Frequency Response of Amplifiers •General Considerations • Miller Effect • Association of Poles with Nodes •Common Source Stage •Source Follower •Differential Pair Hassan Aboushady University of Paris VI References • B. Razavi, “Design of Analog CMOS Integrated Circuits”, McGraw-Hill, 2001. H. Aboushady University of Paris VI 1 Miller Effect • Miller’s Theorem with • Proof Av = VY VX we have Z1 = Z2 = Z 1 − Av−1 Z V 1− Y VX V X − VY V X = Z Z1 Z1 = VY − V X VY = Z Z2 Z2 = H. Aboushady Z 1 − Av Z V 1− X VY University of Paris VI Example 1 • Calculate the input capacitance Cin: Z= 1 sC F 1 sC F Z1 = 1+ A Cin = C F (1 + A) Av = VY should be calculated at the frequency of interest. VX To simplify calculations we usually use low frequency value of Av. Miller’s theorem cannot be used simultaneously to calculate input-output transfer function and the output impedance. H. Aboushady University of Paris VI 2 Association of Poles with Nodes Vout2 Vout1 VM ( s ) = H. Aboushady Vin ( s ) 1 Vin ( s ) = 1 sCin 1 + sRS Cin RS + sCin VN ( s ) = Vout1 ( s ) 1 + sR1C N VP ( s ) = Vout 2 ( s ) 1 + sR2C P Vout A1 A2 1 ( s) = Vin 1 + sRS Cin 1 + sR1C N 1 + sR2C P University of Paris VI Association of Poles with Nodes Vout2 Vout1 Vout A1 A2 1 ( s) = Vin 1 + sRS Cin 1 + sR1C N 1 + sR2C P ω1 = 1 RS Cin ω2 = 1 R1C N ω3 = 1 R2C P 3 poles: each determined by the total capacitance seen from each node to ground multiplied by the total resistance seen at the node to ground H. Aboushady University of Paris VI 3 Example 2 • Calculate the pole associated with node X: The total equivalent capacitance seen from X to ground: C = C (1 + A) X The pole frequency: F ωX = 1 1 = RS C X RS C F (1 + A) H. Aboushady University of Paris VI Example 3 • Neglecting channel length modulation, compute the transfer function of the common gate stage with parasitic capacitances Parasitic capacitances at node X: Input resistance of a common gate amplifier: Pole frequency at node X: Rin = C S = CGS 1 + CSB1 1 g m1 + g mb1 ωX = 1 ⎛ (CGS1 + CSB1 )⎜⎜ RS // ⎝ Parasitic capacitances at node Y: Pole frequency at node X: H. Aboushady ωY = ⎞ 1 ⎟ g m1 + g mb1 ⎟⎠ C D = C DG1 + C DB1 1 (C DG1 + CDB1 )RD University of Paris VI 4 Example 3 (cont.) Low-frequency gain of a common gate stage neglecting channel length modulation: Av 0 = (g m + g mb )RD 1 + (g m1 + g mb1 )RS The overall transfer function is given by: Vout ( s ) Av 0 = Vin ( s ) ⎛ s ⎞⎛ s ⎟⎟⎜⎜1 + ⎜⎜1 + ⎝ ωin ⎠⎝ ωout ⎞ ⎟⎟ ⎠ Note that if we do not neglect r01 , the input and output nodes interact, making it difficult to calculate the poles. H. Aboushady University of Paris VI Common Source Stage Neglecting channel length modulation and applying the Miller’s theorem on CGD , we have: The total capacitance at node X: C X = CGS + (1 − Av )CGD where, Av = − g m RD The 1st pole frequency: ω p1 = RS (CGS 1 + (1 + g m RD )CGD ) The total capacitance at the output node: ( ) Cout = C DB + 1 − Av−1 CGD ≈ C DB + CGD The 2nd pole frequency: H. Aboushady ω p2 = 1 RD (C DB + CGD ) University of Paris VI 5 Common Source Stage The transfer function: Vout ( s ) − g m RD = Vin ( s ) ⎛ ⎞ ⎞⎛ ⎜1 + s ⎟⎜1 + s ⎟ ⎜ ω ⎟⎜ ω ⎟ p1 ⎠⎝ p2 ⎠ ⎝ r0 and any load capacitance can be easily included. Sources of error (approximation): • we have not considered the existence of zeros in the circuit • the amplifier gain varies with frequency H. Aboushady University of Paris VI Common Source : “exact “ Transfer Function To obtain the exact transfer function: Applying Kirchoff Current Law (KCL): VX − Vin + sCGSVX + sCGD (VX − Vout ) = 0 RS ⎛ 1 ⎞ ⎟⎟Vout = 0 sCGD (Vout − V X ) + g mV X + ⎜⎜ sC DB + R D ⎠ ⎝ H. Aboushady University of Paris VI 6 Common Source : “exact” 1st pole After some manipulations, we get: (sCGD − g m )RD Vout = 2 Vin RS RDξ s + [RS (1 + g m RD )CGD + RS CGD + RS CGS + RD (CGD + C DB )]s + 1 with ξ = CGS CGD + CGS C DB + CGD C DB Writing the denominator as: Assuming ⎛ 1 ⎛ 1 ⎞⎟ s ⎞⎟⎛⎜ s ⎞⎟ s2 1+ = +⎜ + D = ⎜1 + s +1 ⎟ ⎜ω ⎜ ω ⎟⎜ ω ⎟ ω ω p1 ⎠⎝ p1 ⎠ p1 p 2 ⎝ p1 ω p 2 ⎠ ⎝ ω p1 << ω p 2 ω p1 ≈ 1 RS (1 + g m RD )CGD + RS CGD + RS CGS + RD (CGD + C DB ) Compare this result with ωin calculated using Miller’s Theorem H. Aboushady University of Paris VI Common Source : “exact” 2nd pole (sCGD − g m )RD Vout = 2 Vin RS RDξ s + [RS (1 + g m RD )CGD + RS CGD + RS CGS + RD (CGD + C DB )]s + 1 with having and then ξ = CGS CGD + CGS C DB + CGD C DB ⎛ 1 1 ⎞⎟ +⎜ + s +1 ω p1ω p 2 ⎜⎝ ω p1 ω p 2 ⎟⎠ 1 ω p1 ≈ RS (1 + g m RD )CGD + RS CGD + RS CGS + RD (CGD + C DB ) D= s2 ω p2 = 1 1 RS RDξ ω p1 ω p2 = RS (1 + g m RD )CGD + RS CGS + RD (CGD + C DB ) RS RD (CGS CGD + CGS C DB + CGD C DB ) H. Aboushady University of Paris VI 7 Comparison between “exact” and Miller’s theorem 1st pole: exact If RD (CGD + C DB ) is negligible 2nd pole: Miller exact ω p1 = ω p1 = ω p2 = RS (CGS 1 + (1 + g m RD )CGD ) + RD (CGD + C DB ) RS (CGS 1 + (1 + g m RD )CGD ) RS (1 + g m RD )CGD + RS CGS + RD (CGD + C DB ) RS RD (CGS CGD + CGS C DB + CGD C DB ) if CGS >> (1 + g m RD )CGD + ω p2 ≈ H. Aboushady Miller ω p2 = RD (CGD + CDB ) RS CGS RD (CGS CGD + CGS C DB + CGD C DB ) 1 RD (C DB + CGD ) University of Paris VI Common Source : transfer function zero After some manipulations, we get: (sCGD − g m )RD Vout = 2 Vin RS RDξ s + [RS (1 + g m RD )CGD + RS CGD + RS CGS + RD (CGD + C DB )]s + 1 ωz = H. Aboushady gm CGD University of Paris VI 8 Source Follower High frequency equivalent circuit X Applying Kirchoff Current Law (KCL) at the output node: sCGSV1 + g mV1 = sC LVout Vout + V1 − Vin + sCGD (Vout + V1 ) + sCGSV1 = 0 RS Vout ( s ) g m + sCGS = Vin ( s ) RS (CGS C L + CGS CGD + CGD C L )s 2 + ( g m RS CGD + C L + CGS )s + g m at node X: H. Aboushady University of Paris VI Source Follower Input Impedance VX = ⎛ IX g I + ⎜⎜ I X + m X sCGS ⎝ sCGS ⎞ 1 ⎟⎟ ⎠ sC L Input Impedance: Z in = g 1 1 + + 2 m sCGS sC L s CGS C L Note the negative resistance: H. Aboushady − gm ω CGS C L 2 University of Paris VI 9 Source Follower Output Impedance Neglecting CGD Applying KCL sCGSV1 + g mV1 + I X = 0 RS V1 + sCGS (V1 − V X ) = 0 RS 1/gm Output Impedance: Z out = sCGS RS + 1 sCGS + g m Zout increases with frequency It contains an inductive component H. Aboushady University of Paris VI Source Follower Output Impedance Equivalent C t Equivalent circuit of source follower output impedance: at ω = ∞ ⇒ Z1 = R1 + R2 at ω = 0 ⇒ Z1 = R2 Z1 = sLR1 + R2 sL + R1 ⎛ 1 ⎞ ⎟⎟ R1 = ⎜⎜ RS − g m ⎠ ⎝ H. Aboushady ⎛1 1 ⎞ sL⎜⎜ + ⎟⎟ + 1 R R2 ⎠ Z1 = ⎝ 1 sL 1 + R1 R2 R2 R2 = 1 gm L= Z out = CGS gm sCGS RS + 1 sCGS + g m ⎛ 1 ⎞ ⎜⎜ RS − ⎟⎟ g m ⎠ ⎝ University of Paris VI 10 Differential Pair Differential inputs: same as common source stage H. Aboushady University of Paris VI Differential Pair Common-Mode inputs: Low frequency ACM with M1-M2 mismatch: ACM = ∆g m R D (g m1 + g m 2 )rO 3 + 1 High frequency ACM with M1-M2 mismatch: ACM where H. Aboushady ⎛ 1 ⎞ ⎟ ∆g m ⎜⎜ RD // sC L ⎟⎠ ⎝ = ⎛ ⎞ (g m1 + g m 2 )⎜⎜ rO 3 // 1 ⎟⎟ + 1 sC P ⎠ ⎝ C P = C DG 3 + C DB 3 + C SB1 + CSB 2 University of Paris VI 11