Noise-FigurePPT优秀课件.ppt

Noise Figure FundamentalsNoise Figure Fundamentals•Why is Noise Important?•Common Types of Noise•Available Noise Power•What is Noise Figure?•Noise and Temperature•Noise Measurement•Y-factor Technique•Cold Noise Method•Difficulties in Measuring Noise Figure•SummaryRF UniversityNoise Figure Fundamentals2 Receiver SystemSmall SignalImperfect AmplifierSignal Larger, But Noiser.Why is Noise Important?RF UniversityNoise Figure Fundamentals3 NameExampleDescriptionImpulseIgnition, TVI Not Random, cure byShielding, etc.Quantizing,Decoding, etcBERDigital Systems, DAC’s &ADC’s. Often Bit Resolutionand/or Bit FidelityShotTransistorsCorpuscular Current Flow,Lots of ImpulsesThermalResistors,AtmosphereThermal Agitation ofElectrons, AC Like SignalCommon Types of NoiseRF UniversityNoise Figure Fundamentals4 R - jXR + jXLLAt Standard Temperature T0 (290K): kTO = 4 x 10-21 W/Hz = 4 x 10-18 mW/Hz = -174 dBm/Hzk = Boltzmann's constant (1.38 x 10-23 Joule/K)T = Temperature (K)B = Bandwidth (Hz)Power Delivered to a Conjugate Load, i.e. R = RL, X = XL P = kTB =av.Available Noise Power (Thermal)RF UniversityNoise Figure Fundamentals5 (Original Definition)(S/N)in(S/N)outF(dB) =(S/N)in(S/N)outTS = T0 = 290K10 log.nf0030What is Noise Figure?TS = source temperatureRF UniversityNoise Figure Fundamentals6 An Equivalent Definitionof Noise Figure (IRE)NoutNinGaNoise addedNaf1f2.inoutGa,NaAn Equivalent Definition of Noise Figure (IRE) =Na+ (kTSB)Ga(kTSB)GaTS = T0 = 290 KF =Na + NinGaNinGaTS=T0Ga = amplifier gainNa = noise added by the amplifier F = (S/N)in = Sin Nout = 1 Nout (S/N)out Sout Nin Ga NinNoise Figure:RF UniversityNoise Figure Fundamentals7Partitioning the Measured Total Noise FigureF1 = DUT contribution F2 = receiver contributionF12 = total noise figureNaNoZS @ TO = 290K RecNin Ga * NinGaF12 = F1 + (F2 - 1)GaRF UniversityNoise Figure Fundamentals8.NaNHoutNoise Output PowerTSSLOPE kGaB0Nout NinGaF = NCT1T2Na, GaZS @ TSAddedNoiseN out = Na +Ga k(Ts )B{Input Noise, NinNinNoise Power is Linear with Temperature{RF UniversityNoise Figure Fundamentals9 .Noise Output PowerNaTSTeoutNa=0ZS@ TSZS@ TeNout = kGaB(Te+Ts)NaZS @ TSN out = Na + kGaBTSDefinition of Effective Input Noise Temperature, TeRF UniversityNoise Figure Fundamentals10How The Measurement is Made1) Measure Gain (small signal available gain)2) Measure Noise (available power)DUTMUXADCLNADUTRF UniversityNoise Figure Fundamentals11Y Factor Method: The Old Way•Measures incorrect gain (transducer vs. available)•Mishandles mismatch between DUT and system•Ignores receiver noise changing with DUT match3) Solve for noise power & gainMUXADCLNADUT1) Measure noise with ‘cold’ noise sourceMUXADCLNADUT+27 V2) Measure noise with ‘hot’ noise sourceRF UniversityNoise Figure Fundamentals12Cold Noise Technique•Vector Network Analyzer•High speed computer workstation•Noise figure calibrationThis is the method used by the Agilent 84000 RFIC Tester!RF UniversityNoise Figure Fundamentals13Measurement Process: Calibration Phase 1. RF source Power Calibration2. 2-Port S-Parameter Calibration3. Characterize the Cold Reference Termination4. Noise Figure Receiver Calibration5. Receiver Noise Parameter Characterization6. All the appropriate Calibration Data is savedRF UniversityNoise Figure Fundamentals14Measurement Process: After Insertion of DUT7. S-Parameter measurement of DUT8. Receiver/DUT Noise Figure measurement9. Calculate the DUT Noise FigureRF UniversityNoise Figure Fundamentals15Cold Noise vs. Y-Factor Techniques•The Y-factor technique is predicted to be more accurate for a well matched DUT. •With the Cold Noise method, the uncertainties of the many measurements result in larger overall calculated uncertainty, even after error correction.•However, the Cold Noise uncertainty is not affected as strongly by increasingly poor DUT match.RF UniversityNoise Figure Fundamentals16Error Correction: Removing the Receiver NoiseThe receiver noise figure will change as the attenuator is changedNo reflectionHigh reflectionMUXADCLNASystem ReceiverIdealVariableAttenuatorShortCircuitRF UniversityNoise Figure Fundamentals17System LimitationsMatch atall pointsADC noiseNoise figureGain stability, noise figure, match sensitivityNoise temperature of terminationAccuracy of noise sourceMUXLNADUTADCRF UniversityNoise Figure Fundamentals18What System Error Dominates?The gain of the device is the single most dominant factor that determines accuracy. If the device has a lot of gain, the output noise is lifted above the receiver noise, reducing sensitivity to most systematic errors. Error in Noise figure due to 1% (.04 dB) receiver gain changeAccuracy in this region will be dominated by mismatch error and noise source calibration accuracyRF UniversityNoise Figure Fundamentals19•IF LO or spurs leak through, the system receiver will be compressed & distorted.•Noise on the LO can convert into the IF and artificially degrade the noise figure.MUXADCLNADUTMUXADCLNADUTLO FeedthruLO NoiseCompressed and distortedFrequency Translating Components: Interaction with DUT LORF UniversityNoise Figure Fundamentals20DUT LO Compressing ReceiverDUT LO leakage incident on the noise receiver can cause the detected noise to go down as receiver compresses.Detected noise goes down as receiver compressesRF UniversityNoise Figure Fundamentals21Removing the Fixture•The fixture (or the probes) insert loss & mismatch•Relatively easy to remove from gain & power•Very difficult to remove from noise measurement•Simple scalar correction won’t work•Simple noise cascade won’t work because of system receiver noise parameter extraction•Need full noise correlation analysis to remove fixture.RF UniversityNoise Figure Fundamentals22Receiver Must Dominate Noise BandwidthDigital receiver sets noise bandwidth during calibration.If DUT sets noise bandwidth during measurement, calibration will be invalid.MUXLNADUTADCRF UniversityNoise Figure Fundamentals23Measurement ExampleDUTInputFilterAmpOutputFilterLOFilterMixerLOIFRFVccRF UniversityNoise Figure Fundamentals24Measurement Results•Measurement Time: 250 ms•Std Deviation: 0.11 dBno Filteringon DUT LOFiltering on DUT LORF UniversityNoise Figure Fundamentals25Summary•Minimize Noise!•Most difficult error to correct: receiver noise•Dominant factor in determining accuracy: device gain•Remove fixture or probe affects!RF UniversityNoise Figure Fundamentals26Appendix A: An Equivalent Definition of Noise Figure (IRE)F1 = Sin /Nin = NoSinSo /NoNinSo= No1where So= GaNin GaSin= NinGa + Nawhere No= NinGa + Na NinGaF1= 1 + NaNinGaRF UniversityNoise Figure Fundamentals27F12= NoNinGaGr=Nin GaGr + NaGr + Nrwhere No= (NinGa + Na)Gr + NrNinGaGr=1+ Na+ NrNinGa NinGaGr=F1+ Nrwhere F1= 1 + Na(from previous slide) NinGaGr NinGa= F1+ (Nr) 1(NinGr) Ga=F1+F2-1where F2=1 + Nr F2 - 1 = Nr GaNinGrNinGrAppendix B: Partitioning the Measured Total Noise FigureRF UniversityNoise Figure Fundamentals28个人观点供参考，欢迎讨论。