Lecture 4c Determination of Poles & Zeros of CS Amplifiers (2) Objectives To calculate different poles and zeros which determine the frequency response of CS amplifiers To calculate the dominant poles and the amplifier bandwidth Direct Determination of High-Frequency Poles and Zeros: The high frequency poles and zeros may be determined by direct analysis. In this case all the coupling and bypass capacitors may be considered short circuit. However, the high frequency model of the transistor should be used since the transistor internal capacitances can't considered open circuit at high frequencies. The following example will illustrate this method for common source amplifier. Example 1: Drive an expression for the high frequency response AvH(s) of the common source amplifier shown in Figure 1. Hence, determine the midband gain Amid, high frequency poles and zeros, and the high cutoff frequency. Figure 1: Common source amplifier Solution: The CS amplifier is redrawn for small signal analysis as shown in Figure 2 by grounding the DC voltage source VDD and replacing the transistor by its high frequency small signal model. The biasing resistors R1 and R2 are combined into RG. ∴ RG = R1 || R 2 All the coupling and bypass capacitors are considered short circuit at high frequency. Figure 2: The CS amplifier in Figure 1 redrawn for small signal analysis with the transistor replaced by its high frequency small signal model The small-signal model can be simplified by using Thevinin Theorem as shown in Figure 3. v th = v sig RG R sig + RG and Rth = R sig RG R sig + RG Also, R L′ = R L || R D Figure 3: Simplified circuit to calculate the frequency response of the CS amplifier Writing the two nodal equations at nodes A and B in frequency domain, vo + g mv gs + (v o −v gs )sC GD = 0 R L′ v gs sC GD + v gs −v th R th + (v gs −v o )sC GD = 0 Solving the two nodal equations found in the preceding slide by eliminating vgs yields, V o (s) = v th (s) (sC GD - g m ) Rth Δ ⎛C ⎛ 1 1 ⎞⎞ 1 Δ = s 2 (C GS C GD ) + s ⎜ GS + C GD ⎜ g m + + ⎟ ⎟⎟ + ⎜ R′ R th R L′ ⎠ ⎠ R L′ .R th ⎝ ⎝ L ∴ AvH (s ) = v o (s ) v sig (s ) ∴ AvH (s ) = 1 (sC GD - g m ) R sig Δ Let us define CT ⎛ R′ ⎞ = C GS + C GD ⎜ 1 + g m R L′ + L ⎟ R th ⎠ ⎝ ∴ Δ = s 2C GS C GD + ∴ AvH (s ) = sCT 1 + R L′ R ′.R th 1 R sig C GS .C GD s 2 + (sC GD - g m ) sC T 1 + R L′ .C GS .C GD R ′.R th .C GS .C GD vo (s) = v th (s) (sC GD - g m ) Δ R th ⎛C ⎛ 1 1 Δ = s 2 (C GS C GD ) + s ⎜ GS + C GD ⎜ g m + + ⎜ R′ R th R L′ ⎝ ⎝ L v o (s ) 1 (sC GD - g m ) AvH (s ) = = v sig (s ) R sig Δ ∴ ⎞⎞ 1 ⎟ ⎟⎟ + ⎠ ⎠ R L′ .R th ⎛ R′ ⎞ = C GS + C GD ⎜1 + g m R L′ + L ⎟ R th ⎠ ⎝ sC 1 Δ = s 2C GS C GD + T + R L′ R ′.R th Let us define C T ∴ ∴ 1 AvH (s ) = R sig C GS .C GD s 2 + (sC GD - g m ) sC T 1 + R L′ .C GS .C GD R ′.R th .C GS .C GD Amid may be found from the last expression by assuming s → 0 . ∴ A mid = - g m R L′ .RG R sig + RG High-frequency response is given by 2 poles, one finite zero and one zero at infinity. Finite right-half plane zero, ωZ = + gm/CGD > ωT can easily be neglected. For a polynomial s2 + sA1 + A0 with roots a and b, if one root is much larger than the other one we can assume that a = A1 and b = A0/A1. ∴ ωP 1 = ωP 1 ≅ 1 R thC T A0 A1 Assume that the midband gain is very large ⎛ R′ ⎞ ( i.e. g m R L′ ⎜1 + L ⎟ , and g m R L′C GD C GS ) ⎝ Rth ⎠ gm ⎛ 1 1 ⎞ + ωP 2 ≅ ⎜1 + ⎟ C GS ⎝ g m Rth g m R L′ ⎠ gm ωP 2 ≅ C GS ∴ ∴ ∴ ωP 1 = A0 1 ≅ A1 R thC T ⎛ R′ Assume that the midband gain is very large ( i.e. g m R L′ ⎜ 1 + L ⎝ R th gm ⎛ gm 1 1 ⎞ ωP 2 ≅ + ⎜1 + ⎟ ≅ C GS ⎝ g m R th g m R L′ ⎠ C GS ⎞ ⎟ , and ⎠ g m R L′C GD C GS ) ∴ Smallest root that gives first pole limits frequency response and determines ωH. Second pole is important in frequency compensation as it can degrade phase margin of feedback amplifiers. Assuming ωp1 ωp2 (ωp1 is the dominanat pole) ∴ Assuming ωp1 ∴ 1 RthCT ωp2 (ωp1 is the dominanat pole) ωH ≅ ωP 1 = ωH ≅ ωP 1 = 1 R thC T Dominant pole model at high frequencies for CS amplifier is shown in Figure 4. Figure 4: Equivalent circuit assuming dominant pole model Miller’s Theorem Another way to find the high frequency poles is by calculating the equivalent capacitance for CGD at the amplifier input. This may be done using Miller's Theorem as shown next. For the circuit shown we can state the following. Figure 5: Miller's Theorem v o (s ) = − Av 1 (s ) i s (s ) = sC [v 1 (s ) − v o (s ) ] i s (s ) = sC (1 + A ) v 1 (s ) v o (s ) = − Av 1 (s ) Y (s ) = i s (s ) = sC [v 1 (s ) −v o (s ) ] Y (s ) = i s (s ) = sC (1 + A ) v 1 (s ) The total input capacitance = C (1+A) because total voltage across C is vc = vi (1+A) due to the inverting voltage gain of amplifier. Applying Miller's Theorem to the CS Amplifier For the common source amplifier referring to Figure 6 we may write. CT CT = C GS + C GD (1 + A ) = C GS + C GD (1 + g m R L ) CT = C GS + C GD (1 + A ) = C GS + C GD (1 + g m R L ) Figure 16a: Applying Miller's theorem to the CS amplifier Figure 6b: Resulting circuit from applying Miller's theorem to the CS amplifier Open-Circuit Time Constant Method to Determine ωH At high frequencies, impedances of the coupling and bypass capacitors are small enough to be considered short circuits. Open-circuit time constants associated with the impedances of device capacitances are considered instead. ωH ≅ 1 m ∑ R ioC i i =1 where Rio is the resistance at the terminals of the ith capacitor Ci with all other capacitors open-circuited. Example 2: Derive an expression for the high cutoff frequency for the circuit in Example 1 using the OCTC method. Solution: Using OCTC method for the circuit in Figure 3. For CGS the resistance seen across the capacitor terminals may be calculated from Figure 7. RGSO = R th So the time constant associated with CGS equals CGS. Rth Figure 7: Circuit used to calculate the open-circuit time constant associated with CGS For CGD the resistance seen across the capacitor terminals may be calculated from Figure 8. vx ix RGDO = RGDO = Rth ⋅ (1 + g m R L′ + R L′ ) Rth vx R′ = R th .(1 + g m R L′ + L ) ix R th RGDO = So the time constant associated with CGD equals CGD · Rth (1 + gmR`L + R`L/Rth) Figure 8: Circuit used to calculate the open-circuit time constant associated with CGD The high cutoff frequency is calculated using the SCTC as: ωH ≅ 1 2 ∑ R ioC i i =1 1 + RGDOC GD ωH ≅ ωH ωH RGSOC GS 1 ≅ R thCT 1 1 ≅ 2 = ∑ R ioC i RGSO C GS + RGDO C GD i =1 = 1 R thC T