The effect of AM noise on correlation PM noise measurements 1/f - PowerPoint PPT Presentation

The effect of AM noise on correlation PM noise measurements 1/f noise in RF and microwave amplifiers Enrico Rubiola, Rodolphe Boudot, Yannick Gruson FEMTO-ST Institute, Besançon, France CNRS and Université de Franche Comté TimeNav ’ 07 – May 31, 2007 Outline Part 1 – The effect of AM noise ... Part 2 – Amplifier noise ... home page http:/ /rubiola.org

1 – The effect of AM noise on correlation phase noise measurements

Effect of AM noise on a saturated mixer 3 mixer LO RF input input D2 D1 V l cos[ ω 0 t ] V r cos[ ω 0 t + ϕ ] LO power → V OS RF power → V OS D3 D4 IF v out o FFT IF load v = k ϕ ϕ + V OS static: P → V OS (offset) phase detection for | ϕ | ≪ 1 noise: AM noise → DC noise mistaken for phase noise origin: diode and balun asymmetry RF mixer: balun asymmetry ≈ const. vs. frequency microwave: balun asymmetry depends on frequency

The AM noise propagates through the system 4 A null of AM sensitivity (sweet point) A delay de-correlates the two inputs, can be found in some mixers thus it destroys the sweet point A C DUT REF 2−port device DUT RF RF analyzer analyzer delay x x FFT FFT dc dc LO LO (ref) phase phase optional tip: use a phase offset, or a DC bias at the mixer IF phase lock v o ( t ) = k ϕ ϕ ( t ) + k lr α ( t ) v o ( t ) = k ϕ ϕ ( t ) + k l α l ( t ) + k r α r ( t ) phase B D meter output DUT detector (noise only) 2−port device RF analyzer analyzer DUT RF x x FFT FFT Δ bridge Σ µ w dc dc REF LO LO (ref) phase and ampl. phase lock v o ( t ) = k ϕ ϕ ( t ) + k l α l ( t ) + k r α r ( t ) v o ( t ) = k ϕ ϕ ( t ) + k sd α ( t ) In a bridge, the AM noise propagates to With two separated inputs, the effect of the output only through the LO. AM noise adds up The effect is strongly reduced by the RF amplification before detecting

Basics of correlation spectrum measurements 5 a(t) phase noise measurements + DUT noise, a, b instrument noise x=c−a Δ Σ normal use c DUT noise analyzer c(t) − FFT background, a, b instrument noise − ideal case c = 0 no DUT y=c−b Δ Σ background, a, b instrument noise b(t) + with AM noise c ≠ 0 AM-to-DC noise S yx = E { Y X ∗ } W. K. theorem S yx = � Y X ∗ � m measured, m samples a , b and c are incorrelated expand X = C − A and Y = C − B Averaging on a sufficiently large S yx = S cc a , b , c independent number m of spectra is necessary � S yx = S cc + O ( 1 /m ) measured, m samples to reject the single-channel noise

The AM noise in a correlation system 6 Should set both channels at the sweet The delay de-correlates the two inputs, point, if exists so there is no sweet point REF A C (ref) AM V OS phase LO RF delay arm a x x dc dc DUT arm a DUT analyzer analyzer RF LO AM V OS common FFT FFT 2 − port device RF LO arm b y y delay dc dc arm b REF LO RF AM V OS phase (ref) B D phase lock phase and ampl. meter output REF AM (noise only) V OS LO LO RF x x Δ AM bridge a µ w arm a dc Σ dc DUT analyzer analyzer RF DUT common FFT FFT 2 − port device RF RF y y Δ arm b dc Σ µ w dc bridge b LO LO REF V OS (ref) AM phase and ampl. phase lock Should set both channels at the sweet The effect of the AM noise is strongly point of the RF input, if exists, by reduced by the RF amplification offsetting the PLL or by biasing the IF pink: noise rejected by correlation and averaging

Measurement of the mixer sensitivity to AM 7 • The measurement schemes follow immediately from the statement of the problem • A lock-in amplifier is used for highest noise immunity • Set the amplitude modulator to the minimum of residual PM (at least in the scheme B-C) A LO & RF → IF: RF amplit. dc in coefficient k lr modulat. amplifier lock−in LO phase v o ( t ) = k ϕ ϕ ( t ) + k lr α ( t ) dc bias osc Σ B−C LO or RF → IF: attenuat. RF (LO) coefficients k l and k r dc in amplifier lock−in LO (RF) amplit. v o ( t ) = k ϕ ϕ ( t ) + k l α l ( t ) + k r α r ( t ) modulat. phase dc bias osc Σ D LO → IF in a sync.-detection RF scheme: coefficient k sd dc in amplifier lock−in amplit. LO modulat. v o ( t ) = k ϕ ϕ ( t ) + k sd α ( t ) dc bias osc Σ

Example of results (microwave mixers) 8 200 k l 11GHz 8dBm The AM sensitivity depends on frequency. k l 9GHz 7dBm k 10GHz 8dBm r k r This is ascribed to the microstrip baluns, 11GHz 8dBm 100 k r 8.5GHz 8dBm l or k r (mV) k 8GHz 8dBm and to the diode capacitances r 0 k −100 Narda 4805 SN0973 7 mar 2006 −200 −50 −25 0 25 50 phase offset, degrees 200 k 7GHz 6dBm r The AM sensitivity can have opposite sign 100 at the two inputs (mV) k r 7GHz 8dBm k r 0 k l 6GHz 8dBm k l 7GHz 8dBm or k l −100 k l 6GHz 6dBm Pulsar MM−02−SC mar 2006 −200 −50 −25 0 25 50 phase offset, degrees

Example of results (microwave mixers) 9 100 100 Narda 4805 SN 0973 Pulsar MM−02−SC 11 mar 2006 mar 2006 80 80 60 60 k l (mV) k l (mV) 40 40 6dBm 6dBm 20 7dBm 20 7dBm 8dBm 8dBm 9dBm 9dBm 0 0 8 9 10 11 5 6 7 8 (GHz) ν ν (GHz) 100 100 6dBm Narda 4805 SN 0973 6dBm 7dBm 7dBm 11 mar 2006 80 80 8dBm 8dBm 9dBm 9dBm 60 60 (mV) (mV) 40 40 k r k r 20 20 Pulsar MM − 02 − SC mar 2006 0 0 8 9 10 11 5 6 7 8 (GHz) ν ν (GHz) The effect of power is somewhat weaker than that of frequency

Example of results (microwave mixers) 10 Mixer k ϕ k lr k r k l k sd Narda 4805 s.no. 0972 272 16 7.9 37 6.5 Narda 4805 s.no. 0973 274 18.3 17.1 44 9.8 NEL 20814 279 51.5 12.1 37.9 2.7 NEL 20814 305 41 1.9 30.2 3.73 unit mV/rad mV mV mV mV Test parameters: ν 0 = 10 GHz, P = 6 . 3 mW (8 dBm) Some relevant facts • The AM noise rejection is of 15–40 dB • Generally, k sd is smaller than the other coefficients • There is no predictable relation between k φ , k l , k r , k lr , and k sd • It is observed that k lr , ≠ k l + k r

Example of results (VHF mixers) 11 60 60 30 30 (mV) (mV) r 0 0 k r k k l k l 200MHz 5dBm 200MHz 5dBm ou k l or k l 200MHz 9dBm 200MHz 9dBm k l k l 6MHz 5dBm 6MHz 5dBm l k l k l k k 6MHz 9dBm 6MHz 9dBm l −30 − 30 k r k r 200MHz 5dBm 200MHz 5dBm k r k r 200MHz 9dBm 200MHz 9dBm k r k r 6MHz 5dBm 6MHz 5dBm TFM10514M2 TFM10514M3 k r k r 6MHz 9dBm 6MHz 9dBm −60 − 60 −50 −25 0 25 50 − 50 − 25 0 25 50 phase offset, degrees phase offset, degrees 60 60 30 30 (mV) (mV) k r k r 0 0 k k l 200MHz 5dBm 200MHz 5dBm l or or k l k l 200MHz 9dBm 200MHz 9dBm k l k l 6MHz 5dBm 6MHz 5dBm k l k l k l k l 6MHz 9dBm 6MHz 9dBm − 30 −30 k r k r 200MHz 5dBm 200MHz dBm k r k r 200MHz 9dBm 200MHz 9dBm k r k r 6MHz 5dBm 6MHz 5dBm HP10514 ZFM2 k k r 6MHz 9dBm 6MHz 9dBm r − 60 −60 − 50 − 25 0 25 50 −50 −25 0 25 50 phase offset, degrees phase offset, degrees • The AM noise rejection is of 15–40 dB • The sweet point is not observed in general • There is no predictable relation between k φ , k l , k r , ( k lr , and k sd are not reported)

12 Warning: even in single-channel measurements, the pollution from AM noise may be not that small −100 PM noise, dBrad 2 /Hz frequency flicker Wenzel 501−04623 dBrad 2 /Hz −67 @ 1Hz specifications −110 −30dB/dec −120 −130 m e a s u r e d A −140 M n 15 dB o i s e ( b e AM noise, dBV/V/Hz s t c a s e ) −150 p o l l u t i o n f r o −160 m A M n o i s e dBrad 2 floor −173 /Hz −170 −180 10 1 10 3 10 4 5 10 10 2 Fourier frequency, Hz E. Rubiola, “The measurement of AM noise of oscillators,” arXiv:physics/0512082, dec 2005

13 Summary (1) The AM noise is taken in via the DC-offset sensitivity to the power The AM noise rejection is of 15–40 dB For a given mixer, there is no predictable relation between the AM noise sensitivity in different configurations The sweet point exists only in some configurations The sweet point is generally not observed in VHF mixers In correlation systems, rejecting the AM noise is possible only in some cases The AM noise can even limit the single-channel measurements home page http:/ /rubiola.org Free downloads (texts and slides)

2 – On the 1/f noise in RF and microwave amplifiers AM/PM noise additive parametric white local (flicker) environmental

Amplifier white noise 15 Noise figure F, Input power P 0 P=FkT 0 B RF spectrum S( ν ) P 0 B B V 0 cos ω 0 t ∑ g N e =FkT 0 ν 0 −f ν 0 ν 0 +f ν n rf ( t ) LSB USB S φ (f) low P 0 power law 0 b 0 = FkT 0 white � b i f i S ϕ = high P 0 phase noise P 0 P 0 i = − 4 f Cascaded amplifiers (Friis formula) The (phase) noise is chiefly that of the 1st stage F 1 F 2 F 3 g 1 g 2 g 3 The Friis formula applied to phase noise + ( F 2 − 1) kT 0 b 0 = F 1 kT 0 N = F 1 kT 0 + ( F 2 − 1) kT 0 + . . . + . . . P 0 g 2 P 0 g 2 1 1 H. T. Friis, Proc. IRE 32 p.419-422, jul 1944