsadiscontinuity,
someofthewaveisreflected.

ThegoalofRFistotransferallthepowertotheloadwithoutloss.Any
reflectionofpowermeansnotallofthepowerisgettingtotheloadso
mismatchisacriticalparameter.Anydifferenceinimpedancebetweencircuit
elementsandthetransmissionlinecausesreflectionsandlossofpower.
InRFapplications,transmissionlinesgenerallyarecoaxialcablesexternalto
circuitboardsandmicrostripswithincircuitboards.Thesecomponentshave
acharacteristicimpedance.Theexpressionforthecharacteristicimpedance
ofatransmissionlinedependsonthegeometryoftheconductors,the
propertiesoftheconductors,andtheinsulatorholdingorseparatingthe
conductors.
ForRFapplications,thecharacteristicimpedanceofthetransmissionlines
andtheinputandoutputimpedancesofcomponentsaredesignedtobe50or
75?.A50?impedanceisusedtooptimizepowertransferinasystem75?
systemsaredesignedforminimumattenuationinapplicationssuchascable
systems.MostRFwirelesstransmissionsystemsoptimizedforpower
transferare50?characteristicimpedancesystems.
Tominimizereflections,RFcablesandcomponentsforwirelesstestand
measurementapplicationsaredesignedfor50?.Conversely,theoptimal
powertransfertakesplacewhenimpedancesarematched.
Awavepassingfromonecharacteristicimpedancetoanothercauses
reflection.Iftheimpedancesarethesame,thereisnoreflection.Incases
wherethereisareflectedwaveduetoanimpedancediscontinuity,therewill
bewavestravelinginbothdirectionsonthetransmissionline.

Atsomepointwherethewavesareinphase,amaximumvoltage(Vmax)will
occurandwherethewavesare180degreesoutofphase(Vmin).Theratio
ofVmaxtoVministhevoltagestandingwaveratio(VSWR).Thisisone
indicationofhowcloseaconnectororacableisto50?.
Figure3givestheformulasfordeterminingtheothermeasuresofmismatch
from50?.Thereflectioncoefficient(?)isadirectindicationofthepercentage
ofthesignalthatisreflectedatadiscontinuityorachangeinimpedance
suchasacabletoinstrumentconnectororantennatolownoiseamplifier.
Thereturnlossisameasureoftheattenuationtoareflectedsignal.Ahigh
returnlossisdesirable.

Figure3.StandardEquationsforDeterminingMismatchFrom50?

Figure4showstherelationshipbetweenthethreeparametersfortheideal
case,aperfectmatch(noreflection),theidealopencircuit(100%reflection),
andthreevaluesbetweentheextremes.Testinstrumentationtypicallyhas
inputoroutputVSWRsinthe1.2:1to1.6:1range.

Figure4.RelationofVSWRtoMismatchParameters

NewConnectors,Cables,andComponents
CableswithBNCconnectorstypicallybegintodegradeabove500MHz.In
theRFworld,cablesoftenareequippedwithNconnectorsandSMA
connectors.Nconnectorscommonlyareusedontestinstrumentation
becausetheyarerugged,canhandlehighpowers,andperformwellupto
about18GHz.TheSMAconnectorismuchsmallerandratedforlower
powerthantheNconnector,butitcanbeusedwellbeyond18GHz.
AllRFcablesarecoaxial.CoaxialRFcablescanbeinflexibleorrigid,flexible
foralimitednumberofbends,orflexible.Careofthecableismuchmore

importantforRFthanlowfrequencycables.Excessivebendingofthecable
and90?bendscandamagethecableandseverelydegradeperformance.
Atlowfrequencies,agoodconnectionmeansthattheconductorsarein
contactwitheachother.AtRFfrequencies,theimportanceofmismatch
meansthatagoodconnectionnotonlyhastheconductorsincontact,but
alsothattheconnectorsareproperlytorquedtogether.Manufacturers
recommendabout7ftlboftorquetoensuregoodcontactandminimal
insertionlossbetweentheconnectors.

Maintainingthe50?Line
ParallelconnectionsormultiplesignalpathsinRFcircuitsarenotassimple
asinlowfrequencycircuits.Maintainingamatchedcircuitpathtominimize
discontinuitiesandsignalreflectionsiscritical.
RFswitchesareprecisionmachinedanddesignedtomaintain50?
impedancethroughtheswitch.Toeffectaparallelpath,devicesknownas
splittersordividersseparateaninputsignalpathintotwoormoreoutput
paths,eachwith50?impedance.Combinersperformtheoppositefunction
byconvertingmultipleinputpathsintoasingleoutputpath.
ThesearejustafewofthespecializedcomponentsneededforRFtest
systems.IfyouarenewtoRFtest,bepreparedforstickershock.RF
componentscostmuchmorethantheirequivalentDCcomponents.
WhatDoYouNeed?
Aswiththebreadthoflowfrequencytestinstruments,theworldofRFtest

instrumentsiswideandvaried,rangingfromsignalsourcesandpower
meterstospectrumandnetworkanalyzers.Theseinstrumentsareusedto
generateRFsignalsandmeasureawiderangeofsignalparameters.

RFPowerMeters
PoweristhemostfrequentlymeasuredRFquantity.Apowermeter
essentiallymeasuresthepowerofRFsignals.Itusesabroadbanddetector
andreportsabsolutepowerusuallyinwatts,dBm,orpossiblydB?V.Forthe
majorityofpowermeters,thebroadbanddetectororsensorisanRF
SchottkydiodeordiodenetworkthatperformsanRFtoDCconversion.
PowermetersprovidethebestaccuracyofanyRFinstrumentformeasuring
power.Highendpowermetersoftenrequiringanexternalpowersensorcan
measurewith0.1dBorbetteraccuracy.Powermeterscanoperatedownto
near70dBm.Sensorsrangefromhighpowermodelstohighfrequency
modelstohighbandwidthmodelsforpeakpowermeasurement.
Powermetersareeithersinglechannelordualchannelinstruments.Each
channelrequiresitsownsensor.Twochannelsprovidethecapabilityto
measureinputandoutputpoweronadevice,circuit,orsystemandcompute
againorloss.
Somepowermetershavehighmeasurementspeedsof200to1,500
readings/s.Somepowermeterscanmeasurepeakpowercharacteristicsof
manytypesofsignalsincludingmodulatedsignalsandpulsedRFusedin
communicationsandotherapplications.Twochannelmetersalsomake
accuraterelativepowermeasurements.Powermeterscanbepackagedinto

smallenclosuresdesignedforportability,makingthemsuitableforuseinthe
field.
Themainlimitationofapowermeterisitsamplitudemeasurementrange.The
widefrequencyrangeisatradeoffformeasurementrange.Inaddition,a
powermeterwillprovidethemostaccuratemeasurementofpowerbutwill
givenoinformationonthefrequencycompositionofthesignal.
RFSpectrumorRFSignalAnalyzer
AspectrumorvectorsignalanalyzermeasuresRFsignalsinthefrequency
domainusingnarrowbanddetectiontechniques.Theprimaryoutputdisplayis
aspectrumofbothabsoluteandrelativepowervs.frequency.Theoutput
alsocanbeademodulatedsignal.
Spectrumanalyzersandvectorsignalanalyzersdonothavetheaccuracyof
powermetershowever,thenarrowbanddetectiontechniquesusedinthese
RFanalyzersenablethemtomeasuredowntolevelsaslowas150dBm.
RFanalyzershaveaccuraciestypicallyatandabove0.5dB.
Spectrumandvectorsignalanalyzerscanmeasuresignalfrequenciesfrom
kilohertzto40GHzandbeyond.Thewiderthefrequencyrange,thegreater
thecost.Themostcommonanalyzersextendto3GHz.New
communicationsstandardsthatoperateinthe5.8GHzregionrequire
analyzerswith6GHzandhigherbandwidths.
Vectorsignalanalyzersarespectrumanalyzerswithaddedsignal
processingcapabilitythatnotonlymeasureasignalsamplitude,butalso
decomposethesignalintoitsinphaseandquadraturecomponents.Vector
signalanalyzerscandemodulatemodulatedsignalssuchasthosegenerated

bymobilephones,wirelessLANdevices,anddevicesoperatingonother
newandemergingstandards.Vectorsignalanalyzerscandisplay
constellationdiagrams,codedomainplots,andcomputemeasuresof
modulationqualitysuchaserrorvectormagnitude.
Traditionalspectrumanalyzersareknownasswepttuneddevicesbecause
alocaloscillatorissweptacrossafrequencyspansothatanarrowbandfilter
canacquirethepowercontentattheindividualfrequencieswithinthe
frequencyspan.Vectorsignalanalyzersalsosweepoveraportionofthe
spectrum,buttheycapturewidefrequencysegmentsofdata.Asaresult,
vectorsignalanalyzerscangenerateaspectrummorequicklythan
spectrumanalyzers.
Akeymeasureofavectorsignalanalyzersperformanceisits
measurementbandwidth.Thenewhighbandwidthcommunications
standardssuchasWLANandWiMaxgenerate20MHzbandwidthsignals.
Theanalyzermusthavealargeenoughbandwidthtoacquirethewhole
signal.Iftestinghighbandwidth,digitallymodulatedsignals,makesurethe
analyzerhasthemeasurementbandwidthtoadequatelycapturethesignal.
Aspectrumanalyzerwillverifythatatransmitterisgeneratingtheappropriate
powerspectrum.Ifdistortioncomponentssuchasharmonicsorspurious
signalsmustbetested,thenaspectrumorvectorsignalanalyzerisneeded.
Examplesofotherteststhatrequireaspectrumanalyzerorvectorsignal
analyzerincludetestingforintermodulationdistortion,thirdorderintercept,
the1dBgaincompressiononapoweramplifierorpowertransistor,anda
devicesfrequencyresponse.

Testingatransmitteroramplifierthatmustprocessdigitallymodulated
signalsrequiresavectorsignalanalyzertodemodulatethesignal.The
vectorsignalanalyzercanmeasurehowmuchmodulationdistortionadevice
iscreating.
Thedemodulationprocessisacomplex,computationintensiveprocess.
Vectorsignalanalyzersthatperformthedemodulationandmeasurement
computationsquicklycansavevaluabletesttimeandsubstantiallycuttest
costs.

RFSourcingOptions
AllRFsignalsourcesgeneratecontinuouswave(CW)RFsinewave
signals.SomesignalgeneratorsalsocanmodulateanRFsignalwhilevector
signalgeneratorsuseIQmodulatorstogeneratedigitallymodulatedsignals.
TypesofsourcescanbefurtherdistinguishedasfixedCWsinusoidalwave
outputs,instrumentsthatsweepoverarangeoffrequencies,andanalog
signalgeneratorsandvectorsignalgeneratorsthataddanaloganddigital
modulationcapabilities,respectively.
Iftestrequirementscallforastimulussignal,anRFsourceisneeded.Key
requirementsforRFsourcesincludefrequencyandamplituderanges,
amplitudeaccuracy,andmodulationqualityforsourcesthatgenerate
modulatedsignals.Frequencytuningspeedsandamplitudesettlingtimes
alsoarecriticalforminimizingtesttime.

Vectorsignalgeneratorsarehighperformancesourcesthatoftenincorporate
arbitrarywaveformgeneratorsfordigitalsignalgeneration.Thearbitrary
waveformgeneratorenablesthevectorsignalgeneratortoproduceanykind
ofdigitallymodulatedsignal.
Manywaveformscanbegeneratedinternally,andinsomecases,a
waveformcanbecreatedexternallyanddownloadedintotheinstrument.If
thetestspecificationsrequireacomponent,device,orsystemtobetested
withthemodulationthatthedeviceundertestwillprocessinitsenduse,a
vectorsignalgeneratoroftenisneeded.
RFsourcingisusediftestspecificationscallforreceiversensitivitytests,bit
errorratetests,adjacentchannelrejection,twotoneintermodulation
rejection,ortwotoneintermodulationdistortion.Thetwotoneintermodulation
testsandtheadjacentchannelrejectiontestrequiretwosources.The
receiversensitivitytestandthebiterrorratetestmusthaveasingleRF
source.
Adeviceusedinthemobilephoneindustrymostlikelywillrequiretesting
withthetypeofmodulatedsignalrequiredbythemobilephonestandard.A
mobilephonepoweramplifierwillbetestedwithamodulatedsourcesuchas
avectorsignalgenerator.Beforeselectingavectorsignalgenerator,
evaluatethespeedatwhichtheinstrumentcanswitchbetweendifferent
modulatedsignalstoensurethegeneratorprovidesthefastestpossibletest
times.
NetworkAnalyzer
Athirdtypeofanalyzerisanetworkanalyzer.Networkanalyzerscombine
aninternalRFsourceandeitherabroadbandornarrowbanddetectortotest

RFdevices.TheoutputdisplaysthedevicecharacteristicsinXY
rectangularcoordinates,apolardisplay,oraSmithchart.
Essentially,avectornetworkanalyzermeasurestheSparametersofa
device.Avectornetworkanalyzercanprovidebothmagnitudeandphase
informationanddeterminetransmissionlossesandgainsofthesedevices
overawidefrequencyrangewithgoodaccuracy.Italsomeasuresreturn
lossandimpedancematchaswellasphasemeasurementsandgroupdelay.
Networkanalyzersareusedprimarilyforanalysisofcomponentssuchas
filtersandamplifiers.Beawarethatnetworkanalyzersworkwithcontinuous
waveunmodulatedsignalsandthatcalibrationoftheanalyzerisextremely
important.Amanufacturerscalibrationkitwillkeepthenetworkanalyzer
incalibration.
Becausenetworkanalyzerscombinesourcingandmeasurementinone
instrumentandbecausetheanalyzerhasawidefrequencyrange,theyare
expensiveinstruments.
TypicalApplication
AnexampleofanapplicationthatrequiresfourmajorRFtestinstrumentsis
poweramplifier(PA)testing.Asourcecanprovidetheinputsignal,andeither
apowermeteroraspectrumanalyzercanmeasureoutputpower.If
accuracyiscritical,suchasinamaximumpowermeasurement,thena
powermeterisneededfortheoutputmeasurement.
TheinputimpedancematchofaPAisakeyparameterforadesigner
developinganRFtransmitter.Itisimportanttoamplifyallthepowersupplied

tothePAandnotloseasubstantialamountduetoreflection.Forthisreason,
PAmanufacturerswillspecifyandmeasurereturnloss.
Alternatively,ifonlythescalarmagnitudeisrequired,thenasourceanda
spectrumanalyzerorpowermetercancombinewithacouplertomeasure
themagnitudeofthereflectedpower.Thesetupismorecomplicated
comparedwiththeuseofanetworkanalyzerbecauseadditionalpassiveRF
componentsarerequired.Thepowermeterwillprovidethemoreaccurate
powermeasurementforthereturnlossscalarmeasurement.
ThecapabilityofaPAtodeliverpowertoaloadwhoseinputimpedanceis
notmatchedtoatypically50?outputimpedanceisakeymeasureofthe
amplifierscapabilitytoperforminrealworldconditionswhereloadssuch
asantennasmaynothaveexactlya50?characteristicinputimpedance.In
suchcases,anon50?resistiveloadisswitchedtotheoutputofthePA.
TheloadcanforcethePAtooutputintoaVSWRofupto20:1a50?load
wouldresultinaVSWRnear1:1.ThePAmustbeabletofunctionproperly
anddeliversomepowertotheloadinthepresenceofalargeamountof
reflectedpower.
Someoutputmeasurementsrequirespectrumanalysis.RFPAsusedin
broadcastormobilephoneapplicationsmustnotgenerateexcesspowerin
frequencychannelsadjacenttothechannelwherethePAisoperating.
Adjacentchannelpower,intermodulationdistortion,andharmonicdistortion
aremeasuresofpowerthataPAgeneratesoutsidetheintended
transmissionchannel.Forthesemeasurements,dynamicrange,the

capabilitytomeasureasmallsignalinthepresenceofalargesignalsuchas
acarriersignal,isanimportantspectrumanalyzerspecification.
Forexample,considerthecasewhenaPAhasaspecificationthatits
adjacentchannelpoweris60dBc.Thedynamicrangeofthespectrum
analyzermustbeatleast6dBgreaterthantheminimumallowedpowerlevel
fortheharmonic,theadjacentchannelpowerlevel,ortheintermodulation
product.
Theadjacentchannelpowermeasurementmustbeperformedwitha
modulatedsignal,whichmeansthesourcesadjacentchannel
performancealsohastobeconsidered.Thesourcesadjacentchannel
poweroutputmustbeatleast6dBlessthanthemaximumallowable
adjacentchannelpowerthatthepoweramplifiercangenerate.
Forharmonicmeasurements,theanalyzershouldhaveafrequencyrange
threetimesgreaterthanthemaximumoperatingfrequencyofthePAto
adequatelycapturethepowerinthe3rdharmonicofthemaximumoperating
frequency.Again,thenoisefloorofthespectrumanalyzermustbeatleast6
dBlowerthanthe3rdharmoniccomponenttohaveareasonablesignalto
noiseratioforanaccurateandrepeatablemeasurement.Theharmonic
measurementsindicatetheamountofdistortionthePAcreates.Excessive
distortioncannegativelyaffectmodulationperformance.
IntermodulationdistortiondetermineshowmuchdistortionthePAgenerates
whensignalsatdifferentfrequenciesorcomponentsofasignalatdifferent
frequenciesareatthePAsinput.Twosourcesarerequiredtogenerate
thetestsignals.Onedualoutputsourceisinadequateduetoinsufficient
isolationbetweenthetwooutputs.Thesourcewouldcreateitsown

intermodulationdistortion,whichwouldleadtohigherandincorrectamplifier
distortionmeasurements.
ModulationqualitymeasurementsoftenaremadeonPAsdesignedforthe
mobilephonemarketandothermarketsegmentssuchasWLAN
applicationswherecomplexmodulationschemesareused.Thisusually
involvesmeasuringtheerrorvectormagnitude.
Conclusion
ThisoverviewofRFtestinstrumentationprovidesoverallguidanceonwhat
typesoftestinstrumentationareneededtomeettestrequirements.Inthe
vastmajorityofcases,anyoneoracombinationofthesefourinstruments
willbeneeded:signalsources,powermeters,spectrumanalyzers,and
networkanalyzers.
AbouttheAuthor
RobertGreenisseniormarketdevelopmentmanageratKeithley
Instruments.Inadditiontohavingmorethan10yearsofexperienceinthe
wirelessmarket,Mr.GreenearnedaB.S.inelectricalengineeringfrom
CornellUniversityandanM.S.inelectricalengineeringfromWashington
University,St.Louis.KeithleyInstruments,28775AuroraRd.,Cleveland,OH
44139,4402480400,email:rgreen@keithley.com
November2006

By:EvaluationEngineering
Publishedsince1962,EvaluationEngineeringhasdeliveredindepthtechnicalinformationtothe
testengineeringmarketformorethan50years,servingengineers,engineeringmanagers,and
corporatemanagersresponsiblefortestmeasurementandqualityofelectronicproductsand
systems.EEisamultimediaresourceprovidingamonthlymagazineanddigitaledition,dailye
newsletters,productshowcases,monthlyspecialreports,tradeshowevents,anda
comprehensivewebsiteforbuyersandspecifiersoftestequipmentinsemiconductors,medical,
communication,RF,microwave,andwirelessapplications.Anannualsubscriptionisfreeto
qualifiedtestengineeringprofessionals.