LIMITATIONS ON POWER GA IN IN FEEDBACK AMPLIFIERS by RONALD L.ESLIE 'l'HOMPSON SUBMITTED IN PARTIAL FULFII..LMENT OF THE 1 RE'~UIREMJ~N rs FOR rHE DEGREE OF 1 MASTER OF SCIENCE at the MASSACHUSETI'S INSTITlYrE OF TECHNOLOGY August, 1953 • Signature redacted Signature of Author ,_,.,.___ . ac: ._. _ . _ Department or'Electrica.l Engineer1ns7Auguet 24, 1953 certified by Signature redacted -------------·---~----rc-..t'~'---T-k_ie_s_i~---S-u_pe_r--v-i~s-or--- Signature redacted Mlllibraries Document Services Room 14-0551 77 Massachusetts Avenue Cambridge, MA 02139 Ph: 617.253.2800 Email: docs@mit.edu http://libraries.mit.edu/docs DISCLAIMER OF QUALITY Due to the condition of the original material, there are unavoidable flaws in this reproduction. We have made every effort possible to provide you with the best copy available. If you are dissatisfied with this product and find it unusable, please contact Document Services as soon as possible. Thank you. Some pages in the original document contain text that runs off the edge of the page. 38 . LIMI'[A'rIOJS o:·: I i~ ro:~ER GAIN 1-'Er.;"CB.'\CK AMPLIFIERS by RONALD LESLIE THOMPSON Submitted to the Depar tment of Electrical Engineering on August 24, 1953 in partial fulfillment of the requirements for the degr_ee of Master of Science. ABS'l'RAC'f A method of maximizing the power gain in feedback amplifiers, subject to a given stability margin, is investigated. As an example, this method is applied to a grounded base junction transistor amplifier and to a grounded emitter junction transistor amplifier. A second function, the unilateral amplification factor, U, is also considered. This function is plotted for the junction transistor amplifier, and this curve is compared with the gain curves of the transistor feedback circuit. Derivation of the equations in this paper is given in the appendix at the end of chapter 3. Thesis Supervisor: s. J. Mason Title: Assistant Professor of Electrical Engineering -ii- ACKNOWLEDGMl~ N'rs •rhe author is indebte d to t.he super visor of this the1,;1 s, Profes sor s. J. Mason, for his guidance in the choice of a sub ject which proved to be very interesting. I would also like to thank him for the many discussions we had, in which the general direction of the research was established. -111- TABLE (I r CON'l'iN'rS Page ABS1'.RACT .......... .. ...... ..... ... .. ... ........ .. ............... ... ...... u- ACKNOWLEDGEMENT Cht1.pter 1. Chapte r 2. iii IUTRODUC'l'ION •.• . . • •. 1 . 1 Gain and StabiliLy 1.2 Purpose of th is Investigation GAIN AND STABILITY CONSIDERATIONS 2.1 General Repl'esentation of Two-Terminal Pair Device 2.2 Stability 2.3 Gain Maximization 2.4 Unilateral Amplification Factor ............. .................................... ......... ...... .......................... Chapter 3. .......... .... ... ... .. .......... ...... ...... ........ ... ............ ......... ............ ..... ................ ............ ... ......... ... ..... ..... .. .... ... ....... .... .......... TRANSISTOR GAIN MAXIMIZA'rION .. . . . .. Introduction Grounded Base Connection Grounded Emitter Connection Determination of u .. 3. 1 3.2 3.3 3.4 CONCLUDING REMARKS ... . . . . . APPENDIX I. Grounded Base II. Grounded Emitter III. U •••• •• •• BIBLIOGRAPHY ........ .................... .......... ........ ....... ....................... .......... .. .... ................ ............................. -iv- l 1 2 3 3 4 5 10 12 12 12 19 25 28 29 29 32 35 LIST OF ILLLJSINJTIONS Fiirre Page 1 General Amplifier Circuit ..................................... 2 Flow Graph for General Amplifier................................4 3 Reduction of Flow Graph ........................................ 4 Vector Diagram Showing Stability Margin ......................... 8 5 Network With Arbitrary Feedback ................................ 10 6 Transformed Unilateral Network ................................. 11 7 Grounded Base Junction Transistor .............................. 12 8 10 Log 9 Grounded Base M, k, and H Curves ..................... 01 11 IHI 3 and OH Grounded'Base Junction Transistor ............ 0 log (1+r)2 vs r 4 14 15 ................................... 17 Maximum Permissible Gain Grounded Base Junction Transistor Case Two ................................... ................. 18 12 Grounded Emitter Junction Transistor 13 10 Log )Hj and eH Grounded Emitter Junction Transistor ......... 21 14 Curves of M, k, and 10 Log 23 15 Maximum Permissible Gain Grounded Emitter Junction Transistor Case Two ............... ................ 24 Unilateral Amplification Factor For Junction Transistor ........ 26 -v - 16 IHI .... Grounded Emitter 20 LTST OF TABIES Table 1 Values of ZG and ZL for Grounded Base Amplifier ............... 2 Values of ZG and ZL for Grounded Emitter .................... -vi- 19 CHAPTER I INTRODUCTI'ON 1.1 Gain and Stability Basic considerations in designing any network are the amount of gain that is available from the given network configuration and whether the circuit will be stable when put into operation. As instability is approached in any network, the gain approaches infinity, so that some compromise between the two must be reached. For analysis of a circuit, the network can be represented as a two-terminalpair device. These devices can be passive circuits, unilateral active circuits, or non-unilateral active circuits. It is this latter type circuit that is of interest. The non-unilateral active circuit is characterized by the fact that it has appreciable internal coupling between the output and input terminals of the amplifier. This coupling, if it is of the right phase and magnitude, leads to regenerative feedback and instability of the network. With the advent of the transistor, interest in non-unilateral active networks has greatly increased. Very little has been written on gain maximization in networks of this type, and what has been written has considered only those. cases in which stability was not a problem. With an unconditionally stable network, the gain was maximized by synthesizing the load and source impedances on a conjugate image impedance match basis. Professor S. J. Mason, of the Massachusetts Institute of Technology, has written a paper1 concerning the problem of power gain maximization in nonunilateral active networks in which there is a stability problem. Mason, S. J., "Power Gain in Feedback Amplifiers", RLE Report, July 1952. -l - 1. 1.2 Purpose of this Investigation The purpose of this investigation is to see how the methods of analysis that Professor Mason has put forward are applied, with specific reference to transistor circuits. The small signal incremental equivalent circuit for a transistor will be used in all the discussions of the specific circuit. -2- CHAPTER I GAIN AND STABILITY CONSIDERATIONS 2.1 General Representation of Two-Terminal-Pair Device The -two-terminal-pair network that will be considered here can be represented as shown in Figure 1. ZG +(L, The impedances that specify the network 7 ~z 12 EZ ZL Z? z2 1 N GENERAL AMPLIFIER CIRCUIT FIGURE 1. N are complex quantities. E E2 = 1Zi The equations characterizing such a device are + 12 Z 1 2 = EG 11 Z 2 1 + 12 Z2 2 = G I2ZL (2-1) (2-2) Because of the active nature of the network N, Z2 1 does not equal Z.. Equations (2-1) and (2-2) can be represented graphically by a signal flow graph,l as in Figure 2. 1. Mason, S. J., "On the Logic of Feedback", MIT E.E. Sc. D. Thesis, 1951. -3- z_12 I E____ 312 ____ Z,,+ Z' Z2 22+ ZL FLOW GRAPH FOR GENERAL AMPLIFIER FIGURE 2. This graph gives a clear picture of equations (2-1) and (2-2), and lends itself to an easy solution of them. 2.2 Stability In any amplifier, the gain is a function of the transfer admittance or impedance of the network. By simplifying the graph of Figure 2, the transfer admittance of the network is plainly shown. two steps in Figure 3. This simplification is done in The resulting solution for the transfer admittance T (o.Z11 (~~ %%(Zz~ZL(I T) (b) FIGURE 3. REDUCTION OF FLOW. GRAPH -4-. Zia) is easily seen from Figure 3(b) to be ( - 1 ) (2-3) An examination of equation (2-3) shows that when T = + 1, the gain of the network becomes infinite and the circuit is unstable. A closer examination of T must be made, in order to determine what measures must be taken to avoid the point +1. T is defined as the loop transmission or loop gain of the feedback network and is of the form Z (Z1 1 12 Z 21 + ZG)(Z22 + ZL (2 -41 Assume for the moment that the circuit parameters are a known function of frequency. A Nyquist plot, in the T plane, of T ( j) as w varies from zero to infinity. can then be constructed, If this plot runs too near the critical point, + 1, then a change in any of the parameters may sweep T across the critical point. In order to take care of such a contingency, a stability margin will be imposed. This margin will be defined as (2-5) M>m= IY In this case M is in an upper bound, which is deterinined by design considerations, such as allowable tolerances in the system parameters. larger the value that is assigned to M, the closer T is to +1. The This reduces the limits of design tolerances and increases the probability of instability. 2.3 Gain Maximization With this stability margin imposed on- the network, gain maximization must be considered. The gain of the network is defined as 'the ratio of some output -5- quantity to some input quantity, such as power,. current, or voltage. The magnitude of the gain can be either gre3.ter than, less than, or equal to unity. The problem here is that of maximizing the power gain in a feedback circuit. The power gain of a network is defined as the ratio of the power delivered to the load PL (2-6) RL 122 to the available source power PS= k1 2 (2-7) G where EG is the peak value of a sine- curve. The maximum available gain is therefore Ga 4 R GRL (2-8) 2 Combining equations (2-3) and (2-5), the transfer impedance of the network can be put in the form 12 1 -- = -Z (2-9) Substituting this result into equation (2-8), the expression for the maximum available power gain becomes Ga = RGRL (2-10) 2m Examination of equation (2-10) shows the relationship between gain and stability margin. As the margin of stability is decreased (m becomes larger) the gain is increased. This is logical, since an oscillator, which is an unstable feedback circuit, has infinite gain. The problem then is to maximize this gain at every frequency by choosing an optimum value of ZG and ZL, while still maintaining a specified stability margin. -6- The problem can be partitioned into three distinct cases: and R22 greater than zero and M > m One R1 Two R11 and R22 greater than zero and M = m Three R and/or R22 less than zero and M = m In case one, with M > m, the power gain is maximized at values of ZG and ZL for which a margin of stability is safely maintained. For this condition, a conjugate image impedance match is used for both the generator and load impedances. Case three is complicated by possible instability of the amplifier and the fact that another parameter, N, must be chosen so that N > n= R1 1 (2-11) The choice of N is dictated by design tolerances. n = N. For maximum permissible gain The conjugate image impedance method is not applicable in case three, and the values of the source and load impedances are based on the value that is chosen for N. The method of solution in both these cases is a semi-graphical one, which will not be considered in this paper.1 A Junction transistor, connected as a grounded base amplifier and as a grounded emitter amplifier, is considered in this paper. The input and output impedances of these devices are positive for all frequencies, and over part of the frequency spectrum, M > m. Therefore, case two is of interest. Again the process of gain maximization is a semi-graphical one, but now the maximum permissible gain, taking stability into account, will be considered. The method of gain maximization will be investigat~ed in more detail, but in order to do this, certain quantities must be defined. 1. These are For further information on the design solution for cases one and three see Mason, S. J., "Power Gain in Feedback Amplifiers", RLE Report, July 1952. -7- H = Z12 Z2 1 f2 (2-12b) h = (1+r )2 R2 F (,+r)2 F = R (2-12a) H Rl R22 where (Z (z= r = + ZG )(Z 2 2 + ZL) Z 1 2 Z21 (2-13a) .. Rl (2-131) R22 H is a complex number which can be wi-itten as H H LH (2-14) With the quantities given in equations (2-12) to (2-14) and the relationship M (2-15) a vector diagram for the determination of ZG and ZL can be drawn. is shown in Figure 4. JM '14 paG-abT FIGURE 4. MA: VECTOR DIAGRAM SHOWING STABILITY MARGIN -8- This diagram In order that a network be considered in case two, E must lie outside the stability margin curve, M, at some frequency or frequencies. If, however, H lies between this M curve and the curve m = oo, conjugate image impedance matching (case one) can be used for maximizing the gain. This violates the given stability margin but the circuit will be stable as long as its parameters stay constant. When H lies outside the curve m = oo, then case two must always be considered, since a conjugate image impedance match will result in an unstable circuit. With a chosen value of M, the method of maximization becomes a graphical one. A value of r must be chosen so that the construction of Figure 4 can be made. From the intersection of f determined. with the m = co curve a value of k can be Using this value of k, and that of r which was previously determined, equations (2-16) give the optimum values of ZG and ZL at any frequency. RG = r R (2-16a) RL = r R2 2 (2-16b) XG = (l+r) k - X11 (2-16c) XL = (1+r) k R2 2 -X22 (2-16d) With these values for the generator and load impedances, the maximum permissible gain becomes Gp = R1 1 R2 2 Z12 (2-17) This process of maximization is a long and tedious one, requiring considerable geometrical construction. However, if considerable computation is expected, construction of curves of constant M and constant k in the 10 log jhj versus eH plane would be worthwhile. A second curve of 10 log (1+r) versus r would be 2 very helpful for obtaining values of r, since 10 log (l+r) can be obtained from a plot of 10 log JHf versus OH and the constant M curves. The inethod of obtaining r, and the use of these curves, will be explained in Chapter 3. It must be remembered that although the circuit is in case two for some frequency range, this is not necessarily true for all frequencies. Therefore, when H lies inside the given M curve, the maximization process reverts back to case one and a conjugate image impedance match. 2. 4 Unilateral Amplification Factor In the preceding discussion, the feedback in the network was only that which the network contained as an integral part. If the network input and output are coupled through an arbitrary lossless reciprocal network, N, as shown in Figure 5, any amount or kind of feedback can be introduced into the FIGURE 5. network. NETWORK WITH ARBITRARY FEEDBACK The problem with the network N is to adjust it so that the overall power gain is maximized, subject to the same stability margin that was previously imposed. With this type of coupling a parameter U, called the unilateral amplification factor, can be defined. It is of the form Z -Z 2 12 21 (Rl222 - R12 R21) U= (2-18) It can be shown1 that this parameter is invariant under permutations of the network terminals and under the addition of a lossless coupling network. Therefore (2-19) U(Z) = U(Z') If jUf > 1, a network N can be chosen which will make Rll and R2 2 greater than zero. Therefore, this network,.N, has operated on the determinant of the network, Z, in such a manner that a zero has appeared in the 1-2 position. The resulting network for such a situation is shown in Figure 6. rI, Z7 0- I----4 FIGURE 6. TRANSFORMED UNILATERAL NETWORK- For the network that is shown in Figure 6 the gain is jj2 (2-20) =U(Z') G = 4 R R22 Equation (2-20) shows that under the foregoing conditions, Z1 2 = 0, that U represents the gain of a two-terminal-pair network, regardless of the way the terminals are connected. 1. Mason, S.J., "Power Gain and Stability Margin in Feedback Amplifiers", RLE Group 35 Report, May 1953. -11- CHNPI'ER III TRANSISTOR GAIN 3.1 XIMIZATION Introduction In the preceding chapter, a method of determining the maximum permissible gain in a two-terminal-pair network was discussed. This method considered stability margin requirements, in order that instability would not be encountered due to parameter changes. In this chapter these methods will be applied to a junction transistor in both the grounded base and grounded emitter connections. The function U will then be examined. to see how its value compares with the gain computed for the unit without coupling. 3.2 Grounded Base Connection The first circuit to be considered is a grounded base junction transistor amplifier. The circuit is shown in Figure 7. The values chosen for the ltic cc r. = 20 ohms Cc = 50 Jgf rb = 350 ohms a r. = 1.5x10 6 ohms FIGURE 7. = .95 GROUNDED BASE JUNCTION TRANSISTOR parameters are the average values obtained from a number of transistors. The gain of the grounded base amplifier was maximized over the frequency range for which the network was in case two. arbitrarily chosen for M. For this work, a value of 2 was Other values could have been chosen since the -12- consideration here is to show how this method applies to a specific circuit and the results that can be obtainQd from it. Before going into the analysis the values of the circuit determinant of the network should be obtained for the grounded base transistor. These are in the form = r Z 11 b + r ((3-ia) c (3-1b) rb Z1 2 ~arc Z Z= rb + Z 2 2 = rb + 22 (3-ic) c (3-id) rc 1+ '7C where Tc.. (3-2a) rccc (3-2b) a = j The first step in the maximization procedure is to obtain a plot of 10 logjH versus 0 H. Combining equations (2-12a) and (3-1), and inserting the values of the circuit parameters, H becomes )(s,3.12x106-jl.82x,07)(Sal.3x i.8 (+3.12x106+.82xl0 (s+8.75x106)(s-8.75xl0(')(s+6.25xi06 ) (3-3) ) H From equation (3-3), by visualizing a pole-zero plot of H, in the s-plane, curves of 10 log in Figure 8. fHI and eH versus log w can be drawn. These curves are shown From these two plots a curve of 10 log jHj versus eH can be drawn, as shown in Figure 9. Before continuing further, the use of Figure 9 will be explained. From these curves the values of r and k, which are needed for the determination of -13- lo -1 _Z l BFRE-QUENCY IN RADIAN5 PER SECOND An __ .0*1 .00 4S + -4 - 4 f - - -" - ; 4+1 -1-' - -I ti H it 4-il , $ :117 -0411- -i T it --I--t- - 4 4 -] 71 :1tT t -1 4 - - - - 4 T 4 7-I'-' - - -121 --- - wpm: 4. - O-- -/ -P .. f. 4. V -I.-. tub 1LL~LL~i- 'K I V -- .1 TI.___________ F -~ ~- ~1 -V4 v-I-; 4-. ~~1. .10 4 4 * -. 4 - rrr~ 4 .1 H4 'Ia-- 77 ~wV~~r 7 - - I ~.-~-1..~~~~. - 'tirh *I tj~~ 14 I at.. ~e'~ FR 4 t~t ;i-i ~ 7 - 17.~ P7 I -- I ZG and ZL, can be obLained. Evaluation of' k is an easy proces included in Figure 9. since curves of constant k are At any frequency, the value of' k is found where the value of eH at this frequency crosses the given M curve. Determination of r requires the use of an auxiliary curve, Figure 10. Figure 9 is a plot in the 10 log Ihi versus the M curve corresponds to a value of 10 log eH plane, so that any point on Ihi. If logarithms are taken of equation (2-12a), it becomes 10 log IH - 10 log Ih = 10 log (l+r )2 (34) The evaluation of equation (3-4) can- be made by taking the difference between the 10 log fH This value curve and the curve for the given stability margin. can be used in conjunction with Figure 10 for obtaining r. With r now determined,relationship (2-17) can be evaluated at every frequency for which case two applies. frequency can then be plotted. A curve of this function, GP, versus log Substituting parameter values, equation (2-17) becomes GP = 16.92 r2 -5) (s+8.75xl05)(s-8.75xl05) (a +1.- 3X104) (a -1. 3x104) Because r is not a known function of s, a break-point method, using the s-plane poles and zeros of Gp, can not be used for plotting this' curve. The plot of Gp, for the frequency range in which the amplifier is in case two, is found in Figure 11. At high frequencies, Op falls off very rapidly. It is here that the s-plane factors in equation (3-5) become approximately equal and r becomes much less than one. Under these conditions, it is probably better to drop the M = 2 -16- - boundary conditions and revert to a conjugate-image-impedance match to improve Fiat 10i +r)so ~#1 1l4T T- -4-i It- -1. -T- fTt: 7 fl- + I I - ~ . " -1 Tl -T 4 + I ~L2 -I iiI ;i, 4 T I i I ij4ii -4 4 T. Ii t1t a1111 f ~TII ti 4: i -i- -1~t~I~11 P .-w J~ H 111 . I4I iii: -17- i;t - 44,+1-41 I I -: 14 I , Tl II++ TTTT T ~VTT i T 4- H -M., -4--4 rr 1 I ;-4-+ 4i~I1Th tT44~t14444P1 Thi 4tH+'~ +4411 i 1 IN 1 11 (P 01 -j M v -4 Akcl 10 dl %J Mb CD "86 w 0 j!~ .4.4. ~i7 .4.... .4.4, .4. 4... ~ FV . I .1. i. 4- 4-" '-F'- 7171 .1 - T 4 -.T .4... 1 ..... T --1~-.4 .. *I .4.... . 5.... ~fIIII7II .1 iii' .1~...... * .. I,. *1I ELI VIT I. , .. T ...- it -'-AL, ... ... . . t . Ii. .. -1 .11~ 7M I.... 5:: H- I7 . 4 .. 4... [,,J,4 K 51777i .1..... *1 44 *1-~ I:j~i~ ... !.77.47.T .4.... j~ i.i~i I .4 4 I. I * Ku *1 Ti :~ 711111 m~44 -46 V. ~1~~~~ iliji; F *~ 44 7.-..- -201~ ... ~ I.1 - - "l4 r. Perrm 4 XI .. .M .. '1.. 4~~L I .4...-.. 44 .4 -to 4 .4.... I I ja .~ -*~ 4~-44.. I --. 4. I -4---- 4 -4--.. 4 ... ..LL..LA...LL..... L - T7. -t 11417 ~ the gain. The reason for this is that the H curve now lies within the m = oo boundary and stability can be maintained by matching in this manner. The only remaining problem is to evaluate the generator and load impedances which will give the maximum gain. With values obtairied from Figure 9 for r and k, equations (2-16) can be evaluated. The values of ZG and ZL at two frequencies are listed in Table 1. ZGohms Frequency cpa ZL ohms Gain db 1.59x10 3 118+jl46 318xio 3-j349xlo 3 37 8.75x1o 5 362+j952 354-j2698 12 Table 1. Values of ZG and ZL for Grounded Base Amplifier In synthesizing the impedances found above, a series R - L connection can be used for the source impedance and a series R - C connection for the load impedance, as long as the values can be practically obtained. 3.2 Grounded Emitter A second connection to be considered is that of a grounded emitter junction transistor amplifier. The method of gain maximization is the same as that for the grounded base amplifier. shown in Figure 12. The circuit under consideration is The values of the circuit parameters are the same as those for the grounded base connection, as given in Figure 7. M = 2 was chosen for the stability margin. -19- Here again, a value of U 7- FIGURE 12. GROUNDED EMITTER JUNCTION TRANSISTOR A set of equations similar to equations (3-1) can be written for this connection. These are Z11 rb" (36a Z1 2 =r Z 21 =r r. = r (3-6b) + (3-6c) arc l+ TCs (1-a)rc (3-6d) 1+ Tcs where 'c and s have the same meaning as before. With the use of equation (2-12a), and the 4 preceding equations, H can be written as a polynomial of s-plane factors, from which Figure 13 was drawn. H is of the form (3-7) ) H H%056(s-6.25x106 Z .056 -~s l.)(s -1-3x10 x 2 4) (s+6.6xlo 4 )(s -6.6x10 4 The approximation comes about because H has two. poles and two zeros at high frequencies (approximately 12 megacycles) which have little effect. 1. See Appendix II, page 36. -20- Combining ~rtt 44 4.IS t Kc. L U11111ILL + 4-- f4 4 T7 -ij--t-V4--tm I- -44- -7-4 - --4-4 1-f -7 ~- I T ~.1 ... .. . : ' i .1- -44 -.4-, 4i 4-4 4 - - - .t- - - - t1 'IiI, _41 -1, -i-I- -'4- .144 + L.a 4- 4 -''-.4 4~; .4 +4 ---1-*-- ~r. 1L2 + ii4 :7 ttj "i " '7 i~W I -A , 11. -i- K 04 a -, 4' K; I .17717: .2 I0' ~0' V10*4auslc Y 101il I7I AL I.,* 44 $I. 44* - I 1-'- j - IT -t-t* 77 L- 7 T", it AI X1 I 22. 144 H- V '4 go ~ -- I. 'I U '1 -i--- I~J I 11. I0' 64sacot 4 the two curves of Figure 13, the curve of 10 log fHI versus plotted. described that be ;ure 14 is the same ascan The method of use of H in Figure 14 - for Figure 9. From Figure 14 and equation (2-17) a gain curve for the grounded emitter amplifier can be drawn. The equation for the gain, with numerical values substituted is Op = 296 r2 (s+7.75xlo7)(s-7.75xlo7)(s+6.6xlo4 )(S-6.6xlO4) (3-8) (s+6.25xlo6)(s-6.25xlOb)(s+1.3x,04)(8-1.3xl00) Again the variation of r with frequency is not a known function of s, and a point by point method of evaluation is needed. The curve that was plotted (Figure 15) was for the frequency range in which the amplifier was in case two. One of the big differences in this curve from that of the grounded base gain curve is the short frequency range it covers. However, it can be noted that only over a very small frequency range (9,86 - 19.1 k.c.) does the 10 log HI curve of Figure 14 lie outside the m = ac curve. Except for that one octave in the frequency spectrum, a conjugate image impedance match could be used, but stability would become a problem at frequencies bordering on this octave. This gain curve drops rapidly at high frequencies, resulting from the fact that r is much less than 1 and the ratio of the vector magnitudes is approaching unity. Therefore, at these frequencies, a conjugate image impedance match should be used in order to try to improve the gain. However, at low frequencies the gain is still rising and this would make one other. piece of information of interest, this being the zero frequency gain of the amplifier. This gain must be evaluated on a conjugate image impedance match basis and is 30.9 db. Although Figure 15 shows the gain to be increasing, it has fallen off somewhat -22- 7 77 IfK -23 - I 7 7 i~ IA. h C' Eh~F~I cl ir ;iI7TTTI7f7[17177--I---TI.i ~1 -~ -- b- 1 -"-4 r - I -.- - - -~ ~t2ttei' ~.-F~ ~ 5 A.41 -- 271V777iTf771 *1 m .4 2 I-7 -K--.7-------~--__ 4 I K0 t ; P"1I 1 Jill - I. .. .. 4- *1~ *1I0 -- W, J ~LLLL~LL 1. t t r, T = 'r'r'P 14 JViii_ Z-4i 1 ii> I 7 4 I*'. -iL.6 L -L A, .4i, 4ft: -7-7 77t- -4 .4~7.2j 4tif i tliftif I LL I.J 200 4O t Fit Go* em ioo T L 4;lt I )Ca :f 120l .: 1G170 IW4 C4 S. '.-. 44 -.. .. . . . + - - .-- + -'-4-.~~ ++ ++ +++-e -e4~~~~~~~~~~ + . - -T- - - ' - 1 1 1 1 ] 4-4- -ft. .4.4.4.. + - 01- I +~ ~ -4--4--+444-41 4 f -1 4-4 zt 4N-4I~T~ - t -T ~#+ I ~l~+**r4*44I~ - .lI[ I "4i" III 444 I .. 1:4IL11A - 4-, ~ III I11l1.~~.Li LL1..1A.iiii 2 II | A -+ Li: 19944; I - 144.1 ~ 'I:: LIII 1 I I 1 ; 11 4 .1 IL: i ~ ~~ Ii I~ Li: IILIILL4II:Lil4:LILIL:ILLLiiL:ILiLL:IIL:LLI En E.444..4..... mL I 111111 4-l--l--9-4-i-I-44+44+-4-l-4-4-4-444i-+-++4444t9l414l+l444$l444-A1t~$14$4ff't"flI----~ 4!; 4lt Ii -4 ii:; ~ 111111111 ~ ' I Ill 4 1 4 ' i i 444 I i I 44 1. f I I. I. I. I I t t TV . . . 1 9- 4~LI4494 4449444 II;HIII4,4444**-4+++-+-~-1-4$t$it4t4t44tt 41IIIt I tII 4. 117 Mi LIU 7777 r .4114.4 1114 T7 I I 4~f~rr1v 44 4 44 41411114 iI I i I I I 71TITIji 1 IT I Ii I I ill I Ii Ii I~~~~~~~~~~~~~~~~~~~~~~~ ,I;;4,i::14.1444iI':II4TILiII4Li:.L:i,::4-'LII4...I I II iii II 41 4 , L.LL:: i:IIL i4LIL :I:Li 4 Li lI!4!;IIlI!I11144i1.4LIIIi;IIIIII:;I.-II 4+4- 41 i11 I- 1- t't 114Il ! SIII 441 " .14 4441 i! : I liii LI 1777171 7. 1i I ILL .!iLI.L -4- I I I It 14444111 rtItitt4t4t"~t- - -4--j---4.--+.--I--+- 1 + 1~ -4- I LI L~.- Ii - 49- ~ ~ I ii 11i - L LII! II ~~+-L-L~ -ttf ;+*~fl4+l4i4+44-4+ 4 .1 I I I I '4.4.4 44114 44 141:4 4~1I 4,4 1 4.141114 149 I ,4--~-44~4-4.44-;-~+4lI-4l~44 ,~.iI~i4...d..4-.l--4.-J--4.-4--4..4-4.-9.4.4,'4--4---~---.--+-4-44--9--l-i'-l--4-. 4. 4 74 L~L 4 I 4 4 1..-,-4'J' LILILI 41~444t994I-4Il4l4:,t4.4-4-l-44.44.l->4.4.494-4-+44i 4 1119: I LI . IL -~tftii :: ILtt4l41 III -I LI L :-.-~ 1-tt 4 Ill i II 44491494 -4-4-4-4-,4-4 144-4---+--+--4-+-4--1--4--+t+-4-++*+-4ii II 'Ii. $ 99994' ii I 4 14 1..4-4-4-.4-4-4--.-~A-4.4 4-4--4--4---4---4---*-4--t---+-+t 414 Itui1iI4414411 - H 11.1 1414 4.~ 44 +t +-+ - 44 44949 944.---.~~4i4,..-444 Th 4-4-4-4 4- -1 1 1 -4-4-4-4- 4 -4+t -++ .4+~+ 14 ~ 6L 1 '- -i 1 j I jai 1 111 M I I I L I I I I 1 Mu I V1 1119 -11 I I I T Tj I I I I I - --T1 J Hill T T I LILL, I ti li 4 T I I F I I f I I I I I 1 1111 . I. 4 '-t * . I I '** + - t4-- -- -+.--,-, I + ++*+ 4-~jV;: -4- ~ -4- L -- - 1-l:::I2~54~+f -~ .4-4 -4-- -~~ .-- .- ++ -- -- -T ++-i4t--- + - d (4 4 L Il I I I.. I A ll I LM 111111 Ufl a. I T t I at zero frequency. acL Evaluation of equations (2-16) .'o frequencies gives the values needed These values at two frequencies are -for ZG and ZL to obtain these gains. listed in Table 2. Frequency ZG ZL' Gain cps ohms ohms db 4.46x103 24+j473 30.2x10 3 Table 2. 59+j386 1080+j21,200 34 18.5 832+j5430 Values of ZG and ZL for Grounded Emitter Amplifier Synthesis of these generator and load impedances calls for a series R - L connection if the element values can be practically obtained. 3.4 Determination of U In Chapter 2, a unilateral amplification factor, U, was defined. Since the value of U is the same regardless of the manner in which the transistor is connected, one curve suffices for both the grounded base and the grounded emitter connection. The numerical value of U is 1 4 (s+9.81x104 )(s-9.81x10 (3-9) ) U = .845xo14 from which Figure 16 is drawn. Since U is the gain in a unilateral amplifier, a 'comparison of the curve of Figure 16 with the gain curves for the grounded base and grounded emitter is in order. First look at the grounded base connection. -25- Comparison at the lower 1T I --Ur1: I 111F111;..;. N A K I *miM -4 'spIF I"" b 9~i It 74 - 7 114,1 a or ! n-A 4' 11e 4. K! H- .2 -1--- -z itt 4 I. - K I,, MmW- 414 ,ji it?- ~ 4 u, m 4.4 trT ITt **1~ Itit! 4-. -.4 tt -4. II '-. .. --.-. ..... -I - t- -- mi1 L!fk~E I4~r1j~44444.4qii2~22J TIPI H--4-. --..4. -4 r - 4 144++ -Iz :1 ti 1~'~ lit 1j h ji H~ I 141:17: -Ii K -1-- .4.. I.. 7111111 ii 4+4- .4-- - ,4 t 4.4. 111; F-"' n V 4--4-- Hj1i T -. 4- To - L- - -- 4,', 4 - -4 14Z ZlN IIIh 1~4 +44 ff -+ 4. - k4 Pt ITi 1, TV -.4' 4-i 4 I- 4414 Id t144-rt I r ' SOt~6S W S 21 ' I 11-t . ! FT S frequencies shows that the grounded base circuit has some uiilt in negative feedback, while at higher frequencies, where the gain is dropping, there is some built in positive feedback. This u.ans that at the low freque-cyend, a unilateral grounded base connection realizes the most gain and at high frequencies, a straight grounded base connection is the best. However, for the grounded emitter connection, the unilateral connection realizes more gain over the whole frequency spectrum, indicating that the grounded emitter transistor has some negative feedback built into it. The calculation of the gan of the grounded emitter circuit at zero frequency showed that even there, this connection could not realize more gain than a unilateral connection. -27- M M MW Concluding Remarks The method of gain maximization thaL is iudicated in this paper requires considerable work. No simple method h&s been arrived at so that the solution for any network can be reached qtickly. Values of the generator and load impedances must be evaluated, frequency by frequency. However, for an application of this method, a load and source impedance can be synthesized, which will maximize the power gain over the frequency range for wkdA tle network is to be used. In determining the function U, which indicates the amount of gain that can be realized by transforming a feedback network into a unilateral amplifier, no method has been stated for realizing the transforming network. Further work along this line is indicated, since, in some cases, the gain of a network can be increased by the use of such a coupling network. -28- I Appendix I Grounded Base In Chapter 3, several numerical formulas were given without the actual mathematics involved in obtaining them. here. The mathematics will be outlined First, the four circuit parameters, Z 1 1 , Z1 2, Z2 1 , and Z22 must be broken down into their real and imaginary parts. For Z very simple, since they are both real impedances at all Z1 1 rb + r= Z12 = r and Z12 this is frequencies. -370 (A -l) =350 (A-2) However, for the remaining two circuit parameters some work is required. Z21 is Z = rb 21b + a0rc 1+ /1a )(1+ frc -S (A -3) where a0 is the zero frequency value of a and a cutoff frequency in radians per second. Ta is the reciprocal of the This function can be written as a product of s-plane poles and zeros or as the sum of its real and imaginary parts. Both methods will be considered. First, Z 2 1 will be written as a quotient of its pole-zero factors. Obtaining a common denominator for equation (A-3) and collecting terms of the same power of a leaves Z21 in the form Z0 l+ c' rb'Ta + rb(Ta %)s + rb oc(I+TS( (+ Tas )(1 + 'Ies) (A J1 Dividing both numerator and denominator by rb Ta Tc, substituting the parameter values given in Chapter 3, and then factoring the numerator of Z2 1 gives the relationship -29- = 350 (a+3.12xlO6 +jl.82x,0 )(S+3.12x10 -jl.82x107 I ) Z ~~ ) (s+l.3x104 )(s+6.25xl06 Obtaining Z 2 1 as the sum of its real and imaginary parts requires that the second term in equation (A-3) be rationalized. Doing this, and then collecting even power terms of s as the real part of Z21 and odd power terms of s as the imaginary part of Z21 results in Z21= R2 1 + J X 2 1 aorc(1+ Ta cs2 ) A ac a Ta+ a T)s (1- I CS Al . -ls (. [ a )(1.' c I aa P 0 (A-6 0 ) rb + -. Substituting in equation (A-6) gives 2+8.3x10 1 0 (3-3 9xlol2)(s2-l.69xl1M) (A-7a) ) R21 = 350 + 1.19 x 1017 (s2 and 5 7.63 x 1023 (s2-39x1012 )(s2 -. (A -7b) 1.69x108 ) X '= - A similar procedure must be followed for Z2 2 , except only the real and imaginary parts need to be obtained. Z2 2 is r (A -8) 1++c - Z22 = rb + Rationalizing this gives rb + T R2 2 350 - 2.7x10l4 (s2-1.69xlob) X22 2x10 1 0 Z22 -r A -9) Substituting ( A -10a) 82-1.69xl0 A-o0b) 5 -30- Because of the form in which H was obtained, it will be necessary to determine R2 2 as a quotient of factors. Evaluating equation (A-10a) in this manner gives (s+8.75xl 5 )(s-8.75xl05) R22 = 350 (A-11) ) (s+1. 3xlO )(s -1.3xl04 Substituting the values for Z1 1 ) Zl2, Z2 1, and Z22 in equation (2-12a) results in equation (3-3) for H. A similar substitution in equation (2-17), with M = 2, gives equation (3-5) for the maximum permissible gain. Evaluation of the generator and load impedances was done at two frequencies. The first was w = 10 k = 0.3. radians per second. At this frequency r = 0.32 and From equation (2-16) optimum values of ZG and ZL at this frequency are ZG = r R + j (l+r) k R 11 =.32 x 370 + J (1.32)(.3)370 = 118 + j 146 (A -12) and ZL = r R 2 2 + j [(1+r) k R22 - X22 = .32x995x103 + j [1.32x3x995xlo 3 - 7.44x105] =318 x 103 - J 349 x 10 3 (A-13) The second frequency at which the generator and load impedances are calculated is to = 5.5x106 radians per second. Optimum values of ZG and ZL at this frequency are ZG = r Rl 1 + j (1+r) k R11 =.98x370 + J (1.98) 1.3 370 362 + j 952 and -31- (A-14 ) k = 1.3. At this frequency r = 0.98 and ZL= r + j [(L+r) k R 2 2 22 =.96x3 66 + = II ' 22] - 3.64x10 3 j l1.98xl.3x36C (A-15) 354 - j 2698 Grounded Emitter An analysis similar to that for the grounded base connection will be done for the grounded emitter connection. be evaluated. As in the grounded base case Z and Z12 are re&l impedances frequencies and are Zi = rb + rc = 370 (A-16 Z1 2 = r (A-17) ) at all First, the four circuit parameters will = 20 Again the remaining two parameters are complex impedances. Z = r. 21. era (1+T (l+ First look at (A-18) aC ) This function will be evaluated both as a quotient of .its s-plane factors and as the sum of its real and imaginary parts. Consider first the quotient of factors. Obtaining a common denominator and collecting terms of the same power of s, Z2 1 becomes Zr a Ic 2 + r6( Ta+ V)s + re-aorc (1+ rcs)(l+ Ta) (A-19) Substitution of the circuit parameter values in equation (A-19) and obtaining the roots of the polynomials results in Z 7 )(s-7.4x107 ) = 20(s+8x10 2~ (s-1.3xl04)(s+6.75xl06) (A-20) Next the real and imaginary parts of Z21 must be obtained. Z21 = R2 1 + j X 2 1 a r (1+ 7 = r22 s2 ) 2 (1- or2)(1:2s2) -32- aorc( a+ 1)s + (II 0 (1_ 7 2 a(A-21) _ 722 Numerical substitution in equation (A-21) gives 2 21 = 20 - 1.19x1017 (6 -39x12 10 (A-22a) 6 9x1) )( -1., and x s (S2 -39xl01-2)(82-j.69x18-)-- 7.63x102 3 = 21 (A -22b) It will be obtained . The only remaining parameter to be evaluated is Z 22 as the sum of its real and imaginary parts. R22 + j X 2 2 Z22 ao + \c + r (A-23) 1 + -cs er 2- 84 - [r 6 (eT2+'?')+ Tar s 2 +rg -r (1-a aIar) (A-24a) - = ) This equation can be broken down into its real and imaginary components. (1- TC 0 X2 2 Tcs3+rcap( Tr ( Q1-(l -; a2(-2a+ 2 2 . (A-24b) Substituting in equations (A-24) and factoring gives 20 (8+7.75x107)(s-7.75xl07)(s+6.6x0o4)(s-6.6x (s.-1 3 x,04)(s+1. 3 x04)(s-6.25xo6.)(s+6.25xlO 6 10) (A-25a) ) = s(s+j6.l7xl06)(s -j6.17xl0 6 ) X22 = 2x10 1 0 (s-1.3xl04)(s+1. 3 x,04)(s-6.25xlO6)(s+6.2 5 xlO6) Introduction of relations (A-16 ), (A-17), (A -25b) (A-20), and (A-24a) into equation (2-12a) results in equation (A-2 6 ) for H. ) H = .056 (s+8xl07)(s-7.4xl07)(s-6.25xlo6)(s-l.3xlo 4 (s+7.75x107 )(s -7.75xl07)(s+6.6x1o4)(s-6.6xi0) (A-26) The factors which occur at the large values of s (about a a 7.5x107 ) have little or no effect on the phase and gain of H at the frequencies of interest. -33- Therefore, for computation purposes, these factors may be nrglected. This is the approximation that is referred to in connection with equation (3-9). A similar substitution of equation (A-24 a) in equation (2-17), with M = 2, results in equation (3-5). The generator and load impedances were evaluated at two frequencies. The first was o = 2.8x104 radians per second, at which frequency r = .065 and k = 1.2. Equations (2-16) then gives for ZG and ZL ZG = r R1 1 + j (l+r) k R 1 1 = .065x370 + j (1.065)(1.2) 370 (A-27) = 24 + j 47 3 and r R2 2 + ZL (l+r) k R2 2 - X22 = .06 5x16 ,6 00 + j 1.o65xl.2x16,600-201 = 1080 + j 21,200 (A-28) The second frequency is o = l.9x10 5 radians per second for which r = 0.16 and k = 0.9. This leads to the following source and load impedances ZG = r RH1 + j (l+r) k RB1 = .16 x370 + J 1-16x9x370 = 59 + j 386 (A-29) and ZL = r R 2 2 + j [(l+r) k R 2 2 - X2 21 = .16x5200 + j [1.16x.9x5200 = 832 + j 5430 - 0.5] (A-30) -34- III U The function U, which is the same for the grounded base and the grounded emitter connections, was determined by using the circuit values for the grounded emitter connection. For the function U, as defined in relationship (2-18) Z 21 - Z - = r 12 e - a 1+ 'sc - =- arc (A-31) 1+ 7c5 and R R 11 22 2R 12 21 -Frao Ta Lb 0r a r + (r+r) ls 2 al + (rb+rc)rG +rbrc(1-ao)- (A-32) Substitution of the transistor parameters in these two relations, followed by subsequent division of equation (A-31) by equation (A-32) times 4, results in equation (3-9). -35- -1 BIBLIOGRAPHY 1. Llewelyn, F. B., "Some Fundamental Properties of Transmission Systems", IRE Proc. 40, March 1952, pp. 271-283. 2. Mason, S. J., "On the Logic of Feedback", MIT E.E. ScD. Thesis, 1951. 3. Mason, S. J., "Power Gain in Feedback Amplifiers", RLE report, July 1952. 4. Mason, S. J., "Power Gain and Stability Margin in Feedback Amplifiers", RLE group 35 report, May 1953. 5. Roberts, Shepard, "Conjugate Image Impedance", IRE Proc. 34, April 1946, pp. 198P-203P. -36-