Phase Noise in RF and Microwave Amplifiers

Transcription

1 Phase Noise in RF and Microwave Amplifiers Enrico Rubiola and Rodolphe Boudot IFCS, Newport, CA, 1 4 June 2010 Outline Noise types (white and flicker) Amplifier networks Experiments Conclusions home page

2 2 AM-PM noise types file: amp-noise-tree parametric AM-PM noise noise originates near DC additive noise originates around! 0 environmental internal environmental internal temperature RF leakage white 50 Hz B fields power supply acoustic flicker (1/f) rnd walk (1/f 2 ) drift radiation

3 The difference between additive 3 and parametric noise additive noise parametric noise u input Σ z noise-free amplifier RF noise, close to! 0 v output u input AM x PM y noise-free amplifier near-dc noise v output PSD file: amp-add-vs-param Sz(f) (noise) stopband sum passband! 0 Sv(f) (output) Su(f) (input) stopband PSD Sy(f) (noise) up-conversion stopband passband! 0 Sv(f) (output) Su(f) (input)!! stopband the noise sidebands are independent of the carrier the noise sidebands are proportional to the carrier

4 White noise in cascaded amplifiers White noise is chiefly the noise of the first stage 4 kt 0 (F 1 1)kT 0 (F 2 1)kT 0 (F 3 1)kT 0 input file: amp-cascaded-friis A 1 F 1 A 2 A 3 F 2 F 3 output N e = F 1 kt 0 + (F 2 1)kT 0 A 2 1 F = F 1 + (F 2 1) A (F 3 1) A 2 2 A2 1 + (F 3 1)kT 0 A 2 2 A Friis formulae H. T. Friis, Proc. IRE 32 p , jul 1944 Noise is chiefly that of the 1st stage b 0 = FkT 0 P 0 white phase noise b 0 = F 1kT 0 P 0 + (F 2 1)kT 0 A 2 1 P 0 + (F 3 1)kT 0 A 2 2 A2 1 P Friis formula for phase noise

5 Parametric noise in cascaded amplifiers 5 There is a nonlinear model that gives exactly the same results, see Chap. 2 of E. Rubiola, Phase Noise and Frequency Stability in Oscillators, Cambridge 2008, ISBN u input! 1 " 1! 2 " 2 AM PM noise-free amplifier AM PM x 1 y 1 x 2 y 2 noise-free amplifier v output file: cascaded-ampli ampli 1 ampli 2! =! 1 +! 2 " = " 1 + " 2 Flicker: the two amplifiers are independent E{α 2 } = E{α 2 1} + E{α 2 2} S α = S α 1 + S α 2 E{ϕ 2 } = E{ϕ 2 1} + E{ϕ 2 2} S α = S ϕ 1 + S ϕ 2 Environment: a single process drives the two amplifiers α = α 1 + α 2 E{α 2 } = E{(α 1 + α 2 ) 2 } ϕ = ϕ 1 + ϕ 2 E{ϕ 2 } = E{(ϕ 1 + ϕ 2 ) 2 } Yet there can be a time constant, not necessarily the same for the two devices

6 Flicker noise in parallel amplifiers 6 E. Rubiola, Phase Noise and Frequency Stability in Oscillators, Cambridge 2008, ISBN u 1 A 1 v 1 ψ 1 input v i m way power divider u k A k v k ψ k m way power combiner output v o ϕ v m ψ m u m A m The phase flicker coefficient b 1 is about independent of power The flicker of a branch is not increased by splitting the input power At the output, the carrier adds up coherently b the phase noise adds up statistically 1 = 1 m Hence, the 1/f phase noise is reduced by a factor m Only the flicker noise can be reduced in this way b 1 cell

7 Volume law 7 The analysis of the parallel amplifier suggests that: For a given technology, the flicker coefficient b 1 should be proportional to the inverse of the volume of the active region Gedankenexperiment - Flicker is of microscopic origin because it has Gaussian PDF (central limit theorem) - Join the m branches of a parallel device forming a compound - Phase flicker is proportional to the inverse size of the amplifier active region

8 Parametric noise in regenerative amplifiers R. Boudot, E. Rubiola, arxiv: v1, Jan Submitt. IEEE Transact. MTT 8 Vin A0 RF filter Vout A = A 0 1 A 0 β file: amp-regen-sch short delay feedback ß phase adj. ampl. adj. A = A m 0 β = Am A m S! (f) [rad 2 /Hz] /f (b 1 /f) file: amp-regen-mecha 1kHz 1MHz roundtrip rad 2 /Hz (few roundtrips) (b 0 ) 1GHz Short roundtrip time, vs. flicker time frame Quasi-static analysis holds carrier f A A = A 0 e jψ 1 A 0 β e jψ A 0 1 A 0 β 1+j 1 1 A 0 β ψ 1 ϕ = 1 A 0 β ψ 2 1 (b 1 ) RA = (b 1) ampli 1 A 0 β (b 1 ) RA = m 2 (b 1 ) ampli.

9 Measurement methods 9 Saturated mixer (common laboratory practice) 3 db atten DUT atten atten RF LO IF LNA FFT analyzer 90º adj Bridge (interferometer) V 0 cos(! 0 t) pump 0º 90º dark hybrid junction 90º 0º 0º bridge DUT hybrid junction 90º 0º 0º " # (microwave) error amplifier coherent detector x AM noise y PM noise FFT 90º phase & ampl. adjustment 90º null x cos(! 0 t) y sin(! 0 t) File: bridge E. Rubiola, V. Giordano, Rev. Sci. Instrum. 73(6) pp , June 2002

10 Flicker noise of some amplifiers R. Boudot, E. Rubiola, arxiv: v1, Jan Submitt. IEEE Transact. MTT 10 Amplifier Frequency Gain P 1 db F DC b 1 (meas.) (GHz) (db) (dbm) (db) bias (dbrad 2 /Hz) AML812PNB V, 425 ma 122 AML412L V, 100 ma AML612L V, 100 ma AML812PNB V, 1.1A 119 AFS V, 171 ma 105 JS V, 92 ma 106 SiGe LPNT V, 10 ma 130 Avantek UTC V, 100 ma Avantek UTO V, 23 ma 137

11 Phase noise vs. power. The 1/f phase noise b 1 is about independent of power The white noise b0 scales as the inverse of the power The corner frequency is misleading because it depends 130 on power 2 R. Boudot, PhD thesis Measured at LAAS Phase noise, dbrad 2 /Hz P= 50dBm P= 60dBm P= 70dBm P= 80dBm Amplifier X H at 4.2 K Data from IEEE UFFC 47(6):1273 (2000) E. Rubiola, Phase Noise and Frequency Stability in Oscillators, Cambridge 2008, ISBN Fourier frequency, Hz P= 80dBm P= 70dBm P= 60dBm P= 50dBm R. Boudot, E. Rubiola, arxiv: v1, Jan Submitt. IEEE Transact. MTT

12 Phase noise in cascaded amplifiers R. Boudot, E. Rubiola, arxiv: v1, Jan Submitt. IEEE Transact. MTT 12 The expected flicker of a cascade increases by: 3 db, with 2 amplifiers 4.8 db, with 3 amplifiers White noise is limited by the (small) input power

13 Phase noise in parallel amplifiers R. Boudot, E. Rubiola, arxiv: v1, Jan Submitt. IEEE Transact. MTT 13 Connecting two amplifier in parallel, a 3 db reduction of flicker is expected

14 Flicker noise in parallel amplifiers 14 E. Rubiola, Phase Noise and Frequency Stability in Oscillators, Cambridge 2008, ISBN AML812PNA0901 (100mA) AML812PNB0801 (200mA) Phase noise, dbrad /Hz AML812PND0801 (800mA) AML812PNC0801 (400mA) Fourier frequency, Hz 10 Specification of low phase-noise amplifiers (AML web page) amplifier parameters phase noise vs. f, Hz gain F bias power AML812PNA AML812PNB AML812PNC AML812PND unit db db ma dbm dbrad 2 /Hz

15 15 Phase noise of a regenerative amplifier R. Boudot, E. Rubiola, arxiv: v1, Jan Submitt. IEEE Transact. MTT Indirect measurement: The RA replaces the two-stage sustaining amplifier in a Opto-Electronic oscillator Thanks to K.Volyianskiy for the OEO noise spectrum A RA is set for the gain of two cascaded amplifiers As expected, the RA flicker is 3 db higher than the two amplifiers Indirect measurement through the frequency flicker

16 Environmental effects in RF amplifiers 16 E. Rubiola, Phase Noise and Frequency Stability in Oscillators, Cambridge 2008, ISBN f 5 f 5 Amplifier phase noise courtesy of J. Ackermann N8UR, comments on noise are of E. Rubiola f 5 TADD-1 Spectracom 8140T b 1 = db HP 5087A and TADD-1 10 MHz b 1 = 133 db b 1 is the 1/f noise coefficient in dbrad 2 /Hz (dbc/hz + 3 db) 8140T HP 5087A TADD-1 background b 1 = 142 db TADD-1 5 MHz b 1 = db It is experimentally observed that the temperature fluctuations cause a spectrum Sα(f) or Sφ(f) of the 1/f 5 type Yet, at low frequencies the spectrum folds back to 1/f

17 a 2 + b 2 + c 2 + d 2 = 1 DB Correlation between AM and PM noise 17 R. Boudot, E. Rubiola, arxiv: v1, Jan Submitt. IEEE Transact. MTT u input file: AM-PM-correl a a=b=0.7 c=d=0 AM c x b PM d y noise-free amplifier correlated noise z v output a=0.4 b=0.4 c=0.2 d=0.8 The need for this model comes from the physics of popular amplifiers Bipolar transistor. The fluctuation of the carriers in the base region acts on the base thickness, thus on the gain, and on the capacitance of the reverse-biased basecollector junction. Field-effect transistor. The fluctuation of the carriers in the channel acts on the drain-source current, and also on the gatechannel capacitance because the distance between the `electrodes' is affected by the channel thickness. Laser amplifier. The fluctuation of the pump power acts on the density of the excited atoms, and in turn on gain, on maximum power, and on refraction index. a=0.4 b=0.92 c=d=0 a=b=0 c=d=0.7 AM and PM fluctuations are correlated because originate from the same near-dc random process

18 Conclusions 18 The model predicts the noise of the amplifier and of networks First noise model of the regenerative (positive-feedback) amplifier Experimental data validate the model Correlation between AM noise and PM noise (needs further work) Thanks to K.Volyanskiy for the measurement of the OEO noise, to Y. Gruson for help with phase noise measurements, to P. Salzenstein and to V. Giordano for support and discussions. This work results from a long-term transverse program on oscillators and frequency synthesis, supported by the following contracts: ANR-05-BLAN , CNES 60265/00, CNES 60281/00, ESA 20135/06/D/MRP, LNE/DRST R. Boudot, E. Rubiola, arxiv: v1, Jan Submitt. IEEE Transact. MTT home page