Main Sources of Electronic Noise

Transcription

1 Main Sources of Electronic Noise Thermal Noise - It is always associated to dissipation phenomena produced by currents and voltages. It is represented by a voltage or current sources randomly variable in time. - It is analytically described by a stationary process - Amplitude distribution: GAUSSIAN with zero mean value - Power Spectral Density: constant (white noise) G(f) η η = Power spectral density =K. T f

2 Main Sources of Electronic Noise Shot noise It arises typically in PN junctions forwardly biased; it is due to the discrete nature of current through the junction, which results randomly variant around the imposed bias value Amplitude distribution: GAUSSIAN with zero mean value Constant spectrum (white noise) G(f) η η = Power spectral density (=q. I) f

3 Main Sources of Electronic Noise Flicker noise (1/f) 1. It arises in semiconductor devices, due to impurities and defects in the crystal structure).. Its spectrum is not constant (energy is concentrated at low frequency). Power Spectrum: a I G ( f ) = K W / Hz b f K: depends on the fabrication process I: DC current through the device a: typically 0.5 b: 1 Nota: The amplitude distribution is not always GAUSSIAN

4 Noise Characterization in microwave devices P Nin -port G a P Nin N DB = Added Power (from the -port) P Nout = G a P Nin +N DB =Total noise Power N DB The noise figure NF: define the attitude of the -port of adding noise at the output: NF PNout = G P a Nin P Nout is the actual noise power at output while G a. P Nin is the noise power at output if the -port would not add noise power (G a is the available power) Actually NF is a function of frquency, so the above powers must be assumed per unit band (i.e. they represent actually power densities) Moreover NF depends also on the source impedance (Γ S )

5 NF dependance on Γ S NF = ( NF ) + 4r min n Γ Γ s min ( ) s 1+Γ 1 Γ min (NF) min = Minimum value of NF Γ min value of Γ S which determines NF=NF min r n = Normalized noise resistance All these parameters are frequency dependent. Typically, they are made available by the manufacturers of commercial devices (directly into.sp data files).

6 Constant Noise Figure Circles If we plot the equation expressing NF as function of Γ S on the Smith Chart (representing Γ S ), we obtain a circle with the following center and radius: C F = Γ min 1+ N i, r F = 1 1+ N i N i + N i ( ) 1 Γ m N i is given by: N i = NF ( NF) 4r n min ( ) 1+ Γ min

7 Noise Figure for cascaded stages G a1, N F1 G a, N F G a3, N F3 NF 1 NF 1 = G G G 3 ( NF ) NF TOT a1 a1 a The noise figure is mainly determined by the first stage NOTE: In general the value of Γ S that determines the minimum value of NF is different by the one that maximizes G T ; the choice of Γ S for the first stage is then the result of a compromise between the noise figure and gain

8 Design of a low noise amplifier The choice of Γ s is a compromise between G T and NF. Some circles with NF=cost and G a =cost are first plotted on the Smith chart (Γ s ). The value of Γ s is selected within the common area of two circles. Considering that NF increases with the radius while Ga decreases with it, we have (with reference to the zones 1 and ): Cerchi a NF=cost NF NF1 G a1 G a Cerchio di Instabilità del gen. In zone 1: NF NF 1 and G a G a NF is previliged NF min 1 In zone : NF NF and G a G a1 G a is previliged Cerchi a Ga=cost Once assigned Γ s,opt, Γ L,opt is computed by imposing the matching at output (then G T =G a )

9 Example of design Amplifier Requirements Frequency Band: GHz Minimum Transducer Gain: 10.5 db Maximum Noise Figure: 1.5 db Substate: Duroid ε r =.54 H= mm t = 35 µ Active Device MGF193 Mitsubishi (GaAs Mesfet) MSG (6.8 GHz): db (with NF=3.1 db) Minimum NF (6.8 GHz): 1.13 db (with Gt=8.06 db) Topology: SUBCKT TLOC m TLIN NET= "transistor" 1 TLIN

10 Biasing network of the active device V GS Bias V DS Bias RF Short Circuit RF Open Circuit NOTE: The S parameters delivered by the manufacturer refers to the red sections. After the biasing network has been assigned, the S parameters changes to the ones referred to the black sections Input reference section Block Capacitors (DC) Output reference section

11 Evaluation of the S parameters of the biased active device MLEF ID= TL14 W= mm L= mm MLIN ID= TL15 W= 0. mm L= 0.5 mm MSUB Er=.54 H= mm Rho= T= mm 1 Tand= 0 ErNom=. Name= SUB1 MLEF ID= TL16 W= mm L= mm MLIN ID= TL17 W= 0. mm L= 0.5 mm MRSTUB ID= TL3 Ri= mm Ro= mm Theta= 60 Deg 3 1 MTEE$ ID= TL MTEE$ ID= TL MRSTUB ID= TL4 Ri= mm Ro= mm Theta= 60 Deg MLIN ID= TL10 W= 1.44 mm L= 0.99 mm CAP ID= C1 C= 15 pf MCTRACE ID= TL13 W= 0. mm R= L= mm 5 mm 3 1 MLIN ID= TL5 W= 1.4 mm L= 0.5 mm SUBCKT ID= S1 NET= "MGF193" 1 MLIN ID= TL6 W= 1.4 mm L= 0.5 mm 3 MCTRACE ID= TL1 W= 0. mm L= mm R= 5 mm 1

12 S parameters for the design Frequency: 6.8 GHz

13 Γ S selected for maximum gain Load (0.74, 15) Γ L ( O pt) : (0.771, ) Gen. GT=15.1 NF=3.18 G T : db NF: 3.36 Γ : (0.68 (0.73,, 167.5) 173.5) S ( O pt)

14 Selection of Γ S as a compromise between G T and NF Load NF=1.5 db Source X Γ L X Γs G T = 10.7 db NF=1.36 db Γ L = Γ S = Ga=10.5 db

15 Input matching network 107 x Γ s x 1+j1.15 LOAD ID= Z1 Z= 50 Ohm TLIN ID= TL1 Z0= 50 Ohm EL= Deg F0= GHz TLOC ID= TL Z0= 50 Ohm EL= Deg F0= GHz PORT P= 1 Z= 50 Ohm Γ S

16 Output matching network x Γ L TLIN ID= TL1 Z0= 50 Ohm EL= Deg F0= GHz PORT P= 1 Z= 50 Ohm 96.3 LOAD ID= Z1 Z= 50 Ohm TLOC ID= TL Z0= 50 Ohm EL= Deg F0= GHz x 1+j1.3 Γ L

17 Scheme of the overall amplifier MLEF ID=TL4 W=3.5 mm L=.43 mm PORT P=1 Z=50 Ohm MLIN ID=TL6 W=1.386 mm L=1 mm 3 MTEE$ ID=TL5 1 MSUB Er=.54 H=0.508 mm T=0.035 mm Rho=1 Tand=0 ErNom=. Name=SUB1 MLIN ID=TL1 W= mm L=1.47 mm SUBCKT ID=S1 NET="transistor" 1 MLIN ID=TL3 W=0.364 mm L=1.617 mm MTEE$ ID=TL8 1 3 MLEF ID=TL9 W=1.443 mm L=4.96 mm MLIN ID=TL7 W=1.387 mm L=1 mm PORT P= Z=50 Ohm

18 Amplifier Layout

19 Amplifier Response Optimized Initial Response Response 6.5 GHz GHz db db 6.5 GHz db 6.5 GHz db DB(GT()) (L) DB(GT()) Ampli (L) Ampli DB(NF()) (R) DB(NF()) Ampli (R) Ampli Frequency 6.4 (GHz) Frequency (GHz)

20 Scheme of a power microwave amplifier Reti di Polarizzazione R 0 V in MATCH In MATCH Out R 0 Γ S Γ L The concept is identical to the ones seen before. In this case however the values of Γ S e Γ L to be assigned have to maximize the power delivered to the load (for specified biasing conditions). Instability must be obviously avoided (commercial devices are generally pre-matched internally for unconditional stability). The active device must be characterized for large signal operation

21 Example of PA design Amplifier Requirements Frequency Band: GHz Output Mean Power (-tone): 0 dbm Carrier-to-IM3: >35 db Minimum GT: >9 db Active Device MGFK5V4045 Mitsubishi (GaAs FET) IP3: 36 dbm (CI=IP3-Pm-6=38 db) P1dB: 5 dbm (min) Device unconditionally stable. Optimum Loads: Freq. Γs Γs ΓL ΓL G T,MAX

22 Ideal design The maximum Transducer Gain is obtained by imposing the optimum loads. The values so obtained are not constant with frequency: they decreases, reaching the minimum at 14.5 GHz (9.6 db). It is then convenient to design the transforming network at the highest frequency using then optimization for getting the requirements satisfied in the whole band. The networks used are double stub networks with shunt open stubs (no use of via-hole). The design can be approached in two (equivalent) ways: Γ S(L) Φ Φ 1 Φ 50Ω Γ S(L) Φ Φ 1 Φ 50Ω

23 Design of the Double Stub with the Smith Chart (case 1) Insert Γ S(L) and store Draw the circle g=1 rotated of -70 (toward source) Draw the circle g=cost passing for the current point Select one on the intersection betwen the two circles as the new current point. The susceptance b 1 is given by the imaginary part of Delta Y with the sign reversed. Give an increment of +70 to the phase of gamma of the current point (the new current point must be on the circle g=1) The imaginary part of the Y of the current point represents b. It has: φ 1 =tan -1 (b 1 ) and φ =tan -1 (b )

24 Result of initial design Ω MGFK Ω Zc=50 Ω for all the lines G T,max G T Γ in Γ out

25 Optimized result (imposing optimum loads) 150Ω, Ω, Ω 150 Ω Ω MGFK 0 Ω Ω Ω G T,max G T Γ in Γ out

26 Optimized result (imposing GT>9.5 db) 150Ω, Ω, Ω 14.5 Ω Ω MGFK 0. Ω Ω Ω DB( S(1,1) ) (L) Ampli -30 DB( S(,) ) (L) Ampli DB( S(,1) ) (R) Ampli DB(GMax()) (R) Ampli Frequency (GHz) 5

27 Balanced Amplifier Vin 1a 0 90 C=3 db a 3a V in / Γ in jv in / Γ in A A V in jv in A A b 3b C=3 db b Vout V = AV, V = j A V Gain: 3 Reflection: b in b in V V P V = V = j + = j AV = A b 3b out out 4b in Pin V = V, V = AV, V =Γ AV, V = j AV, + + 1a in a in a in in 3a in V3a = jγin AV in, V1 a = ( jv3a + Va ) = Γin AV in +Γin AV in V Γ in = = 0 V 1a + 1a

28 IP3 in balanced amplifiers Pin db HYB P in / jp in / A A P out P j out 3 db HYB 90 0 Pout [ ] P = 3( P 3) IP, P = 3( P 3) IP int,1 out 3 int, out 3 P = 3( P 3) IP + 3 = 3P IP 6 = 3P IP int out 3 out 3 out 3 IP = IP Result: the equivalent IP3 of the overall amplifier is doubled with respect the one of the single amplifiers. This means that for the same overall output power the power of the intermodulation products is 6 db lower.

29 Result of balanced configuration (amplifiers optimized for G T >9.5 db) 150Ω, Ω, Ω 14.5 Ω Ω MGFK 0. Ω Ω Ω Scheme of each amplifier