17.6 High-Frequency Limitations of Common-Emitter and Common-Source Amplifiers 1309 17.6 HIGH-FREQUENCY LIMITATIONS OF COMMONEMITTER AND COMMON-SOURCE AMPLIFIERS Now that the complete hybrid-pi model has been described, we can explore the high-frequency limitations of the three basic single-stage amplifiers. This section begins with an example of direct analysis of the common-emitter amplifier; the approximation technique for estimating ω H , the open-circuit time-constant method, is then presented. 17.6.1 Example of Direct High-Frequency Analysis — The Common-Emitter Amplifier The common-emitter amplifier circuit discussed earlier in this chapter is redrawn in Fig. 17.31. At high frequencies, the impedance of the coupling and bypass capacitors are negligibly small, and the three capacitors can once again be considered short circuits, resulting in the high-frequency ac model in Fig. 17.31(b). In this figure, R L represents the parallel combination of R3 and RC (100 k4.3 k), the emitter is connected directly to ground through bypass capacitor C3 , and base bias resistor R B equals the parallel combination of R1 and R2 (30 k10 k). Load capacitance C L has been added to the circuit for completeness. In most multistage amplifiers, the common-emitter or common-source amplifiers will be driving an equivalent load capacitance presented by the input impedance of the next amplifier stage. Analysis of the circuit in Fig. 17.31 begins by replacing the transistor with its small-signal model, as in Fig. 17.32. In this figure, the signal source and associated resistors have been replaced VCC = 12 V R2 RI 1 kΩ vI RC 4.3 kΩ C2 30 kΩ C1 0.1 µF R3 100 kΩ 2 µF + vO – CL R1 RE C3 1.3 kΩ 10 µF 10 kΩ (a) Thévenin transformation RI 1 kΩ vi 7.5 kΩ RB RL 4.12 kΩ CL (b) Figure 17.31 (a) Common-emitter amplifier. (b) High-frequency ac model for amplifier in (a). 1310 Chapter 17 Frequency Response Rth Cµ rx v1 rπ vth Cπ gmv1 v2 CL RL Figure 17.32 ac Model for a common-emitter amplifier at high frequencies. Cµ Rth + rx vth rπ is v1 is rπ o gmv1 v1 rπ o v2 CL RL Cπ Figure 17.33 Norton source transformation for Figure 17.34 Simplified small-signal model for the high-frequency C-E amplifier. common-emitter amplifier. by a Thévenin equivalent circuit in which vth = vi RB RI + RB and Rth = RI RB RI + RB (17.72) Direct analysis proceeds using nodal equations. The network in Fig. 17.32 can be reduced to a two-node problem by using a final Norton transformation of the resistors and voltage source attached to node v1 , as indicated in Fig. 17.33. The short-circuit current and Thévenin equivalent resistance for this network are expressed by is = vth Rth + r x rπ o = rπ (Rth + r x ) and (17.73) Figure 17.34 shows the common-emitter amplifier in final simplified form. Writing and simplifying the nodal equations in the frequency domain for the circuit in Fig. 17.34 yields Eq. (17.74): Is (s) 0 = s(Cπ + Cµ ) + gπ o −sCµ −(sCµ − gm ) s(Cµ + C L ) + g L V1 (s) V2 (s) (17.74) An expression for the output voltage, node voltage V2 (s), can be found using Cramer’s rule: V2 (s) = Is (s) (sCµ − gm ) (17.75) in which represents the determinant of the system of equations given by = s 2 [Cπ (Cµ + C L ) + Cµ C L ] + s[Cπ g L + Cµ (gm + gπ o + g L ) + C L gπ o ] + g L gπ o (17.76) From Eqs. (17.75) and (17.76), we see that the high-frequency response is characterized by two poles, one finite zero, and one zero at infinity. The finite zero appears in the right-half of the s-plane at a frequency ωZ = + gm > ωT Cµ (17.77) 17.6 High-Frequency Limitations of Common-Emitter and Common-Source Amplifiers 1311 The zero given by Eq. (17.77) can usually be neglected because it appears at a frequency above ωT (for which the model itself is of questionable validity). Unfortunately, the denominator appears in unfactored polynomial form, and the positions of the poles are more difficult to find. However, good estimates for both pole positions can be found using the approximate factorization technique in Sec. 17.6.2. Note that even though there are three capacitors, the circuit only has two poles. The three capacitors are connected in a “pi” configuration, and only two of the capacitor voltages are independent. Once we know two of the voltages, the third is also defined. 17.6.2 Approximate Polynomial Factorization We estimate the pole locations based on a technique for approximate factorization of polynomials. Let us assume that the polynomial has two real roots a and b: (s + a)(s + b) = s 2 + (a + b)s + ab = s 2 + A1 s + A0 (17.78) If we assume that a dominant root exists — that is, that a b — then the two roots can be estimated directly from coefficients A1 and A0 using two approximations: A1 = a + b ∼ =a a∼ = A1 so, and A0 ab ∼ ab = =b = A1 a+b a b∼ = and (17.79) A0 A1 Note in Eq. (17.78) that the s 2 term is normalized to unity. The approximate factorization technique exemplified by Eqs. (17.78) and (17.79) can be extended to polynomials having any number of widely spaced real roots. For a third-order polynomial, for example, (s + a)(s + b)(s + c) = s 3 + A2 s 2 + A1 s + A0 Assuming c b a, (17.80) c∼ = A0 /A1 b∼ = A1 /A2 , and a ∼ = A2 17.6.3 Poles of the Common-Emitter Amplifier — The C T Approximation For the case of the common-emitter amplifier, the smallest root is the most important because it is the one that limits the frequency response of the amplifier and determines ω H . From Eq. (17.79) we see that the smaller root is given by the ratio of coefficients A0 and A1 : g L gπ o A0 ω P1 ∼ = = A1 Cπ g L + Cµ (gm + gπ o + g L ) + C L gπ o or (17.81) ω P1 1 ∼ = rπ o [Cπ + Cµ (1 + gm R L )] + (Cµ + C L )R L 1312 Chapter 17 Frequency Response The dominant pole is set by two time constants: rπ o times the input capacitance consisting of Cπ and the Miller amplification of Cµ , plus the load resistance times the total capacitance (Cµ + C L ) connected to the output node. We can also view the lower-frequency pole as determined by resistor rπ o and the total effective input capacitance C T expressed by Eq. (17.82): RL RL C T = Cπ + Cµ 1 + gm R L + + CL (17.82) rπ o rπ o C T consists of Cπ plus Cµ multiplied by (1 + gm R L + R L /rπ o ) plus an additional term due to load capacitance C L . The factor multiplying Cµ can be quite large because gm R L represents the intrinsic voltage gain of the common-emitter amplifier and can be expected to be of the order of 10–20VCC . As we will see in Chapter 18, the second pole often plays an important role in frequency compensation of operational amplifiers because the pole can degrade the phase margin of feedback amplifiers. The location of the second pole of the amplifier is estimated from coefficient A1 alone (in properly normalized form): Cπ g L + Cµ (gm + gπ o + g L ) + C L gπ o ω P2 ∼ = Cπ (Cµ + C L ) + Cµ C L or (17.83) ω P2 gm gm ∼ ∼ = = CL Cπ + C L + CL Cπ 1 + Cµ in which the Cµ gm term has been assumed to be the largest term in the numerator, as is most often the case for C-E stages with reasonably high gain. We can interpret the last approximation in Eq. (17.83) in this manner, particularly when Cµ is large. At high frequencies, capacitor Cµ shorts the collector and base of the transistor together so that C L and Cπ appear in parallel, and the transistor behaves as a diode with resistance of 1/gm . Overall Transfer Function for the Common-Emitter Amplifier An overall expression for the gain of the common-emitter amplifier can be obtained by combining Eqs. (17.72) to (17.76), (17.81), and (17.83): Vo (s) = Vth (s) Rth + r x g L gπ o (sCµ − gm ) s s 1+ 1+ ω P1 ω P2 and s 1− Vth (s) ω Z Vo (s) = (−gm R L rπ o ) s s Rth + r x 1+ 1+ ω P1 ω P2 (17.84) For C L = 0, ω Z and ω P2 are both greater than ωT , and Eq. (17.84) can be approximated by Vth (s) (gm R L rπ o ) Vo (s) ∼ =− s Rth + r x 1+ ω P1 (17.85) 17.6 1313 High-Frequency Limitations of Common-Emitter and Common-Source Amplifiers Recognizing that rπ o = rπ (Rth + r x )/(Rth + r x + rπ ) yields βo R L Vo (s) ∼ Rth + r x + rπ Avth (s) = =− s Vth (s) 1+ ω P1 (17.86) or Avth (s) ∼ = Amid s 1+ ω P1 (17.87) in which Amid = − βo R L Rth + r x + rπ and ω P1 = 1 rπ o C T (17.88) Equation (17.87) indicates that the high-frequency behavior of the common-emitter amplifier can be modeled by a single dominant pole, as in the circuit in Fig. 17.35. Rth + rx vth rπ v1 CT gmv1 vo RL Figure 17.35 Dominant-pole model for the common-emitter amplifier at high frequencies. Before we leave this section, it should be noted again that we have neglected the right-halfplane zero ω Z that appears in Eq. (17.84). Since ω Z is high in frequency (beyond ωT ), it has minimal impact on the magnitude of the transfer function. However, the additional phase shift that this zero introduces at high frequency can degrade the stability of feedback amplifiers that we study in Chapter 18. EXAMPLE 17.6 HIGH-FREQUENCY ANALYSIS OF THE COMMON-EMITTER AMPLIFIER Find the midband gain and upper-cutoff frequency of a common-emitter amplifier. PROBLEM Find the midband gain and upper-cutoff frequency of the common-emitter amplifier in Fig. 17.31 using the C T approximation, assuming βo = 100, f T = 500 MHz, Cµ = 0.5 pF, r x = 250 , and a Q-point of (1.60 mA, 3.00 V). Find the additional poles and zeros of the common-emitter amplifier. Assume C L = 0. SOLUTION Known Information and Given Data: Common-emitter amplifier circuit in Fig. 17.31; Q-point = (1.60 mA, 3.00 V); βo = 100, f T = 500 MHz, Cµ = 0.5 pF, and r x = 250 ; expressions for the gain, poles, and zeros are given in Eqs. (17.81) through (17.88). Unknowns: Values for Amid , f H , ω Z 1 , ω P1 , and ω P2 Approach: Find the small-signal parameters for the transistor. Find the unknowns by substituting the given and computed values into the expressions developed in the text. 1314 Chapter 17 Frequency Response Assumptions: Small-signal operation in the active region; VT = 25.0 mV; C L = 0 Analysis: The common-emitter stage is characterized by Eqs. (17.82), (17.83), and (17.88). βo R L gm Amid = − ωZ = + Rth + r x + rπ Cµ 1 1 1 gm 1 1+ ω P1 = ω P2 = + + rπ o C T R L Cµ Cπ gm rπ o gm R L RL rπ o = rπ (Rth + r x ) C T = Cπ + Cµ 1 + gm R L + rπ o The values of the various small-signal parameters must be found: βo 100 gm = 40IC = 40(0.0016) = 64.0 mS rπ = = = 1.56 k gm 0.064 gm 0.064 Cπ = − Cµ = − 0.5 × 10−12 = 19.9 pF 2π f T 2π(5 × 108 ) R L = RC R3 = 4.3 k100 k = 4.12 k Rth = R B R I = 7.5 k1 k = 882 rπ o = rπ (Rth + r x ) = 1.56 k(882 + 250 ) = 656 Substituting these values into the expression for C T yields RL C T = Cπ + Cµ 1 + gm R L + rπ o 4120 = 19.9 pF + 0.5 pF 1 + 0.064(4120) + 656 = 19.9 pF + 0.5 pF(1 + 264 + 6.28) = 156 pF and 1 1 = = 1.56 MHz 2πrπ o C T 2π(656 )(156 pF) 1 1 gm 1 1+ = + + R L Cµ Cπ gm rπ o gm R L 1 0.064 1 1 = + 1+ + (4.12 k)(0.5 pF) 19.9 pF 0.064(656) 0.064(4120) ω P2 f P2 = = 603 MHz 2π gm 0.064 fz = = = 20.4 GHz 2πCµ 2π(0.5 pF) f P1 = ω P2 βo R L 100(4120) = −153 =− Rth + r x + rπ 882 + 250 + 1560 Remember that the overall gain from vs to the output will be reduced to Amid = 0.882(−153) = −135 since vth = 0.882vs . Avth = − Check of Results: We have found the desired information. By double checking, the calculations appear correct. Let us use the gain-bandwidth product as an additional check: |Amid f H | = 239 MHz, which does not exceed f T . 17.6 High-Frequency Limitations of Common-Emitter and Common-Source Amplifiers 1315 Discussion: The dominant pole is located at a frequency f P1 = 1.56 MHz, whereas f P2 and f Z are estimated to be at frequencies above f T (500 MHz). Thus, the upper-cutoff frequency f H for this amplifier is determined solely by f P1 : f H ∼ = 1.56 MHz. Note that this value of f H is less than 1 percent of the transistor f T and is consistent with the concept of GBW product. We should expect f H to be no more than f T /Amid = 3.3 MHz for this amplifier. Note also that f P1 and f P2 are separated by a factor of almost 1000, clearly satisfying the requirement for widely spaced roots that was used in the approximate factorization. It is important to keep in mind that the most important factor in determining the value of C T is the term in which Cµ is multiplied by gm R L . To increase the upper-cutoff frequency f H of this amplifier, the gain (gm R L ) must be reduced; a direct trade-off must occur between amplifier gain and bandwidth. Computer-Aided Analysis: SPICE can be used to check our hand analysis, but we must define the device parameters that match our analysis. We set BF = 100 and IS = 5 fA, but let VAF default to infinity. The base resistance r x must be added by setting SPICE parameter RB = 250 . Cµ is determined in SPICE from the value of the zero-bias collector-junction capacitance CJC and the built-in potential φ jc . The Q-point from SPICE gives VC E = 2.70 V, which corresponds to VBC = 2.0 V if VB E = 0.7 V. In SPICE, VJC faults to 0.75 V and MJC defaults to 0.33 (see Sec. 17.4). Therefore, to achieve Cµ = 0.5 pF, CJC is specified as 20 V 0.33 CJC = 0.5 pF 1 + = 0.768 pF 0.75 V Cπ is determined by the SPICE forward transit-time parameter TF as defined by Eqs. (5.39) and (17.50): TF = Cπ 19.9 pF = 0.311 ns = gm 64 mS After adding these values to the transistor model, we perform an ac analysis using FSTART = 100 Hz and FSTOP = 10 MHz with 20 frequency points per decade. The SPICE simulation results in the graph below yield Amid = −135 (42.6 dB) and f H ∼ = 1.56 MHz, which agrees closely with our hand calculations. Checking the device parameters in SPICE, we also find r x = 250 , Cπ = 19.9 pF, and Cµ = 0.499 pF, as desired. Av (dB) +45 +40 fH +35 +30 +25 +20 +100 +1000 +10K +100K Frequency (Hz) +1Meg +10Meg 1316 Chapter 17 Frequency Response Exercise: Use SPICE to recalculate f H for V A = 75 V. For V A = 75 V and r x = 0? Answer: 1.67 MHz; 1.96 MHz Exercise: Repeat the calculations in Ex. 17.6 if a load capacitance CL = 3 pF is added to the circuit. Answers: 1.39 MHz, 71.6 MHz (A significant decrease in f P2 !) Exercise: Find the midband gain and the frequencies of the poles and zeros of the common-emitter amplifier in Ex. 17.6 if the transistor has fT = 500 MHz, but Cµ = 1 pF. Answers: −153, 835 kHz, 578 MHz, 10.2 GHz 17.6.4 Gain-Bandwidth Product Limitations of the Common-Emitter Amplifier The important role of r x in ultimately limiting the frequency response of the C-E amplifier should also not be overlooked. If the Thévenin equivalent source resistance Rth were reduced to zero in an attempt to increase the bandwidth, then rπ o would not become zero but would be limited approximately to the value of r x . Let us look at the asymptotic behavior of the gain-bandwidth product of the C-E amplifier, GBW = |Av ω H | ≤ βo R L Rth + r x + rπ 1 rπ o C T (17.89) using these approximations: Rth = 0, r x rπ so that rπ o ∼ = r x and C T ∼ = Cµ (gm R L ) (17.90) 1 r x Cµ (17.91) Substituting these values in Eq. (17.89) yields GBW ≤ The product of the base resistance r x and the collector-base capacitance Cµ places an upper bound on the gain-bandwidth product of the C-E amplifier. From Eq. (17.91), we can see that one important consideration in the design of transistors for very high frequency operation is minimization of the r x Cµ product. Exercise: Compare the gain-bandwidth product of the amplifier in Ex. 17.6 to the limit in Eq. (17.91) and to the fT of the transistor. Answers: 239 MHz < 1.27 GHz, 239 MHz ∼ = 500 MHz/2 17.7 1317 Miller Multiplication C2 RI C1 M 1 kΩ vi 0.1 µF RG RD 0.1 µF R3 vo 100 kΩ 4.3 kΩ 996 Ω 243 kΩ RS Rth C3 1.3 kΩ vth v1 10 µF CGD CGS RL v2 4.12 kΩ gmv1 (b) (a) Figure 17.36 (a) Common-source amplifier. (b) The high-frequency small-signal model. 17.6.5 Dominant Pole for the Common-Source Amplifier Analysis of the C-S amplifier in Fig. 17.36 mirrors that of the common-emitter amplifier. The small-signal model is similar to that for the C-E stage, except that both r x and rπ are absent from the model. For Fig. 17.36(b), Rth = R I RG R L = R D R3 vth = vi RG R I + RG (17.92) The expressions for the finite zero and poles of the C-S amplifier can be found by comparing Fig. 17.36(b) to Fig. 17.34: ω P1 = ω P2 1 Rth C T and 1 gm = + RL CG D CG S RL C T = C G S + C G D 1 + gm R L + Rth 1 1 gm 1+ ωZ = + gm Rth gm R L CG D (17.93) Exercise: What is the upper-cutoff frequency for the amplifier in Fig. 17.36 if CGS = 10 pF, CG D = 2 pF, and gm = 1.23 mS? What are the positions of the second pole and the zero? What is the fT of this transistor? Answers: 5.26 MHz; 58.7 MHz, 97.9 MHz; 16.3 MHz 17.7 MILLER MULTIPLICATION The amplification of Cµ by the voltage gain of the C-E amplifier is an example of Miller multiplication. Miller originally derived an expression for the input admittance of the general amplifier shown in Fig. 17.37, in which capacitor C is connected between the input and output terminals of an inverting amplifier. Let us calculate the input admittance of this amplifier. The output voltage vo is equal to the negative gain times the input voltage vi , and the input current to the amplifier is the admittance of the capacitor (sC) multiplied by the total voltage across the capacitor: Vo (s) = −AVI (s) and Is (s) = sC [VI (s) − Vo (s)] (17.94)