of Phase Noise Measurement
Transcription
of Phase Noise Measurement
I I o The "Arl" of Phase Noise f,lleasurement DieterScherer March1985 o Rf & Mlcrowave Measurement Symposlum ancl txhlbltlon ftE'HF$|IfJJ o O o o INTRODUCTION Measuring and specifying phase noise has become inereasingly important as phase noise is the limiting facior in many RF and microwave systems,like Doppler radar and space telemetry sy.te-. or communicationlinks. The complexity and subtleties of the 1nuurur.-unt have earned it the reputation of being more art than science. This paper gives an introduction to phase noise measurementfocusing on the two most and useful techniques,the "Two Source Phase Detector" and the "Delay Line "o--o1 Frequenc;- Discriminator" methods. System limitations are pointed 9ut a-swell as the potential ,our"". of erroneousdata when phase noise is measuredin a bench set up. The lt72gc Carrier Noise Test Set is HP's latest contribution in this field, designed to simplify and automate the complex task. A table of comparisons will show where the ttTigC and the various other HP solutions are optimally employed. O Why MeasurePhaseNoise? 1. Phasenoisecauses. . . environment in multi-signal o Lowerreceiversensitivity Transmiller H q- DopplerSignal DecorrelatedClutler Noise J_ o Glutternoisein DopplerRadarSystems r Phaseerrorsin digitalcommunicationsystems in the presenceof a phasenoiseimposesfundamentallimitations wherevera weak signal is processed receiverlocal oscillatorare .tiorrg interferini rij""r. In the aboveexample,phasenoisesidebandsof the to thI Ifpioduct of the strong interfering sjsnll and cover up the weak wanted signal. ti"".i"*"a signal in this caseis A similar situatio;exists with cohereit Dopplerradar. The strong interferring unwanted return of this p;;;;J ty reflectionsfrom large stationarl' f5;".ts. Phasenoise side bands signal' weak Doppler lien"i"r. ,iecorrelateJbvdelav ind potentially coveras clutter noisethe Phasenoise adds to b,,r"n in the world of d-igitaliata tiansmission phasenoise is a limiting faetor. the carrier or data clock o*,"r^il system rroir. in.rJ"sing Bit Error Rate and may causea cycle slip in recovery. 5 2. of PhaseNoise BasicRepresentation 2.1 In the Time Domain: A ,0(t) v(t): Signatwithrandomphasefluctuation v(t): Vscos f?rtsl+ J O (t)l DisPlaY OscilloscoPe V(t) = Vs cos 2rlsl ar <> ao(t) andPhasearerelatedbY Frequency f(r):## In the time domain phase noise can be observedas phase jitter of the signal on an oscilloscopedisplal' or a time interval counter may detect the instantaneous time fluctuations of the zero crossings. phase or frequencyfluctuations are the same physical phenomena.One can be derivedfrom the other as angular frequency is the first derivative of phase with respect to time. 6 o O,, In the FrequencyDomain: FJd(t):Jd(f) S4 SpectralDensityof PhaseFluctuations SO(f)-Jd2rms(f) f'3 F l i cke r FM I d2rms f' 2 Random W alk Phase t-t FlickerPhase fo White Phase | -€ FourierFrequencY s3 Fluctuations spectralDensityof Frequency : fzSO(f) Sr t(f) = Jf2rms(f): l2J @2rms(l) domain.Terms like Spectraldensity distributions describefrequencyor phasenoisein the frequency .,White,,',,Flicker,'and ..RandomWalk" refei to the slopeof f , f-t, and f-2 of thd densitydistribution. time domain rrluiiipri."tion by the Fourier frequency- conespond.rlsto differentiation in the noise. converti the spectraldensity of phasl noiie to the spectraldensity offrequency 7 2.3 Relationof PhaseNoiseto CarrierSidebandNoise offset The modulari6n pf the signal phase manifests itself in modulation sidebandsto the carrier at are levels sideband The rFourier frequencyr. rate frequenciesu.hich ar" muliiples of the modulation first onl5'the modulation, For small'angle terms. by Bessel relatedto the magnirucleof the phasedeviation b1' Besselterm is signrficanr and the relation betu'eenphasedeviation and sidebandlevel is approximated Forsmallanglemodulalion ( I ar. < l radian) ?P = J d2rms {-t d2rms(f) A common,though indirect.representationof phasenoiseis,7{fr,the ratio of the single sidebandnoise po$.er in a I Hz band*'ldth to the total carrier power specified at a given offset, f of the carrier' produce This representation is applicable only for very small phase deviations, sufficiently small to negligible higher order sidebands. Definitionol Y (t) y(tt=ry -+ fogl= | +-fott \ Refationbetween2,5 5, Sl f: lor -\6<< lradian st 8i $rb E! E: 1_ 11 Y (t)= iS o(t)=i -t, srf (t) €i E.b .E'E *l1 HzL 69 o o 3 PhaseNoiseMeasurementTechniques o DirectSpectrumAnalysis ) o Two Sourcesand PhaseDetector o FrequencyDiscriminator ) o DelayLine FrequencyDiscriminator o CarrierSuppression (OndriaBridge) o Heterodyne PeriodGounter From the multitude of techniquesto measurephaseorfrequency noisetwobasicmethods are singled out and explained in somedetail onthe following pages:the "Two Sourcesand Phase Detector" technique and the "Dela5' Line Frequency Discriminator" technique' Both methods are ielatively simple to implement and are broadband solutions.They also compliment each other well. The first method rnl.rur"s phase noise;high sensitivity is obtainable and two sourcesare needed. The second method measures frequency noise, has inherently lower sensitivity close in but no secondsourceis required. 3.1Two Sourcesand PhaseDetectorTechnique Source UnderTest lsolationAmPl. lv' PhaseDetector (DoubleBalancedMixer) lsolationAmPl. Calibration Allenuator t_ tl Low PassFiller ^Al i+ I Low Noise Amplifier , 1! F = F v L-- NarrowBand PhaseLock LooP sd (t) y (t) sd (t)=Jdimsltl=1/z Hz) JV273ns(1 (forJd<<lrad. V2grms forld<<1rad. = 9 $ l = 1 1 2s 6 ( l l 1 / q J V2rms(1 Hz) vzBrms generatesa signal With the LO and RF input in phasequadraturesteadystatewise,the mixer IF output phase detectorcoxstant the operation In linear sources. two tl" of phase difference oroportional to the is frequen-cv-offset. produced when Bource either signal beat sinusoidal the of ;;;i;;;;;rr. phaselock "orftg" Most sourc", ,nould-6riftout of quadratureover the periodof measurement.A narrow band ioicesthetwo signalsinto phasequadrature.Atrates lessthan theloopbandwidth the l";;;;t;atically phase fluctuations are ,ou-r.", now track each other; at rates higher than the loop bandwidt\ t-he unaffected.The two isolation amplifiers shouldpreventinjection locking of the Bources. phase noisemeasuredon the IF port representsthe nns-sumof the noisecontributionsof eachsource. Eoureeshould be either For definite data on the sourceut d"r ttst, the phase noise of the reference well characterized.In the absenceof any information on both sources,onecan at least state ", sourcenoise is worse than the measureddata. If 3 unknown sourcesare available, 3 ""gfigiUi. that neither accuratelyeach i""rui"rn"nts with B different sourcecombinationsyield sufficient data to calculate individual noiselevel. 10 o O 311 The Mixeras PhaseDetector Vg cos t,,11 Vs cos(-nt - 6(l)) v (t)= 15 cos(t,n - ,L) t. O(t)) + ffi. oiL)t * o(r))+... vB peak V (t) = VBpeakcos({rn - col1t. O(t)) KL = ConversionLoss VB peak= PeakVoltageof BeatSignal for c,rL= <,.,8 and d(t) = (k +1) 90" - Jd(t) JV (t) = (t) VB peaksin Jd(l) for ld<<l radian Jv (t) = (t) K6 -\d(t) I VoltsI K6=vBpeak L ,d J basedon one signal gating the The frequencyconversionof signals applied.toa.mixer |f qhvsically a multiplying process. represents case Either diodes. other or on the nonlinear conductanceoriite mixer frequencyof the input difference in the product resulting generates a Among many terms,ih" mixer also signals. (linearrelationbetweenRF and IF with a sinusoidalinput at the RF port and assuminglinearoperation VB peak voltage a with peak'The phaseslopeof this beat voltage),the beat sigrial is also ,it" wave " signal at zerocrossingsequals= VB peak-' t; trreiaf;6 hequencyand at a.phaseoffset of 90ophasefluctuationsJdrt) setting both inpui ft;;ir related.The mixer actsas resultin voltagefluclua"tionsJv(t rwhich for smalllhase devjationsarelinearly VB to equal K6 of peak' constant a phasedeteciorwith a phasedetector 11 3 . 1. 2 Ko = vB peak : fvg 76s Jv(f) =Kd r O(f) J d76s(f):;; 1 . JV 765(f)- S O(t): J d2rms(f)= 172 forJ 6<<1 g (t') o Calibrationwith BeatSignal 1 ffi-, -JV rms(f) JV2rms (1 Hz) v2Brms JV2 rms(l Hz) -: 1 / 4 e (l 1/2so(f) v'B ,r, J V 2 r m s ld B m Phasenoisespectrumin dBm correcledfor bandwidthand analyzer characteristics. -V2BrmsldBm Beatsignalin dBm. -Kel dBm -6 dB Atlenuation(if beatsignalwas attenuatedin calibration). Accounls lor rms valueof beat signal(3 dB) and conversionof s 6(f)roS (f) (3 dB). +2.5dB Accountsfor log amPlifierand of noise videoliltering(averaging) (if analogspectrumanalYzerls used). Assuming VR ,.ooL as the phasedetectorconstantdictateslinear operationof the mixer. The calibra' for iz tf ror Sl6rfrcan then be read from the aboveequations. ti""lilr,"."p"."trtifln"lyzer PM or FM Othercalibration methodsare basedon thllharacterization of the phaseslope,on calibrated signal. spurious sidebandson one of the two sourcesor on the injection of a calibrated 12 O 313 SystemNoiseFloor lVn MtX Jvn t.t.tr Low NoiseAmplifier t = system(f) JV2ntlttx +JV2nLNe v2g rms The noise floor of the measurement system is establishedby the equivalent amplifier and mixer noise tshot noise, flicher noiser at the IF port redueed by the phase detector gain K6 (VB peak for linear operation r.System sensitivity is therefore maximized by using a high level mixer, selecting a mixer u'ith low flicker noise and follou'ing the mixer with a low noise amplifier. -80 dB c 5 MHz- 1.6GHz Example:3047AMixer-Amplifer - SyslemNoisewlth High LevelDoubleBalanced Mixerand Low NoiseAmplifier .rN ob -100 F@ urg 9 - ^Y et -120 =\ urO O- gE -140 ACE -tr 6g \ U'E E -160 \ \ o -180 lHz - 1 0H z 1 00Hz 1 kHz 10kHz 100kHz l MHz f [HzJ t3 31.4 PotentialError Sources o Suppression or peakingol noisenearor belowthe phase lock loop bandwidth. o Inieclionlocking r Deviationfrom phasequadrature(resultingin an effectivelylowerK6) o Non-linearoperalionof mixer(if linearitywasassumed in calibration) o Distortionof RF-Signal(resultingin a deviationol K6 from VB peaf) o Saturationof the preamplilieror spectrumanalyzerin calibrationor by high spurioussignals(e.9.line spurious) o Changeof impedanceinterfacewhile going lrom calibrationto measurement e lnsuflicientcancelationof commonsourcenoisein a residualnoisetest r AM-noise(in particularAM noiseof the common sourcein a residualnoisetesl) o Closelyspacedspurious(e.9.line multiples) as Phasenoise misinterpreted o Noiseinjectedby peripheralcircuitry (powersupply, phaselock looP) (€.9., o Noiseinjectedby peripheralinstrumentatioh DVM) oscilloscope, o Microphonicnoise. o Technique Os: DelayLine FrequencyDiscriminator Source underTesl PowerAmPlifier Delay Line ( r) Srt(t) = lf2r|nr{f) tort=fi2JVrms = K62tr Jf *. Jvtrms (t Hz) O srt(f) Srt(l) 7 (f) derivedfrom Srt(f) :I (tl=*irs-rtt n =i + rorro<<1 JV"r. (1 Hz) 6"t- A delay line and a mixer operating as a phase detectorhave the combinedeffect of a frequency discriminator.The delayline transformsany frequencyfluctuationinto a phasefluctuationand the mixer with the L and R inputs at g0ooffsetlinearly convertsthe phasefluctuationsinto voltagefluctuationsat the IF port. As derivedon the next page,the equivalentfrequencydiscriminatorconstantKp is proportional to r and K6. = The frequen"yio phaseconversionhas a sin x/x characteristicwith a null at f 1/r. Measurementsare thereforelimited to f < L/Zt twith correctionl. The systemcan becalibratedby characterizingthe frequencyslopegt KOand ", or moreconvenientlyb5' .aaing fftl sidebandsto the sourc€under test ror a substitute)and determiningtheir level as reference level with an RF spectrumanalyzer. 15 s.2.1 Theoryof DelayLine FrequencyDiscriminator rd(r)( v(r)=Kd rd(r) Delay time r t(t) = lo +Jt sin 2;ft V(t) = Ve cos (2;fst - f .ot 2nltl FM Carrierwithsinusoidal fs = CarrierfrequencY | = FM rate Jf = FM deviation on PhaseDetectorJO(t) PhaseDifference 2rrl(r-r) - 2ilsl-f Jd(t) = Zrts(t - r)..*"ot Jd(t) = -2rrls' .,+ (cos 2rrl(t - rl - cos 2iltl "ot 2trlt JO(t)=- 2nlsr * 2f sinzrfr sinp'r 1t- 7\ = PhaseDetectorOutput:V(t) K6 -\@(t) = (2k+1)?(PhaseQuadrature): lor 21116r V(f) = KO2f sin z:'lrsin 2trt(r -il TransferFunction: Jv = Ko2{sin nlr= K62,..rlt # forr=fi: sin-zrfz = | tlr JV = K6 2nr Jl r_- Kp t#l = FrequencyDiscriminator Constant= K62nr o ()szz SystemNoiseFloor lVn urx JVn lxa L o w N o i s eA m p l i f i e r Srt(f ) system DelayLine !'r t' Psz (if used): Jdnr = ol RFAmplitier Noise Phase ffi 1t * f t Kd = VB peak= fiffiott Ko (in linearoperation): with L = ConversionLoss(Power)of Mixer noise.This also Mixer noiseand basebandamplifier noiseare interpretedby the systgmas frequency uv un nr amplifier.The amplifiermight have beenaddedto boostthe holdsfor phase "oir.l"tioJu."a signal level lowered b1'the lossy deial'line' ':{:l;#- JV2n..nl (X5Z--1' srt(t) 1(tt=++ sYstem , (#l' ('*f ,lJ.t Jr-l2pp (2;rl2 for J6<<1 - *f - rr+23r1 closein and Mixer, basebandamplifier and RF amplifiernoiseall have a flicker noisecharacteristic Ko u'hich reduced showwhite noisefarther out.The contribution. of mixer and basebandamplifier are !v emplol' a is to though' p"i"t. out the benefit of the RF amplifier as it maximizesKg'"A bet*:8:Itllon, il Jon-r" and Yith amplifier u.r*.-itr. signal split and eliminatethe needfor an RF amplifier ;;;;; ' f-2 and f-3 towards the with characteristic, noise L(f)increases white rndth"r, flicker S;;;;a:fn", but poor sensitivity as " carrier. This means sood sensitiyity far out (limited by the sinx/x response) comparedwith the phasedetectormethodclosein' N {(, 6 !{ -3 =\ ApproximaleSyslemNoise Floor of the DelaYLine DiscrlminalorTechnique 2Z'. uJ= .h l- ICE Ag cg Y! - r e G, o o t!t,**;"""G F 1 Hz 10 Hz 100Hz 1 kHz 10 kHz 17 is to lump A different*.a5.ofesrimatingthe s)'stemnoisefloor of the delal'line discriminatortechnique line delal' The perturbation J02r-tor "rrr' phase noise all noise contributions into an equivalent of a factor phase by noise equivalent this enhancei discriminator.interpreting Jc) as frequency rrbi.u | /2;f ;. Example: = -160 dBc/Hz' Phase noise of the mixer at i kHz ! t 't kHz' noise floor is system with a 100 nsec dela5'time the resulting J2=!t l| i l' ' . 2rf; 'Hz = -160 dBc + 61 dB = -96 dBc -100 -90 -80 -70 -60 9 rttl e, tn -50 Losl *40 -30 Enhancemenlof equivalentPhasenoise perturbationbYdelaY -20 t,tll=gxl=rl l " r ?( r )g r ( t ) \ 2 ; f I rti ;r +10 0 100Hz 18 O with Resonator s.2.3 FrequencyDiscriminator Low Noise plifier t-r[rro,r'h 56(l) for t '* CavityResonalor CrystalFiller Bandwidth InsideResonator 1t. $ I' o ddl E q u i va l e nDt e l a yT i me r =Eilfo=r[ KD= K52;r =*ri} ,?a Jvrms(f)= Kdfi rttmsltl Although not a broadbandsolution,it can be advantageousto replacethe delay line with a cavity resonatoior crystal filter. Groupdelay td6/du\ producesan equivalentr of Q/ rfo for Fourierfrequencies less than half the bandwidth. Inside fo/zq, the systemthereforemeasuresthe spectraldensity of frequencynoise*'ith a discriminatorconstantof K6 2Q'/fo. Bandwidth OutsideResonator 1f' $ t, Jvrms(Q= Kd Jdrms(f) Outside fot}Q,the system ma]' be vieg'ed as 2 signals applied to a phase ddtector u'ith the sideband noise of one"signalstripped off by the cavity resonator.The system then measuresJd216s with a phase detectorconstant K6. 19 Solutions 4 HP PhaseNoiseMeasurement o HP 11729C/8662A PhaseNoiseTestSystem (phasedetectormode) o HP 11729CCarrierNoiseTestSet (delayline discriminatormode) o HP 3047APhaseNoiseMeasurement System o HP 11740AMicrowavePhaseNoise System Measurement Following is a comparisonof HP phase noise measurementsolutions with particular focus on the Carrier NoiseTest Set,HP's most recentcontributionto phasenoisemeasurement. LLTZ1C 20 o 4.1 PhaseNoiseTestSystem HP 11729C/8662A Principleol Operation (PhaseDetectorMethod) H P 11729CCarrierNoiseTesl Set r I ill*Li 640 MHz Low Noise Reference Signal Microwave Source UnderTest 5- 1280M Hz I Generation Selection .01-18 GHz al lF-Amp I HP 8662A Synthesized SignalGenerator Opl.003 x0 DC FM EFC Inpul Inpul Low Noise I Amp Speclrum Analyzer 5-1280 MHz R 1 0H z 1 0M H z I I I I I r I I ,l l l Conlroller I I .J L J astateof'theartlownoieereferenesignalat640MHz The HP 86624Synthesizedsignal Generatorprovides is applied to a steprecoverydiodemultiplier which gensignal to the Hp 11Z2gCCarrier NoiseTest Set.This 18 GHz. A switchedin filter selectsan_appropriate beyond ranging MHz eratesa combof signals spacedby 640 an lFrange of 5 themicrowavesourceutdertestinto converts down combline.This microwavireferencesignal to 1280MHz. The doublebalancedmixer phasecomparesthe IF signalwith the outputsignal o{the HP 8662A(.01to 1280 MHz). phase quadratureis enforcedby phaselocking the HP 8662Ato the signal rurdertest either thru the limited electroniccontrol of the HP 8662A's10 MHz crystal oseillator or thnr its DGFM port. The loop band' widths can be set over a 4 decaderange. A firstorder phaselock loopis enabledbythecapturefeaturemakinglock acquisitioneasy.Then thesecond To enablephasenoise orderphaselock loop (with very high dc gain) maintains lock and phasequadr-ature. via the looptest be characterized phase can noise suppression the Larrd*idth, within the loop -"".u".-.nts ports. 21 N N HP 11729CCarrierNoiseTest Sel I 1 r I r+_ - r--1 i o<---o_n= ^,n*:JOLi 2{- ^"-ilAfj r lI I - --'l ' 't I----1---T--0ri_i8GHr Microwavc Sortrco Undor Tesl Hi'I"'Tti i IHIHH tfgl^::;-;;; combGenerator r---1 (@: ! i , u L - - + - J :I- f I lt+-t llII"::: ltooi."ii cED (w!E PowerAmP rr-amoV L -l- J l*i I I I 6a0 mHt SAtt lillcr I I 640 MH2 Inpul 5-1280 MHz I 40 dB LNA I I lo"p I resl I '1.5MHr I l0 Hr-l l0 MHzl HP 3582A/356t4 Low Frequency Speclrum Analyrer Loop Conlrol Loop Conlrol Oulpul DC'FM Aur. OulPul HP 98364 DG3kTop Compuler 40 MHz Xlal lo MHz Slal osc' I I 160 MHz sral 640 MHz Low Noise Relerence Signal I I I I I L- o 0ll0 - 20 MHz I Hz Sleps 1 2 8 0M H z O r r l P t t l' | .l Hr SlePs J o 0 dBm 04 PhaseNoiseTestSystem 11 NoiseFloorof HP 11729C/8662A (PhaseDetectorMethod) N I o o -24 E -40 I F -60 \ tr tr, g E tr o o F llJ L o z \ \ I \ -80 ;2in DC FM \-86! \ \ \ I -100 \ \ 12A \ 640 MHz ReferenceSignal ' from HP 8662A tl -140 $ytt"m Noise@ 10 GHz \ ll\- t-.a \ \ \ \ ltJ o rlpicalHpr'rz2eclaeize \ \ -,---\_,r -160 -- --- I o. o -180 o @ Hz 1 0H z 1 0 0H z 1 k H z 10 kHz 100kHz Oflsetfrom Carrier | [Hz] ts Js J1 / s y s t e m= 1 0 l o g ( N zr ' l Q r o- 1 0 1 0- 1 0 1 0 ) N = Multiplicalionnumber of the 640 MHz signal / r = Absolule SSB PhaseNoise of lhe 640 MHz reference signal (dBclHz). ? a = AbsoluteSSB PhaseNoise of lhe 5 lo 640 (1280)MH: lunable signal (dBc/Hz). ,lg = Two-porl noise of HP 11729C(dBc/Hr). Ollset flom carrier Typical 640llHz low noise oulput lHz 10Hz 100H2 1 kHz 10 kHz 100 kHz 1 IlHz -54 dBc -94 dBc -114 rlBc -126 dBc -149 dBc -159 dBc -159 dBc MHz 1 0M H z Typical Typical Typical 320-640 Two-porl Two-porl llHz tunnolseof noise ol ableE662A HP 11729CH P 1 1 7 2 9 C oulput @ 5 G H z @ 10 GHz -64 dBc -99 dBc - 9 3 d B c -94 dBc -112 dBc -106 dBc -1ltf dBc -125 dBc - 1 1 9 d B c -125 dBc -132 dBc - 1 2 6 d B c -136 d8c -137 dBc - 1 3 1 d B c -136 rlBc -1t18rlBc - 1 4 2 d B c -146 dBc -153 dBc -147 dBc For most microwaveapplications,11, the SSB phase noise of the 640 MHz referencesignal, is the dominatingfactorin settingthe noisefloorof the systemas its phasenoiseis multipliedup to microwave frequencies. 23 o Mode HP 11729COperatingin Discriminator 4.1.2 --l -I I Calibration source f- I t Signalal Carrierlreq. or lF wilh I CalibratedFM Sidebands I I I L- H P 1 1 7 2 9 CC a r r i e rN o t s eT e s l S e t fcalibration , CombLine Comb Generalion Section Calibralion b -6'rl MicrowaveSource u n d e rT e s t i I o.o1to 18GHz lF-Amp ExternalDelay Line 5 l o 1 2 8 0M H z Baseband Spectrum Analyzer iox.toI 'i"l I t_ a a o o Low noisedown convercionof microwavesignalunderlest Delaylineoperatingat lF (5-1500 MHz) Calibrationsignalprovidedby HP 8662A PhasequadratureIndication. The HP LL729Cmay also be usedin a delay line discriminator mode.The microwave referencesignal generatedby the 640MHz signal down convertsthe Bourceunder test into the IF range.The powersplitter ind auxiliaiy IF output allows one to simply ineert an appropriate delay line ext*nally. Delayingt-hesignal at IF meanslower delayline lossesand alsoa wider choiceof delaydevices,like e.g. SAW d;lav line. Any signal generator with FM sidebandsat IF or RF can Eenleas a calibration tool. 24 4.1.3 of HP 11729C Sensitivity Mode) Discriminator (DelayLineFrequency -20 o o ! -40 1OOns t Delay o G cg -60 \ C) (! o o o ,9 o z o o G 3 r @ o o -'*r "^J./ -,..11Hz \..' usclllalor -trof -ruoL TypicalFree'Running - -y'Source at 10 GHz I 10Hz at 10GHz I \ \.\ \\\ \.- tOOHz 1 kHz 1OkHz lOOkHz Offset From Carrier(Hz) L.-..-. 1 MHz 10MHz PhaseNoiseMeasurement on PulsedRFSignals 4.2.1 P u l s e dR F S o u r c e( D U T ) I I I Anti-aliasingFiller J t- A PulseDrive NarrowBand PhaseLock LooP The basebandsignal ofa pulsedRF signal phaselockedto a referencesourceequalsthe basebandsignal (0 is of a. ut pulsedRf signal jated by the pulseenvelope(seeFig. A, B, C). The CW noisespectrumJ multiplied by the envelopefunction of the pulse, h(t)= +**3, (-t)n n n nt Sln -It n2'rt cos -T\ Observedin the frequencydomain,this gating processmeansconvolvingthe two spectrato obtain the pulsedbasebandsp""tto- (seeFig. C). The resulting spectrumhas J (f) folded around multiples of the part of this spectrumis yieldedby the DC tcrm of the pulse iuise repetitionfrequency(PRF).The relevant it"t.ttshowsthattheleveloftheunpulsedphasenoisespectrum.c(0isreducedbythe enrrelopefunction, unPursrd rqu"r.'J the duty cycle. J $1,*o= (+)' Jl3por""a A lowpass filter, the antialiasing filter, with a corner frequency of <PRF/2 needsto be employedto eliminaie all the aliasing products (SeeFig. D). It also sen'eeto filter any phaeetransients associated ;th lh; pulse modulatoi which otherwise would limit the low noiee amplifier. ftr. ,vrt"- sensitivity decreaseswith decreasingduty cycle.A sampleand hold circuit in the baseband iri"en by the puisesignal,wouldimprovetheeystemaensitivityforlowduty cvcle!.Aeimplerway "ir.uit, to reducesystemnoisein the pulse off period is to gate either the LO or the R drive of the mixer'phase deteetor. A side effect of phaselocking to a pulsedRF signal is the reduction of the effectiveloop bandwidth it becomesincreasingly important to have a well balancedmixer-phase by t/T, For duty cycles <.:|ttu, detector. 26 o F r e q u e n c yD o m a i n Time Domain -+l PRF F i g A : P u l s eE n v e l o p e zPar Speclrumof PulseEnveloPe t c A c G s F i g B : U n p u l s e dB a s e b a n dS i g n a l ( U n P u l s e dP h a s eN o i s e ) UnpulsedPhaseNoiseSpectrum t G g A ! g G G o !ut pulsed l\/l\ -\ ,\ -2 Fig C: PulsedBasebandSignal beforeLow Pass \ PulsedPhaseNoiseSpeclrum belore Low Pass t G o A E 3 G a o G a Fig D: PulsedBasebandSignal afler Low Pass PulsedPhaseNoiseSpectrum aller Low Pass 4.2.2. PulsedPhaseNoiseMeasurements (PhaseDetectorMode) withthe HP 11729C LPF ffi HP 11729C Noise Spectrum Output 640 Input 5 to 1280 lnput PLL Gontrol Voltage Output Selectingpulse mode (phasenoise) on the HP 11729Capplies an error cunent to the mixer/phase detector and switches in a user-suppliedlow pass filter to remove the pulse repetition frequency feedthrough. l 4.3 System HP 3047APhaseNoiseMeasurement ExamPle: Apptication on lrequencydivider phasenoisemeasurement Two-por4 HP 3047A Phase Noise llelcurement Syrlem HP 98il6A or9816A Derk Top ComPuter 125 Mllz 1 2 5M H z t_ Comparison Device representsa high sensitivity phasedetectorsystem The Hp 304?4.phase NoiseMeasurementsystem automatesthe calibration and measurement with controllableprr"". r""r, circuits. Extensive softwlre procedures ri--!r^-^ i^ ^- example o -acidrrql meas phasenoise noise measureresidual nhecp of a is.an ^oax-l^ ^f Determining the phasenoise of frequencydividers sum of rms the will measure svstem the and ment. Here,the phasenoiseof the 8662;. i;t;;i;;".J"a noiseof both dividers. LrI i lLrL Eri F[rt]F r P L E S ! , i i 5 E t 0l H Z +tstshr -3S -.ig -59 - E gF -i6 -BU -so t- . - r8g - ilBF. -ra6 -t30 -r{8 - l5Ei -r661-. -r79 lH= 29 4.4 MicrowavePhaseNoise HP 11740A System Measurement HP 11729CCarrierNoise Tesl Sel Deviceunder Test ' 1 0 . 4G H z G u n nV C O a l HP 3(X7A Spectrum AnalYzerSYslem HP 9836A or 9815A Desk Top GomPuler 800 MHz 540 MHz RelerenceSignal providesa broadbandand In the Hp 11?40AMicrowaveNcise MeasurementSystem,the HP ll7290 source' comparable -- noisemicro*'avereferencesignal eliminating the needfor a eecond lo*, with either the 11729C HP of the The Hp gOaZephasedetectorsystemthen comparesthe IF output IF eource. b-12g0 MHz outfut of the HP 8662A or any other suitable and tunable -\ ---T --T \l- -\ ''i". :? 30 ''rl . (. ) lek .dBc./P.z r tez< v's { !:12-l Ir'! reY ?er. Comparisonof SystemSensitivities Iou -20 -40 l tP 11729C N E (, tr ;AW@\ C -60 t,nz 00ns elay a\ l, \ o G E L o L L- -100 G o \ o o -120 I 0 2 o o G a -140 o, \ \\ -HP 117: l9c/8662A @ 1 0 cHz t.\ \ 'a F. \ t \ o o o = \ * l--i \ -rHP 3047A@ 10 GHz* \ -:a;= -160 -180 iT; 1 0H z 100Hz 1 z 10 kHz 100kHz 1 MHz 10 M H z Otfsetfrom Carrier(Hz) .Performanceshownrequiresa second,usersuppliedsource of comparablePhasenoise. 31 SystemsComparison 46 PhaseNoiseMeasurement l- - ;;;^- I H P 8 5 6 6 AE HP 35854 HP3582A S p e c t r u mA n a l Y z e r HP 11729C C a r r i e rN o i s e TestSel HP 35854 Speclrum Analyzer HP 3585A Spectrum AnalYzer HP 3582A Spectrum Analyzer HP 3582A Speclrum AnalYzer HP 35601A S.A. lnlerface HP 35601A S.A. Inlerface - J--;,;;; I I HP 8562A Synlhesized Sig Gen Reference HP 8662A xp seesa I HP 8340A ] xe aozza i_ _:':3: i--I"r''Gl:l t Principle ol Operalion SourceUnderTesl /a\./ @-' _ iI Low Noise Microwave Relerence Signal sigGen HPrs36A-1816A l ,836 -rst6^ I l-rP SourceUnclerTest HP 8556A/8 HP 8568A'B HP 3561A Spectrum Analyzer li .l f "*'"-' I Source Under Tesl Reference Relerence 1 0M H z- 1 8 G H z 5 M H z- 1 8 G H z 5 llHz -'18 GHz 10 MHa - 18 GHz yes no yes yes - 123dBc at 10kHz - 145dBc noisefloot (withHP 1r729C) -1't2 dBc al 10 kHz - 145 dBc noise floor 123dBc al 10 kHz 145 dBc noise lloor - 170dBc al 10 kHz' - 170 ctBcnoise floor Ollsel FrequencY : 1 0M H z <40 MHz <10 llHr (<40 MHz) <10 MHz no yes yes no Aftl-Noise (Miuowave) yes w/erlemal deleclor yes yes Manual Operation yes no yG8 yes Aulomalic Op€relion yes yes te8 yes user provided €xlensive crlemive ur€r piovided (HP-rB) .Specification shown reguires a second. user suPplied source of comparable phas€ noise. i i I i Two-pod Notse 32 ti I Typical Syslem Noiie al 10 GHr Sotlware I %G \r; 6_ Reference Frequency Range HP11729C CarrreN r oise T e s lS e l t-'- - - - - -' f .*..r^ "*t.*l I Synthesized I I _l I Source Under Test (trequencY discriminalofmodej HP 11729C Canler Noise Tesl Set I I HP11729C CarrierNoise Tesl Set HP 11740A Microwave P h a s eN o i s e Measuremenl Syslem H P 1 1 7 2 9 C / 8 6 6 2 A HP 3047A PhaseNoise PhaseNoise Measurement Tesl System System t F h a ! ' d € l e c t o rm o o ( ) GLOSSARYOF SYMBOLS Availablenoisepowerin bandwidthB CarrierlrequencY CornerlrequencYof flickernoise Fourierfrequency(sideband-,olfset-,modulation' requency) baseband-f f(t) IHzJ InstantaneouslreguencY frequencylluctuation Instantaneous rf(t) IHzl K6 Ivolt/radJPhasedetectorconstant Kp lvolVHzl Frequencydiscriminatorconstant [dBc/HzJ Ratioof singlesidebandphasenoiseto total signal 9ttt powerin a 1 Hz bandwidthI Hertzfrom the carrier Signalpower Pg lmwl Powerof singlesideband Pssb lmwl Qualityfactorof resonator o srt(f) tllzzttlz! Spectraldensityof lrequencyfluctuations s6(f) lrad2lHzl Speclraldensityof phasefluctuations t Isec] Time voltage Instantaneous Y(t) Ivoltl vB peak IvoltJ Peakvoltageof sinusoldalbeatsignal Inslantaneous Phasefluctuation a0(r) lradl InstantaneousPhasePerlurbance Ao(t) lradl Delaytime T lsecl FKTB fe fs I lmwl IHz] lHzl lHzl REFERENCES 1r Transmilters."IEEE Transactiorts of Noisein Microu'ave ctn Ashler'.J.R..et al. "The l\leasurement pp. MTT-25, 29a'318, Technigues, Vol. No. 4, tApril, 19ii t. Mierouat'cTheorl'and 0r for MeasuringSpectralPurity", Microu'aves, Vol. 16. Fischer.M.."An Overvieu'of ModernTechniques No. 7, pp.66-75, July 1979. J} Hewlett-Packardstaff. "Understandingand MeasuringPhaseNoise in the FrequencyDomain," ApplicationNote207,October1976. staff,"MeasuringPhaseNoisewith the HP 35854SpectrumAnalyzer",Application 4 ) Hewlett-Packard IMay1981. Note246-2, D) of Low PhaseNoiseSignalsUsing the8662A staff,"Applicationsand Measurements Hewlett-Packard SignalGenerator",ApplicationNote 283'1,November1981. Synthesized A Tutorial Introduction",NBS Technical 6 t Howe,D.A.,"Frequenc]'DomainStability Measurements: Note679,March 1976. 1t Lance,A.L., Seal,\\'.D.,Hudson,N.W.,Mendoza,F.G.,and Donald Halford."PhaseNoiseMeasureAnalysis," Conferenceon PrecisionElectromagneticMeasurements. ments Using Cross-Spectrum Ottawa,Canada,June 1978. 8 t Ondria F.G.,"A lMicrowaveSystemfor Measurementsof AM and FM NoiseSpectra",IEEE TransacVol. MTT-16,pp.767-781, September1968. tionson Microu'aueTheorl'and Techniques, 9 t Scherer,Dieter,"Design Principlesand Test Methodsfor l,ow PhaseNoiseRF and MicrowaveSour- ces",Hewlett-PackardRF and Microu'aveMeasurementSymposium,October1978.(Also editedverApril sion in Microwaves,"Today'sLesson- Learn aboutLow-NoiseDesign",part 1, pp. 116-122, May 1979r. 1979,part 2,pp.72-77, l 0 ) Shoal J.H., Halford, D., Risley,A.S., "FrequencyStability Specificationand Measurement:High Frequencyand MicrowaveSignals", NBS TechnicalNote 632,January 1973. Technique- Conceptsand Implementation", l l t Temple,R., "Choosinga PhaseNoiseMeasurement Hewlett-PackardRF and MicrowaveMeasurementSymposium,February 1983. 34 HP Archive This vintage Hewlett-Packard document was preserved and distributed by www.hparchive.com Please visit us on the web! The HP Archive thanks George Pontis for his contribution of this material. On-line curator: John Miles, KE5FX jmiles@pop.net