Noise - A stench to the ear, Undomesticated music, The chief product and authenticating sign of civilization Ambrose Bierce - “The Devils Dictionary” 1907 Craig Nelson National Institute of Standards and Technology 325 Broadway Boulder, CO 80303 Email : craig.nelson@boulder.nist.gov Newport San Francisco Beach201 2010 1 1 Topics • • • • • • San Francisco 2011 Introduction and Review Noise Types Measurement Methods Measurement Calibration Common Measurement Problems � Conclusions 2 Weather Monitoring RADAR Satellite Communication INDUSTRY and MILITARY NEEDS for SPECTRAL PURITY Imaging RADAR •Navigation •Defense & Homeland Security •Secure Communication •Astronomy & Geodesy San Francisco 2011 Noise Metrology Atomic Frequency Standards & Spectroscopy 3 NOISE PM Noise AM Noise 2. 5 2 2 1. 5 1. 5 1 1 0. 5 0. 5 0 0 0 90 180 270 360 450 540 -0. 5 0 90 180 270 360 450 540 -0. 5 -1 -1 -1. 5 -1. 5 -2 -2 -2. 5 60 60 50 50 40 40 30 30 20 20 10 10 0 0 1 San Francisco 2011 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235 244 253 1 10 19 28 37 46 55 64 73 82 91 100 109 118 127 136 145 154 163 172 181 190 199 208 217 226 235 244 253 4 Difficulty of using a Spectrum Analyzer • IF bandwidth too wide • SA internal reference too noisy • Not enough dynamic range • Phase noise is often below -170 dBc • Cannot distinguish between AM and PM noise San Francisco 2011 5 Power Single Sideband Modulation Frequency San Francisco 2011 6 Power Amplitude Modulation Frequency San Francisco 2011 7 Power Phase Modulation Frequency San Francisco 2011 8 San Francisco 2011 9 Theory Review Basic Model for Noisy Signal V (t ) = A (1 + α (t )) cos(2π υ 0 t + φ (t )) where: A = average amplitude α (t ) = fractional amplitude fluctuations υ0 = average frequency φ (t ) = phase flucuations San Francisco 2011 10 Basic Definitions V (t ) = A(1 + α (t )) cos(2πυ0t + φ (t )) phase = 2πυ0t + φ (t) d ω (t ) = [ phase] dt 1 d 1 d υ (t) = [2πυ 0t + φ ((t)] t)] = υ0 + φ ((tt) ) 2π dt 2π dt Fractional frequency deviation υ (t ) − υ 0 1 d y (t ) = = φ (t ) υ0 2πυ 0 dt 2 1 YT ( f ) 2 S y ( f ) = P S D [ y ( t )] = T San Francisco 2011 0< f < ∞ 1 [ ] Hz 11 Definition of Phase Noise 2 2 1 Sφ ( f ) = PSD (φ (t)) = Φ T ( f ) T 0< f <∞ rad 2 [ ] Hz Single sideband phase noise 2 L( f ) ≡ 1 2 Sφ ( f ) [dBc / Hz] ⎛ υ0 ⎞ Sφ ( f ) = ⎜ ⎟ S y ( f ) ⎝ f ⎠ Definition of Amplitude Noise 2 2 Sα ( f ) = PSD (α (t )) = ΑT ( f ) T 1 San Francisco 2011 0< f <∞ 1 [ ] Hz 12 Noise Types • Additive Noise f 0 • Thermal • Shot noise • Multiplicative Noise • Flicker f −1 • Higher order colored noise types San Francisco 2011 f −2 , f −3 , f −4 ... 13 Pow er Power Additive Noise Frequency San Francisco 2011 14 Pow er Power Additive Noise Frequency San Francisco 2011 15 P ower Power Pha se Noise Phase Additive Noise Frequency Fourier Frequency Since it is uncorrelated to the carrier, additive noise always appears as equal amount of AM and PM Noise. For Amplifiers: San Francisco 2011 kTB Sϕ ( f ) = S a ( f ) = NF = −174 + NF − PI @ 300 K PI 16 Pow er Power Multiplicative Noise Frequency San Francisco 2011 17 Pow er Power Multiplicative Noise Frequency San Francisco 2011 18 P ower Power Pha se Noise Phase Additive and Multiplicative Noise Frequency San Francisco 2011 Fourier Frequency 19 Additive and Multiplicative Noise 10 GHz, Gain=32.5dB, NF=1 -100 Pin=-43.3dBm Pin=-33.2dBm Pin=-29.2dBm Pin=-24.28 Pin=-21.22dBm, 1dB compression L(f) [dBc/ z] [dBc/H Hz] -110 -120 -130 -140 -150 -160 1E+2 1E+3 1E+4 1E+5 1E+6 Offset Frequency [Hz] San Francisco 2011 20 Noise Figure •Noise figure is only a figure of merit of thermal additive noise. •It has ZERO correlation to flicker or any other type of multiplicative noise •In the linear region of operation. The phase noise floor for a device can be determined from input power and noise figure as follows: Sϕ ( floor ) = Sα ( floor ) = L ( floor ) San Francisco 2011 kT0 BNF Pin = −174 + NF − Pin [dBrad 2 / Hz ] or [dB / Hz ] = −177 + NF − Pin [dBc / Hz ] 21 Multiplicative AM Noise Ic g m ≅ VITc Unlike additive noise, multiplicative AM and PM are not fundamentally correlated because they up convert by different mechanisms. San Francisco 2011 22 Multiplicative PM Noise Ic High Linearity => Low Multiplicative PM Noise (Flicker) Separation between P1dB and IP3 is a simple indicator > 15 dB has potential for low flicker San Francisco 2011 23 Typical PM Flicker Noise for Different Semiconductor Types � -110 X-Band GaAs AMP SSB Phase Noise [d [dBc/Hz] Bc/Hz] -120 -130 -140 X-Band HBT X-Band Schottky Mixer L-Band BJT HBT Amp -150 -160 -170 1E+0 HF-VHF BJT HF-UHF Schottky Mixer 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 Offset Frequency [Hz] San Francisco 2011 Courtesy of M. Driscoll 24 Noise Summary Additive Noise Summary • • • Noise power is uncorrelated to signal power AM = PM AM and PM noise levels var varyy inversel inverselyy with carrier p power ower Multiplicative Noise Summary • • • Noise power is correlated to signal power AM ≠ PM AM and PM noise levels are independent of carrier power � San Francisco 2011 25 Typical Noise Types • Passive Devices • Thermal f 0 • Some have flicker (magnetics, carbon resistors) f -1 • Higher order noise may come from temperature effects f -4 • Active Devices • Almost all have thermal and flicker • Possible temperature effects f -4 � -1 f 0 and f -1 � • Sources ( May some or all of the higher order types ) • • • • • White PM or Thermal Flicker PM White FM Flicker FM Random Walk San Francisco 2011 f 0 � f -1 � f -2 f -3 f -4 26 Noise Types f −4 −3 White frequency Flicker Frequency f −2 f 0 log[(σy)(f)] f log[(Sφ)(f)] Random Walk τ −1 τ τ −1 1 2 2 τ0 White phase f −1 Flicker phase log (Fourier Frequency) log (τ) White phase Flicker phase White frequency Thermal Noise, Shot Noise Electronics, recombination-generation, traps Resonator, integrated white phase Flicker Frequency Resonator, integrated flicker phase Random Walk Temperature, Shock, Vibration, Resonator San Francisco 2011 27 Leeson’s Effect Resonator Barkhausen Gain = 1 Phase Shift = 2 π n Phase Shifter San Francisco 2011 Amplifier 28 Noise Phase Noise Leeson’s Effect - Low Q 2 fc ⎛ f0 ⎞ ⎛ fc ⎞⎤ FkT ⎡ ⎢1 + + ⎜ Sϕ ( f ) = ⎟ ⎜1 + ⎟ ⎥ Pa ⎢ f ⎝ 2 fQL ⎠ ⎝ f ⎠⎥ ⎣ ⎦ f −3 Oscillator Noise f −1 Half f0 Resonator 2Q L Bandwidth f −2 f0 fc Amplifier Noise Fourier Frequency San Francisco 2011 29 Leeson’s Effect - High Q Half Resonator Bandwidth Phase Noise Noise f0 2QL 2 ⎡ fc ⎛ f0 ⎞ ⎛ fc ⎞⎤ FkT ⎢1 + + ⎜ Sϕ ( f ) = ⎟ ⎜1 + ⎟ ⎥ Pa ⎢ f ⎝ 2 fQL ⎠ ⎝ f ⎠⎥ ⎣ ⎦ f −3 f −1 Oscillator Noise fc f0 Amplifier Noise Fourier Frequency San Francisco 2011 30 Four Sources at Different Frequencies and Similar Power � 0 SSB Phase Noise [[d Bc/Hz] dBc/Hz] -20 10 GHz DRO -40 -60 -80 -100 -120 -140 -160 -180 1E+0 500 MHz SAW 100 MHz OCXO 5 MHz OCXO 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 Offset Frequency [Hz] San Francisco 2011 31 All Sources Normalized to 10 GHz 0 10 GHz DRO SSB Phase Noise [[d dBc/Hz] Bc/Hz] -20 500 MHz SAW -40 -60 100 MHz OCXO -80 -100 -120 5 MHz OCXO -140 -160 -180 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 Offset Frequency [Hz] San Francisco 2011 32 Composite Phase Noise Synthesis 0 10 GHz DRO SSB Phase Noise [[d dBc/Hz] Bc/Hz] -20 500 MHz SAW -40 -60 100 MHz OCXO -80 -100 -120 5 MHz OCXO -140 -160 -180 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 Offset Frequency [Hz] San Francisco 2011 33 Effects of Frequency Manipulation on Phase Noise Translation or Mixing S Aφ ( f ) S B San Francisco 2011 φ (f) S SMφ( f ) A φ (f)+S • • • B φ (f )+ S M φ (f) Residual Noise AM to PM Conversion Noise Folding 34 Frequency Multiplication v2 (t ) = cos[ N(ωt + ϕ (t )] ω2 = Nω Phase Noi se Noise 2 2 Sφ 2 ( f ) = N Φ1,T ( f ) = N 2 Sφ ( f ) T 20 LogN dB Ideal Response Typical Response Multiplier Noise Floor Oscillator Noise San Francisco 2011 Fourier Frequency 35 Frequency Division 1 v2 (t ) = cos[ (ωt + ϕ (t )] N ω2 = ω N 2 Phase Nois e Noise Sϕ ( f ) 2 ΦT ( f ) Sφ 2 ( f ) = = T N N2 Oscillator Noise 20 LogN dB Typical Response Divider Noise Floor Ideal Response San Francisco 2011 Fourier Frequency 36 Extracting Noise Two noisy signals V1(t) = A1(1+α1(t))cos(ω1t +ϕ1(t)) V2 (t) = A2 (1+α2 (t))cos(ω2t +ϕ2 (t)) Multiply them together ⎡ A1 A2 (1 + α1 ( t ) + α 2 ( t ) + α1α 2 ( t )) ⎤ V1 ( t ) i V2 ( t ) = ⎢ ⎥ i 2 ⎣ ⎦ {cos ⎡⎣(ω + ω ) t + ϕ ( t ) + ϕ 1 San Francisco 2011 2 1 2 ( t )⎤⎦ + cos ⎡⎣(ω1 − ω2 ) t + ϕ1 ( t ) − ϕ2 ( t )⎤⎦} 37 Phase Noise Set ω = ω1 = ω2 and lowpass filter ⎡ A1 A2 (1+ α1 ( t ) + α 2 ( t ) + α1α 2 ( t )) ⎤ V1 ( t ) i V2⎡(At )A= (1+ ⎢ α1 ( t ) + α 2 ( t ) + α1α 2 ( t )) ⎤ ⎥ i 1 2 2 ⎣ ⎢ ⎥ cos ⎡⎣ϕ1⎦( t ) − ϕ2 ( t ) + φ3 ⎤⎦ 2 ⎣ {cos ⎣⎡(ω (2+ωω)t ) t + ϕ ( t ) + ϕ 1 When φ3 = 90° 2 1 ⎦ 2 sin x ≅ x, ( t )⎦⎤ + cos ⎣⎡(ω1 − ω2 ) t + ϕ1 ( t ) − ϕ2 ( t )⎦⎤} cos(x +90) ≅ x ⎡ A1 A2 (1+ α1 ( t ) +⎡ α A12A(2t ⎤) + α1α 2 ( t )) ⎤ cos ⎡φ t − φ t + 90⎤⎦ ⎢ ⎢ 2 2 ⎥ ⎡⎣φ1 ( t ) − φ2⎥ ( t ) ⎤⎦ ⎣ 1 ( ) 2 ( ) ⎣ ⎦ ⎦ ⎣ hLPF ∗ ⎡⎣V1 ( t ) iV2 ( t ) ⎤⎦ ≅ kd ⎣⎡ϕ1 ( t ) − ϕ2 ( t ) ⎤⎦ when ω1 = ω2 ∧ φ3 = 90° San Francisco 2011 38 Amplitude Noise ω1 = ω2 V1(t) =V2 (t) = A1[1+α1(t)]cos[ω1t +ϕ1(t)]= A2[1+α2 (t)]cos[ω2t +ϕ2 (t )] ⎡1 A ⎡ A1 A2A (12+(1α+ (2tα)(+1( tα+)22+(αtα)(2+t ()αt+1)α)α2 ( t()t))⎤) ⎤ co⎡s φ φ ⎢ ⎢ ⎥ ⎥ cos ⎣φ[1 (3t ]) − 23 ( t ) + φ3 ⎤⎦ 2 22 ⎦⎦ ⎣ ⎣ 2 When φ3 = 0° and 2 os(0) = 1 c Remove DC term and neglect higher order term 2 2 ⎡ A2 ⎤ A α 2 α t + t + A () ( ) ⎥ cos [0] ⎢ 2 ⎣ 2 ⎦ A2 hLPF ∗ ⎣⎡V ( t ) ⎤⎦ ≅ + kaα ( t ) when φ3 = 0° 2 2 San Francisco 2011 39 Phase Noise Measurement Types • Homedyne or Residual – Single source � • Noise Floor • Any two port device measurement • Heterodyne - Two source measurement � • Frequency Discriminator • Delay Line • Cavity Resonator • Digital Measurement Systems San Francisco 2011 40 Two Oscillator Measurement PLL φ Reference LPF IF AMP V(f) = Kd (f)δφ(f ) Source S φ m easu red ( f ) = S φT ( f ) + S φ R ( f ) + San Francisco 2011 P SD [V rm s syste m ( f )] Kd2( f ) 41 Noise Floor Measurement PLL φ LPF Reference Source �Phase noise IF AMP of source cancels [ Δ φ R ]2 ( f ) − [ Δ φ R ]2 ( f ) PSD[Vrmssystem ( f )] Sφ NoiseFloor ( f ) = + 2 Kd 2 ( f ) K d ( f ) BW 0 San Francisco 2011 42 Two Port Device Measurements DUT φ LPF IF AMP Reference DUT �Phase noise of source cancels �Two devices are needed if they have a long delay �Two devices are needed if they change the frequency San Francisco 2011 43 System Calibration • Static Phase shift (PM) • Kd or Beat Frequency Method (PM) � • Modulation • Single Sideband (AM/PM) • Phase Modulation (PM) • Frequency Modulation (PM) • Amplitude Modulation (AM) • Noise Standards (AM/PM) San Francisco 2011 44 Static Phase Shift •A known phase shift is introduced, and the corresponding voltage change measured •Adjustable phase shifter (mechanical or electrical) •Switched delay lines •Programmable phase shift in a synthesizer Kd V oltage C hang e = P hase S hift ⎡ V ⎤ ⎢⎣ rad ⎦⎥ P S D [V rm s ( f )] Sφ ( f ) = Kd2 Remember: L( f ) = 12 Sφ ( f ) [dBc / Hz ] San Francisco 2011 45 Two Oscillator Beat Frequency Calibration PLL φ Reference LPF Source San Francisco 2011 46 Calculating Mixer Sensitivity Kd 1.5 1.0 0.5 K d (v / rad ) = 0.0 Slope Slope -0.5 Slope(v / s ) ⋅ T ( s ) 2π ( rad ) -1.0 -1.5 0 1.57 3.14 Period T � 4.71 6.28 Typically 0.3 to 0.4 V/rad for a saturated double balanced Schottky diode mixer Make sure both positive and negative slopes are equal in magnitude and symmetric San Francisco 2011 47 Single Source Kd Calibration φ LPF Power Meter 50Ω Substitution Source San Francisco 2011 48 Mixer Sensitivity • Frequency • RF and LO power • Mixer termination at all three p ports orts • Cable lengths Calibration conditions must replicate the measurement conditions as closely as possible San Francisco 2011 49 Possible Errors Using Kd Calibration � Determines gain only at a single frequency Noise is suppressed inside PLL bandwidth Requires open loop PLL or substitution source � Beat Beat configuration configuration does does NOT match match actual � actual measurement configuration � • Injection Locking • • • • San Francisco 2011 50 SSB Calibration Method φ Δ Source Baseband f SA f f +Δ f +Δ San Francisco 2011 51 Single Sideband Modulation AM PM SSB n dB n-6 n-6dB dB = Generates equal amounts of AM and PM Noise Phase and amplitude modulation is detected at ½ the SSB ratio Sϕ ( f c ) ≅ Sα ( f c ) ≅ San Francisco 2011 Pν 0 + fc 2 Pν 0 ≅ + A SSB tone at n dB below the carrier creates phase and amplitude noise with Pν 0 − fc Sϕ ( f cal ) = n − 3dB 2 Pν 0 Sα ( f cal ) = n − 3dB or L ( f cal ) = n − 6dB 52 Additive Noise Calibration Method φ Baseband Reference f bw 2 Source f BPF San Francisco 2011 fc,bw 53 AM/PM Modulator • Can be adjusted for pure PM or AM modulation • Extremely flat frequency response • Calibrates Calibrates Kd(f) with with system locked locked • Can be used to find true quadrature for minimizing AM leakage San Francisco 2011 54 AM Modulator Calibration Port Diode Detector and Power Meter or Spectrum Analyzer Output -20dB -20dB MAXIMIZE detected AM signal By adjusting phase shifter Sα ( f c ) ≅ Pν 0 − fc + Pν 0 + fc Pν 0 φ RF IF Modulation Frequency LO -20dB Input San Francisco 2011 55 PM Modulator Output Monitor Port Diode Detector and Power Meter -20dB -20dB MINIMIZE detected AM signal By adjusting phase shifter Sϕ ( f c ) ≅ Pν 0 − fc + Pν 0 + fc Pν 0 φ RF IF Modulation Frequency LO -20dB Input San Francisco 2011 56 Tips for Measuring Gain vs. Fourier Frequency using Swept Modulation � • Measure power spectrum not PSD • Use flattop windows for FFT • Only small number of averages required • 3-5 points per decade • Create gain curve with cubic spline or linear curve • Make sure tone does not saturate IF amplifiers San Francisco 2011 57 Calibration Curve at X-band Relative Ga in (dB) Gain 30 20 10 0 -10 -20 1 Gain : 23.1 San Francisco 2011 10 100 1K 10K 100K 1M 10M 100M Fourier Frequency(Hz) 58 Phase Noise of X-band Synthesizer -50 Phase Noise (dBc (dBc/Hz) /Hz) -70 -90 -110 -130 -150 -170 -190 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8 1E+9 Fourier Frequency (Hz) San Francisco 2011 59 Tips for Measuring Noise • Measure Power Spectral Density in Vrms/√Hz • Use Hanning window • Confidence interval depends on number of averages � • Confidence interval de depends p � ends also on resolution and and � video bandwidth for swept analyzers � • Measure system noise floor. San Francisco 2011 60 Noise Floor Reduction Techniques • Cross-Correlation • Carrier Suppression San Francisco 2011 61 Cross-correlation PM Noise System for Two Port Measurements � φ LNA DUT Channel 1 φ REF LNA on NoiseCrosscorrelated = NoiseCorrelated1,2 + San Francisco 2011 Channel 2 off NoiseUncorrelated1 + NoiseUncorrelated2 N Averages 62 Correlated Noise Measurements -140 Channel 1 Channel 2 Cross Spectrum Voltage N Noise oise (dBV/rtHz) -150 -160 -170 -180 -190 -200 1E+6 1E+7 Frequency (Hz) 100,000 Ave : San Francisco 2011 35 min 63 Cross-correlation Oscillator PM Measurements φ LNA Channel 1 Reference 1 DUT φ Reference 2 Sφ ( f )Cross1,2 = Sφ ( f )DUT + San Francisco 2011 LNA Channel 2 Sφ ( f )Ref1 + Sφ ( f )Ref2 + Sφ ( f )System1 + Sφ ( f )System2 NAverages 64 Cross-correlation Oscillator PM Measurements Three oscillator cross-spectrum measurement -100 CH1 CH2 Cross -110 -120 L(f), [dBc/H [dBc/Hz] z] -130 -140 -150 -160 -170 -180 -190 1E+0 San Francisco 2011 1E+1 1E+2 1E+3 Frequency [Hz] 1E+4 1E+5 65 AM Measurements* v(t ) = kα P(t ) α ϕ or DUT α Diode Power Detector CAL Δv kα P0 = ΔP P0 In-phase non­ saturated Mixer v(t ) = kα P0 (1 + α (t ))2 ≅ kα P0 (1 + 2α (t ) + α 2 (t) ) SVrms ( f ) Sα ( f ) = 4kα2 P02 Sα ( f ) = San Francisco 2011 * SVrms ( f ) Ka , Ka = 4kα2 P02 *The Measurement of AM noise of Oscillators – Rubiola2005 66 Cross-correlation Source AM Measurements ChannelA * DUT ChannelB CAL * Sα ( f )CrossAB = Sα ( f )DUT + Sα ( f )CrossAB = Sα ( f )CrossAB = San Francisco 2011 SSystem1 ( f ) + SSystem2 ( f ) 2N Ave SvAB ( f ) 4kα2A PA2 kα2B PB2 SvAB ( f ) Ka , K a = 4kα2A PA2 kα2B PB2 *The Measurement of AM noise of Oscillators – Rubiola2005 67 Two-port AM Measurements ChannelA DUT REF ChannelB CAL Sa ( f )Cross1,2 = Sα ( f )DUT +Sα ( f )REF + Sα ( f )SystemA +Sα ( f )SystemB 2NAverages Reference AM noise must be less then DUT noise Saturation can help in reducing AM Source noise San Francisco 2011 68 Digital Measurement Systems • Carrier frequencies are sampled directly with A/D converters and phase information is extracted extracted mathematically. mathematically. • No phase lock required • Signals can be of different frequencies. • Limited frequency range 1 - 400 MHz. • Fourier Range of 1 mHz to 1 MHz. • Limited sensitivity San Francisco 2011 69 Integral PM and AM Noise Standards • Low noise signal source • Two outputs with extremely low differential AM and PM noise • Calibrated noise source • Greatly simplifies AM and PM measurements • Generates a calibrated level of equal AM and PM noise San Francisco 2011 70 Block Diagram of NIST PM/AM Noise Standard � φ on Si Signal gnal off Phase Shifter Splitter Reference Summer Modulated on White Noise San Francisco 2011 off Bandpass 71 Added Noise Appears as Equal Amounts of AM and PM AM PM υ0 − f υ0 υ0 + f V0 S a ( f ) = Sφ ( f ) = PSDVn (υ 0 − f ) + PSDVn (υ 0 + f ) 2V02 While ∞ San Francisco 2011 ∫S 0 φ ( f ) << 0.1 72 Basic Carrier Suppression Technique (Interferometric) � φ LNA Analyzer DUT San Francisco 2011 73 Basic Carrier Suppression Technique (Interferometric) � φ LNA φ Analyzer DUT DUT San Francisco 2011 74 Delay Line Measurement Systems Residual Delay line PMMeasurement MeasurementSystem System ϕ Source f τd Sv ( f ) Sϕ ( f ) = 2 2 k H ( jf ) v(t ) ∝d Δϕ t ) 2 H ( jf ) = 4sin 2 (π f τ d ) San Francisco 2011 75 Delay Line Transfer Function � Transfer function of delay line 10 10 us 1 us 0 IH f ( jf )I 2 -10 -20 2 ⎤ 10Log ⎡ 10Log H H j jf f ( ) ⎢⎣ ⎥⎦ -30 2 H ( jf ) = 4sin 2 (π f τ ) -40 -50 -60 1E+3 1E+4 1E+5 1E+6 Offset Frequency [Hz] System sensitivity is proportional to τd Long delays reduce sensitivity due to insertion loss Transfer Function has nulls at n/τd San Francisco 2011 76 Delay Line Transfer Function � L(f), [dBc/Hz] L(f), [dBc/Hz] PM noise of a DRO at 10 GHz 3 km Delayline 30 20 10 0 -10 -20 -30 -40 -50 -60 -60 -70 -80 -90 -100 -110 -120 -130 1E+0 Raw Data Processed Data 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 Frequency [Hz] System sensitivity is proportional to τd Long delays reduce sensitivity due to insertion loss Transfer Function has nulls at n/ τd San Francisco 2011 77 Single and Dual Source Measurements Direct Frequency Comparison Direct Phase Comparison (Delay Line) Advantages: Doesn’t require reference source No PLL effects Simple basic calibration Lowest noise floor Noise scales as f-1 near carrier Noise floor easy to determine Very wide band performance Can be used for residual measurement Disadvantages: Noise floor scales as f-3 near carrier Noise floor harder to determine Multiple delay lines required to cover different measurement conditions Not wideband – Frequency response has nulls Not useful for residual measurements Long delays have higher insertion loss San Francisco 2011 Requires reference of comparable quality Requires PLL to maintain phase lock between sources Calibration needed for measuring inside PLL loop bandwidth 78 Optical Encoded Delay Line Measurement � ~0.2 dB/km insertion loss vs 1 dB/m 0 4 m Microwave Delay line -2 0 L(f), [dBc/Hz] -4 0 Ebay 3 km Optical -6 0 -8 0 -1 00 -1 20 -1 40 -1 60 1 E +0 8 km state-of-the-art correlated system 1E +1 1 E +2 1E +3 1 E +4 1E +5 1 E +6 F re q u en c y [Hz] San Francisco 2011 79 Comparison of Noise Floors Noise e [dBC/Hz] SSB Phase Nois Typical Noise Floors @ 10 GHz -100 -110 -120 -130 -140 -150 -160 -170 -180 -190 -200 1E+0 San Francisco 2011 Single Mixer Correlated Mixer Correlated Photonic Delayline Correlated Interferometer (Rubiola) 1E+1 1E+2 1E+3 1E+4 Offset Frequency [Hz] 1E+5 1E+6 80 Measurement Problems • Mechanical Instabilities � • Phase locked loop effects � • IF Gain Flatness • Delay effects • RBW effects • AM Leakage • FM Port noise • Ground Loops San Francisco 2011 81 Phase Locked Loop Effects � -20 Phase Noise [dBc/H [dBc/Hz] z] � -40 -60 -80 -100 -120 -140 1 San Francisco 2011 10 Offset Frequency [Hz] 100 1000 82 IF Gain Reduction in Presence of Noise Floor � Measured Noise -70 -90 -110 Frequency -130 -150 Corrected Noise Phase Noise (dBc (dBc/Hz) /Hz) Gain -50 -170 Frequency -190 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8 1E+9 Fourier Frequency (Hz) Frequency San Francisco 2011 83 Delay Mismatch in Residual Measurements � Frequency Multiplier Phase Noise -90 Phase Noise [dB [dBc/Hz] c/Hz] -100 10 50 ns 1 ns 0 -10 IH f ( jf )I 2 -20 -30 -40 -50 #1 #1 vs #2 #2 Transfer function of delay line -110 -120 -130 -60 -70 -80 -140 1E+5 1E+6 1E+7 1E+8 Offset Frequency [Hz] -150 -160 10 100 1000 10000 100000 1000000 10000000 Offset Frequency [Hz] San Francisco 2011 84 Insufficient Resolution Bandwidth Phase Noise -80 -90 -110.0 10 Hz RBW -100 -110 -120 -120.0 L(f), [dBc/ [dBc/Hz] Hz] L(f) [dBc/ [dBc/Hz] Hz] Phase Noise -100.0 100 Hz RBW -130 -140 -150 -160 -170 -180 1E+1 -140.0 -140.0 -150.0 -160.0 -170.0 -180.0 -190.0 1E+2 1E+3 1E+4 Frequency [Hz] San Francisco 2011 -130.0 1E+5 1E+6 -200.0 1E+1 1E+2 1E+3 1E+4 1E+5 Frequency [Hz] 85 80 60 Sensitivity (dB) PM Sens AM Sens DC Level DC Offset 0.4 0.2 40 0 20 -0.2 0 -0.4 -20 -0.6 -40 -0.8 -60 -1 Mixer IF (Volts) AM to PM Leakage in Mixer AM Null Offset -80 0 45 90 135 -1.2 180 Phase Angle (degrees) San Francisco 2011 86 AM to PM Leakage in Mixer Residual Measurement 80 α 60 SATURATED AMPLFIER or LIMITER Absolute Measurement (dB Sensitivity (d B)) Modulator PM Sens AM Sens DC Level DC Offset 0.4 0.2 40 0 20 -0.2 0 -0.4 20 --20 -0.6 0 .6 -40 -0.8 -60 Mixer IF (Volts) ϕ -1 AM Null Offset -80 0 45 90 135 -1.2 180 Phase Angle (degrees) ∑ Move Operating Point To Here α Modulator San Francisco 2011 87 FM Port (PLL) Noise Phase Noise -70 -80 f −2 slope -90 L(f) [dBC/Hz] [dBC/Hz] -100 -110 -120 -130 -140 -150 -160 -170 1E+1 1E+2 1E+3 1E+4 1E+5 Offset Frequency [Hz] 2 2 ⎛υ ⎞ ⎛ K ⎞ Sφ ( f ) = ⎜ 0 ⎟ S y ( f ) = ⎜ υ ⎟ PSD (V rms ) ⎝ f ⎠ ⎝ f ⎠ San Francisco 2011 Kv is VCO sensitivity in Hz/V 88 Ground Loops and EMI Noise f loor at X-Band -80 -90 L(f) L (f) [[dBc/ dBc/Hz z]] -100 -110 -120 -130 -140 -150 -160 -170 -180 -190 -200 1E+1 San Francisco 2011 1E+2 1E+3 1E+4 1E+5 Frequenc y [Hz] 1E+6 1E+7 89 Ground Loops and EMI San Francisco 2011 90 Ground Loops and EMI • • • • • • • • • San Francisco 2011 Power devices from batteries Shielding – Faraday and magnetic DC Blocks Replace IF Gain with RF gain (Interferometric) Isolate or remove computer connections Plug in AC power all at same place EMI filtered power strips/ Ferrite Cores Ground lifting plugs (dangerous) Isolation transformers (dangerous) 91 Conclusions • Great care must be taken when calibrating the mixer sensitivity. • Calibrating Kd versus Fourier can be achieved by utilizing a modulation technique. • For ultra-low noise floors cross-correlation or carrier-suppression techniques can be used. San Francisco 2011 92 References 1. Algie L. Lance, Wendell D. Seal, and Frederik Labaar, Phase noise and AM noise measurements in the frequency domain, Infrared and Millimeter Waves, vol. 11, 1984, pp. 239–284. 2. D.B. Sullivan, D.W. Allan, D.A. Howe, and F.L. Walls, Characterization of Clocks and Oscillators, NIST Tech. Note 1337, 1-342, 1990 3. E. Rubiola, The measurement of AM noise of oscillators 4. E. Rubiola, V. Giordano, Advanced interferometric phase and amplitude noise measurements , Rev. Sci. Instrum. vol.73 no.6 p.2445-2457, June 2002. 2002. 5. Enrico Rubiola, Phase noise and frequency stability in oscillators, Cambridge University Press, Cambridge, UK, November 2008. 6. F. L. Walls, Secondary standard for PM and AM noise at 5, 10 and 100 MHz, IEEE Trans. Instrum. Meas. 42 (1993) no. 2, 136–143. 7. J. R. Vig, IEEE Standard Definitions of Physical Quantities for Fundamental Frequency and Time Metrology--Random Instabilities, IEEE Standard 1139-1999 8. W. F. Walls, Cross-correlation phase noise measurements, Proc.Freq. Control Symp., IEEE, New York, 1992, May 27-29 1992, pp. 257-261. San Francisco 2011 93