Mixer. General Considerations V RF VLO. Noise. nonlinear, R ON

Transcription

1 007/Nov/7

2 Mixer General Considerations LO S M F F LO L Noise ( a) nonlinearity (b) Figure 6.5 (a) Simple switch used as mixer (b) implementation of switch with an NMOS device. espect to espect to It is F LO desirable port : port : that the linear, nonlinear, time -variant system time - variant system switch turn on and off as ON abruptly as L possible.

3 3 Mixer t t ( ) ( ) ( ) ( ) (.4). / sin (.3) / sin in in out = = = = n n T n f n n T n f n n f f π π δ π π () ω π δ / function, delta Dirac the is convolution, denotes where = T = f f f

4 Mixer Table 6. Typical mixer characteristics. NF db IIP 3 Gain 0 db Input Impedance(Heterodyne) 50 Ω Port - to - port Isolation 0 0 db Passive and Active Mixer Passive mixer: no gain If LO is 50% duty cycle: 5 dbm LO Active mixer: Gain M M 3 DD Amplitude of fundamental ω LO in LO: /π F M Linearity and speed Figure 6.6 Active mixer. 4

5 Conversion Gain voltage conversion gain = Mixer rms; IF rms; F P L ; IF power conversion gain = Pavailable;F voltage conversion gain = power conversion if Z in Z L Z in = Z L = Z S = 50 Ω = 50 Ω (conjugate matching) 500 ~ 000 Ω (heterodyne) ~ kω (homodyne) voltage conversion gain, gain power conversion gain (in db) 5

6 Mixer Linearity P db and IPx 6

7 Mixer Linearity Mixer nonlinear model Signal : ω sig Interferers : ω, ω IM3 Caused by A: Caused by A: ω ω ± ω = ω ( ω ± ω = ωsig or LO or ( ω ± ω ) ± ( ω LO LO ± ω ) ± ( ω ± ω ) LO = ω ± ω ) IF = ω IF sig 7

8 Mixer Linearity Mixer nonlinear model Signal : ω sig Interferers : ω, ω IM Caused by A: Caused by A: ω = ± ω ωsig LO ± ω) ± ( ωlo ± ω) ( ω = ω IF 8

9 Mixer SSB and DSB Noise Figures SSB Noise Figure in S X Y ω LO Spectrum Spectrum at X Thermal Noise at Y Signal Band ω LO Image Band ω ω IF Figure 6.7 Folding of F and image noise into the IF band. ω Mixer (noiseless, gain = ) SSB NF = 3 db. 9

10 Mixer SSB and DSB Noise Figures DSB Noise Figure in S X Y ω LO Spectrum Spectrum at X Thermal Noise at Y Signal Band ω LO ω Figure 6.8 Down-conversion of an AM signal. Mixer (noiseless, gain = ) DSB NF = 0 db. General mixer SSB NF = DSB NF 3 db. ω IF ω Other harmonics in LO also convert noise to IF. relative unimportant due to () F bandwidth () higher harmonics 0

11 Port-to-port Isolation Mixer LO-F feedthrough: LO leakage to LNA or antenna F-LO feedthrough: strong interferes in F interact with LO LO-IF feedthrough: LO may be desensitized F-IF feedthrough: even-order distortion problem in homodyne receivers Single-Balanced and Double-Balanced Mixers Single-balanced mixer: less input-referred noise, but more LO noise converted. Worse LO-IF feedthrough Double-balanced mixer (Gilbert cell): less even-order distortion, higher input-referred noise.

12 Mixer Single-Balanced and Double-Balanced Mixers DD DD M 5 M 6 LO M 3 M 4 M M 3 LO M M F F M (a) (b) Figure 6.9 (a) Single-balanced mixer, (b) double-balanced mixer.

13 Mixer Direct feedthrough Important in homodyne receivers Negligible in heterodyne receivers CC CC out out I F I F I F I F ( a) (b) Figure 6.0 Simple mixers with (a) single-ended and (b) differential outputs. 3

14 Mixer Conversion of differential currents to single-ended output DD L L C I out M M Figure 6. Conversion of differential currents to single-ended output. Mixer Spurious esponse Frequency products m ωf ± nω LO Ensure products except for ωlo ω F not in the IF band. Need careful frequency planning 4

15 Mixer 5

16 Mixer 6

17 Mixer 7

18 Mixer 8

19 Bipolar Mixers CC CC C C C C LO Q P Q 3 LO Q P Q 3 CC F S Q E E : linearity F S X E Q b out I F (a) Figure 6. Bipolar mixers with F signal applied to (a) base, (b) emitter of input device. (b) 9

20 Bipolar Mixers F common emitter with degenerate stage in 6.(a) in = β(r π E ) >> 50 Ω, not suitable for heterodyne architecture (image reject filter impedance = 50 Ω) Common base configuration in 6.(b), in = E /g m can be chosen to 50 Ω impedance. oltage and power conversion gain of the mixer in 6.(b) The small- signal collector current of Q : If the F( t) Ic ( t) = S E gm LO-signal is 50% duty cycle: F( t) C 4 out ( t) = cosω g π S E m ( t) cosωlot shift ( ω) by ± ωlo LO t and F 0

21 Bipolar Mixers. ) ( ) ( domain : frequency output in the IF The F IF π ω ω ω C m E S LO g = ( ) ( ) ( ) ( ). gain : conversion voltage The node IF m E C m E S m E m E S C X F g g g g A = = = π π. : input, For matched S C m E S A g π = = ( ) ( ). 4 load : the to power delivered IF average The F, F, IF, IF π π m E S C rms C m E S C rms C rms g g P = = =

22 Bipolar Mixers The The available source F, rms P in =. 4S power conversion PIF 8 AP = = P π For matched A P in input, = π S C power : S. gain : S C ( ). g = S E E g m : m (In general, AP A In the special case, (but rare): C = S, AP = A. out L S A = = P A AP and A have a difference: 0log db db in S L In reality, the conversion gains are lower than above due to parasitics..) S L Simulation!

23 Bipolar Mixers Nonlinearity in mixers LO-F Isolation CC CC LO C Q P Q 3 C nonlinearity ω LO LO C Q P Q 3 C F S Q E noise linearity power trade-off! F S Q E ω LO E : linearity 3

24 Bipolar Mixers Double-balanced bipolar mixers (Gilbert cell) CC Q5 Q6 Q 3 Q 4 LO F Q Q 4

25 Bipolar Mixers Linearization of F port by Schmook s technique I out in A I out na in na A na A I out in G m n = n = 4 n =0 I out ( a) (c) (b) Figure 6.3 (a) Asymmetric differential pair, (b) Schmook s linearized pair, (c) variation of equivalent G m with input level. in in 5

26 Bipolar Mixers I out LO C S F C C 3 I EE Figure 6.4 Mixer using transformer at F input. 6

27 Subharmonic mixers Bipolar Mixers CC CC o 90 o 0 o 70 o 80 o 90 o 0 I I- Q Q- Q Q- I I- Q Q F Q Q F 7

28 CMOS Mixers Active mixers conversion gain noise DD DD Total switch ~00m LO D D F Linearity: O G m, NF M M 3 P (a) M C W, I switch abrupt D LO M 5 M 6 M 3 M 4 Figure 6.5 CMOS active mixer. (b) Figure 6.9 (a) Single-balanced mixer, (b) double-balanced mixer. F M M 8

29 CMOS Mixers Passive mixer F M M LO Drawbak:. no gain, NF (/π or -4dB). M, M size capacitive coupling LO IF LO Advantage:. IIP3 high. no power consumption Turnon point of M F TH on ( a) (b) LO Figure 6.6 (a) Passive CMOS mixer, (b) variation of switch on-resistance. t 9

30 Square-law CMOS mixer CMOS Mixers C L C L F LO M F M Bias LO Bias 30

31 .5GHz Mixer in TLSI um CMOS Power =.8 f F =.4 GHz f IF = 40 MHz Power consumption: 8.4mW NF = db Gain = 7 db IIP 3 = 5.5 dbm GPS application Current-reuse Mixer 3

32 Diode Mixers Double Balanced Diode ing Mixers 3

33 IF Port Design Active Load Differential Output with Common Mode Feedback 33

34 IF Port Design Active Load Differential Output with Common Mode Feedback 34

35 Current-Switched CMOS Mixers Operation Mode DD LO > O L L Saturation mode M turns on and M turns off < < LO O O I F Amplification mode Both M and M turn on DD LO < O L L Saturation mode M turns off and M turns on LO M M I F 35

36 Current-Switched CMOS Mixers Saturation Mode LO > O < IF = ( I F IB ) L IF = ( I F IB ) L Desired signal LO leakage Desired signal LO leakage LO O DD DD L L L L LO M M LO M M I F I F 36

37 Current-Switched CMOS Mixers Amplification Mode < LO < O O v LO = vlo = I = i i F = k LO d d [( v v ) ( v v ) ] O = k 4 i i d = k = k = k O LO v y v y y LO O LO d [( O vlo vy ) ( O vlo vy ) ] [ v v ( v )( v v )] LO O LO LO k v O y LO y LO y LO I I B B v LO v LO i d L M M v y O I F y DD L i d Assume <<, v << IF = ( i k d O LO LO leakage id ) L I LOL F O LO O L Desired signal 37

38 Noise in Mixers F Noise LO Qualitative Analysis IF Noise Figure 6.7 Conceptual view of a mixer. Source of noise a) F path: thermal noise, (r b, r e of Q ), E, collector shot noise of Q b) IF path: C and C c) Q and Q 3 contribute noise SPICE Few timedomain noise analysis F LO CC C C S Q P Q E Q 3 Figure 6.8 (a) Single-balanced mixer. 38

39 Noise in Mixers n F r b C S Q Q I n CC Figure 6.8 (b) noise contribution of Q when Q 3 is off. P E C P LO = Q and/or Q : area r b but C P LO = Q and Q both ON Amplify the thermal noise of r b and I n How to minimize thermal/shot noise of Q and Q :. large LO swings (without saturating). lowering C P, size Q ~Q 3 3. reduce r b, size Q~Q3 4. I C of Q and Q 39

40 Noise in Mixers I C shot noise: I C Noise Q, Q /Q 3 CC Advantage: C C. eduction of noise by Q and Q 3. Given voltage drop on C / C : I C, C / C conversion gain LO Q P I S Q 3 Disadvantage:. I C of Q and Q 3 in into emitters of Q and Q 3 F signal shunt to C P. Noise due to I S added directly into F signal F S Q E C P NF improvement is not significant. Figure 6.9 Addition of current source I S to lower the collector current of Q and Q 3. 40

41 Example in JSSC um CMOS Power = 3 f F =.4 GHz f IF = MHz Power consumption: 30 mw NF = 8.5 db Gain = 5 db IIP 3 = -9 dbm Noise in Mixers Bluetooth application 4

42 Noise in Mixers Noise from LO buffer CC CC LO Double-balanced If output single-ended C C Q P Q 3 Q F S E Figure 6.30 Inclusion of LO output noise in a mixer. 4