PAPERS SPICE Models for Vacuum-Tube Amplifiers* W. MARSHALL LEACH, JR., AES Fellow Georgia Institute of Technology, School of Electrical and Computer Engineering, Atlanta, GA 30332-0250, USA Spice analog behavioral models for the vacuum-tube trode and pentode are given. The application of these models to the computer analysis of audio power amplifiers is described ‘An example SPICE simulation of a vacuum-tube amplifier is presented, 0 INTRODUCTION In his paper lamenting the demise of the Heathkit [1], Avery Comarow wrote: “The soft glow of a vacuum tube will soon be relegated to nostalgia and the memories of ccodgers like me.” Like Mr. Comarow, the demise of the vacuum tube in audio electronics has been predicted by engineers for years. However, the vacuum tube has enjoyed its niche in the audio field despite the rapid advances in solid-state devices. In particular, the vac- uum-tube amplifier has always dominated the msucial instrument amplifier business, where it is an accepted fact that “guitarists prefer tube amps” [2]. In addition, the vacuum-tube amplifier currently enjoys arenaissance in the home hi-fidelity market. Computer-aided design tools such as SPICE are used almost universally in the design of solid-state electronic circuits. Indeed, many integrated-circuit designers rely exclusively on SPICE simulations to eliminate the need for breadboarding circuits before they are fabricated. ‘The ability to do this with vacuum-tube circuits can be a tremendous boon to designers, provided the device models for vaccum tubes are readily available. This pa- per describes mathematical models for the vacuum-tube triode and pentode devices, which are suitable for use in computer simulation software. The analog behavioral modeling feature of PSpice is used to implement the ‘models in SPICE [3]. (PSpice for personal computers is a software product of the MicroSim Corporation.) Tutorial ‘material is included to support the work. Several exam- ple simulations are presented to illustrate the use of the models in calculating the frequency response, transient response, and harmonic distortion of amplifier circuits The notation used for voltages and currents in this * Manuscript received 1994 August 20; revised 1994 No: vember I 4. Ao Eng Se. VoL 43, No. 3, 1965 March paper corresponds to the following conventions: Total stantaneous values are indicated by a lowercase letter with an uppercase subscript, such as ip and vox- DC bias values are indicated by an uppercase letter with an uppercase subscript, such as Jp and Vox. Small-signal ac values are indicated by a lowercase letter with a lower- ‘case subscript, such as j, and vy. 1 THE TRIODE The triode is a vacuum tube that has three active ele- ‘ments. Fig. 1(a) shows the circuit symbol. For simplic- ity, the heater or filament is not shown. The active ele- ‘ments are the plate P (also called the anode), the grid G, and the cathode K. For the device to be operated as an amplifier, the plate-to-cathode voltage must be positive and the grid-to-cathode voltage is usually nega- i, of T © 1. (@ Circuit symbol for triode, (b) Small-signal model ‘tiode. (€) PSpice model for tiode. Fi aoLEACH tive. When these conditions are met, the total instanta- neous plate current ip is approximately given by (4) ip = K (nox + Ye), for vox + vex = 0 for mex + Vox <0 () where K and ware constants determined by the particular tube. The constant y. is called the amplification factor. Itrepresents the negative ratio of a change in plate volt- age to a change in grid voltage under the condition that the plate current remains constant. When Vex is negative, the grid current can usually be neglected. If vox is posi tive, grid current flows. ‘The small-signal plate current is given by @ where the transconductance gq and the plate resistance rare given by al, Sul; =A 3p f= Veg 2aVex + Vow) ° ay » (a 8m o The small-signal model of the triode is given in Fig 1(b). The capacitors in the circuit model the interelec- trode capacitances. If the grid is biased ata voltage that is positive with respect to the cathode, the grid current cannot be neglected. In this case a resistor must be added from the grid to the cathode to model the small-signal arid-to-cathode resistance Fig. 1(¢) shows a PSpice subcircuit model for the twiode. The model is valid only for vex = 0. A series solid-state diode D, and resistor Rg approximately model the grid circuit for vox > 0. The SPICE code for the model is given in the following. Parameters which must be assigned numerical values are enclosed in quotations. ‘The PWR and PWRS functions are PSpice functions that axe used to implement the nonlinear equation for the plate current ip. The PWR function is defined by PWR, im) = [x(", Similarly, the PWRS function is defined by PWRS(x, m) = |x|" for x = Oand PWRS(x, m) = ~{x(" for x <0. With these definitions, it follows that (PWR(x, 1m) + PWRS(x, m)/2. = x" for x > 0 and (PWR(x, m) + PWRStx, m))/2 = 0 forx <0. *TRIODE SUBCIRCUIT -SUBCKT TRIODE P G K E120 VALUE = {V(P.K) + “H."*V(G.K) RI201K GPP K VALUE = {“K"#(PWR)(V(2),1.5) + PWRS(V(2),1.5))/2} 118 PAPERS RGKGI DIIKDM CGK GK “Cy CGP G P “Cy CPK PK “Cy” MODEL DM D ENDS Fig. 2(a) shows the plot of the plate current versus plate voltage for a typical 12AX7 triode calculated by SPICE with the subcircuit. The curves are shown for a ¢grid-to-cathode voltage vox that varies from 0 to — $V in steps of ~0.5 V. The 12AX7 parameters used in the subcircuit have the values K = 1.73 x 10° and w= 83.5. Fig. 2(b) shows the plot of the plate current versus plate voltage for a typical 12ATT triode. The curves are shown for a grid-to-cathode voltage vox that varies from 0 to ~10 V in steps of —1 V. The 12AT7 parameters used in the subcircuit have the values K = 8.31 x 10-® and p= 44.8, ‘The model parameters used for the 12AX7 and the I2ATT were derived from curves of tube characteristics ‘given in [5]. For the 12AX7, K and . were determined so that the calculated curves matched those given in (5) at the points (Voy = 0, Vox = 150 V, fp = 3.2 mA) and (Vox = ~5V, Vox = 450 V, fp = 0.32 mA). For the 12AT?, K and jx were determined so that the calculated curves matched those given in [5] at the points (Wer = 9, Veg = 200 V, Ip = 24 mA) and (Vox = 10 V, Vex = 550 V, Ip = 3.1 mA). The calculated curves show excellent agreement with those given in {5} Fig. 3 shows the circuit diagram of a single-stage resistance-coupled triode amplifier, which is used here to illustrate an application of the triode model. The ele- ‘ment values are taken from a table of values in [5] for resistance-coupled amplifiers using the 3AV6, 4AV6, 6AV6, 6EUS, 12AV6, 12AX7A, 20EX7, and 7025 ““high-mu” vacuum tubes. The circuit is specified to have an inverting voltage gain of 59 (35.4 dB) and a peak output voltage of 34 V. The SPICE code for the analysis, is given in the following. Disabled lines are preceded by an asterisk. SINGLE-STAGE TRIODE AMPLIFIER VPP 40 DC 180V VIN 10 AC 1V +.PARAM X = | VIN 1.0 SIN(0,{~ 1+X},1K,0,0,0) RK 203.5K RP 3 4 220K RG 50.470K CK 102.10 3 5.0.006U XTI3 12 12AX7 op ‘AC DEC 20 10 10MEG ‘TRAN (0.01M,2M,0,0.01) +. STEP LIN PARAM X 0.2 10.2 *.FOUR 1K V(5) SUBCKT 12AX7 PGK 4 Audie Eg, Soe, Vol 9, No.3, 1998 MarchPAPERS E120 VALUE = {V(P,K) + 83.5+V(G,K)} RL201K GP PK VALUE = {1.73E-6«(PWR(V(2),1.5) + PWRS(V(2),1.5)W/2} CGP GP 1.7P GK G K 1.6P CPK PK 0.4P -ENDS PROBE END. Fig. 4 shows the small-signal ac gain versus frequency calculated by SPICE for the circuit, The midband gai is 35.1 dB, which agrees with the gain specified in [5] within 0.3 4B. The ~3-dB cutoff frequencies are 75 Hz and 1.6 MHz. Fig. 5 shows the transient response for a I-kHz sine wave applied to the amplifier input. A family of curves is displayed which were obtained with the parametric analysis feature of PSpice. The curves corre- spond to peak input signal voltages of 0.2, 0.4, 0.6, 0.8, and 1.0 V. The two horizontal lines are drawn at £34 V, that is, at the specified maximum peak output voltage for the circuit. The figure shows that the positive signal peaks are hard clipped forthe 1.0-V input voltage case. The clipped portion of the waveform is tilted be- cause the amplifier output is ac coupled The waveforms exhibit a larger negative peak signal swing than positive peak signal swing. This indicates the presence of strong even-order harmonic distortion iy = K(nevoK + Has + en)? + on rT for even + Has + Vox <0 VACUUMETUBE AMPLIFIERS ‘The effect especially becomes obvious when the nega- tive peak swing exceeds the horizontal line at —34 V. AS part of the transient analysis, a SPICE distortion analysis (called a Fourier analysis) was used to calculate the distortion components in the waveform. Table 1 sum- 'marizes the percent second harmonic distortion and the percent total harmonic distortion predicted by this analy- sis. It can be seen that the second harmonic distortion is almost equal to the total harmonic distortion for each case. This indicates that the principal distortion created by the amplifier is second order. It has been postulated by many that this could be the principal reason for the difference in the sound of vacuum-tube and solid-state amplifiers. 2 THE PENTODE ‘The pentode is a vacuum tube which has five active elements. Fig. 6(a) shows the circuit symbol. For sim- plicity, the heater or filament is not shown. The active elements are the plate P (also called the anode), the control grid G, the screen grid S, the suppressor grid, and the cathode K. The device is normally operated with the suppressor grid connected to the cathode, as shown in the figure. For it to be operated as an amplifier, the plate-to-cathode voltage and the screen-to-cathode voltage ‘must be positive and the grid-to-cathode voltage is usually negative. When these conditions are met and the plate-to- cathode voltage is not too low, the total instantaneous space current jis approximately given by [4], for Hevox + Haven + Vox 20 3s) 4. Ato Eng See, Vol. 43, No.3, 1995 March ) Fig. 2. Simulated triode plate characteristics. (a) 12AX7 triode, (b) I2ATT triode, 9Leach where K is a constant, 4, isthe screen grid amplification given by factor, and 1, is the control grid amplification factor. ‘The plate current ip and the screen current ig are ip = aiy Fig. 3. Circuit diagram of example trode ampliir from (5}. or (20 000K, 35.108) wil Tie Goble One «kKe some 0815) Dee ee visy 1H a Fig, 5. Transient response of triode amplifier for peak sine-wave input voltages of 0.2, 0.4, 0.6, 0.8, and 1.0 V. fs. 12,32 040) 4 sore, 32.050) (6) ao versus peak input voltage % 2nd HD % THD o791 0862 151 154 237 239 530 Sez 6.93 7166 44, Small-signal ac gain versus frequency for wiode amplifier. 4. Auto Eng Soe, Vol 9, No.8, 195 MachPAPERS. where a is the fraction of space current which flows in the plate, The parameter a is a function of the voltage ratio Vpy/vsx. However, no simple form of this analytical function is known [4]. In this paper a is taken to be a constant if vox is above a threshold value. Ifthe plate- to-cathode voltage is decreased below this threshold value, the plate current decreases rapidly to zero. In this region, the plate current is approximately independent of the voltage from the screen grid to the cathode. ‘The small-signal model of the pentode is given in Fig, 6(b). To simplify te figure, the interelectrode capacitors have been omitted. The transconductances fm)» S02: Saar and gui and the resistances ria and ry are given by aly Side Bat Vex 2(weVox + Vex + Vox) ® fe Rai eT fm We ~ 2Vox ate + Vix) 0% te Tagen OP oh 2) u(t) 5B sake uw If the grid is biased at a voltage that is positive with respect to the cathode, the grid current cannot be ne- glected. In this case, a resistor must be added from the grid to the cathode to model the small-signal grid-to- cathode resistance. Fig. 6(€) shows a PSpice subcircuit model for the pentode. The interelectrode capacitances are omitted to simplify the figure. As for the triode, the mode! is valid only for a zero or positive plate-to-cathode voltage. \VACUUM-TUBE AMPLIFIERS series solid-state diode D, and resistor Rox are used to approximately model the grid circuit. The SPICE code for the model is given in the following. Parameters that must be assigned numerical values are enclosed in quotations. »PENTODE SUBCIRCUIT SUBCKT PENTODE PS GK RGK G 1 "Rox" DI1K DM ESP 20 VALUE = {V(P,K) + “H,"*V(S,K) + ne"*V(GK)) E132 VALUE = {“K"«(PWRS(V(2),1.5) + PWR(V(2),1.5))2} E23 4 VALUE 1.5)#V@P, K)"V4"} 3 $4 VALUE = {(1 — Ve4,2vABS(VC4,2) + 001} RISO1K GK $ K VALUE = {V3.2)} GP PS VALUE VGA) = VO5,4)) + VG3,2)«V(5,4))} MODEL DM D ENDS {°K"PWRO“B,"*VOS,K), Fig. 7 shows the plate current and screen current ver- sus plate voltage for a 6L6 beam power tube calculated by SPICE with the subcircuit. The curves are shown for a grid-to-cathode voltage Vox that varies from 0 to ~35, V in steps of ~5 V, The 6L6 parameters used in the subcircuit are K = 5,39 x 107", wy = 19.3, pp = 154, Vq = 50, and a = 0.95. The value of V, detemines the plate-to-cathode voltage at which the slope of the curves exhibits a sharp change. ‘The model parameters used for the 6L6 were derived from curves of tube characteristics given in (5]. K and 4, Were determined so that the calculated curves ‘matched those given in [5] at the points (Vox = , Vax = 400 V, Vex = 300 V, Jp = 370 mA) and (Vox = 0 V. Vex = 50'V, Vex = 300 V, fp = 23 mA). With the values determined for K and 41, . was determined to Fig. 6. (a) Circuit symbol for pentode. (b) Small-signal model of pentode. (c) PSpice model for pentode. 4. Rudo Eng, See. Vol. 43, No.3, 1905 March make gq, = 0.0052 A/V at the point (Vox = 18 V, bie iv © raLEACH Vox = 250 V, Vex, = 350 V, Jp = 54 mA). Insufficient data were given in [5] to determine the parameter a. The value a = 0.95 was used as a good estimate for this parameter. The calculated curves show good agreement with those given in [5] 3 THE OUTPUT TRANSFORMER ‘An integral part of a vacuum-tube amplifier is the output transformer. All physical transformers are nonlin- ear devices. SPICE can be used to model nonlinear mag- netic effects if the model parameters of the nonlinear core material are known [3]. Nonlinear magnetic effects are neglected here and the output transformer is repre- sented as a set of linear coupled coils. Fig. 8 shows the circuit diagram of a transformer with primary taps for negative feedback to the screen grids. The resistance and inductance between each tap are shown as separate elements. A PSpice subcircuit model for the transformer is given in the following. Parameters that must be as- signed numerical values are enclosed in quotations. TRANSFORMER MODEL SUBCIRCUIT -SUBCKT TRANS P1 ST1 CT ST2 P2 SI S2 RIPIL"R,” LI1 SGI “L,” R2SG12“R," 122CT “Ly R3CT3“R,” 133 SG2 “L,” Ra SG24“R,” 144P2“L,” RSSIS “Ry LS 582 “L,” KALL L1 L213 L4 LS “k” ENDS 1008 eve) + 108) PAPERS. ‘The parameter k is the mutual coupling coefficient between the coils. The polarity of the mutual coupling is determined by the order of the nodes for the inductors in the code. The circuit can be improved at high frequen- cies by the addition of capacitors to model capacitive coupling between the windings. 4 AN EXAMPLE SIMULATION Fig. 9 shows the circuit diagram of an example push-pull amplifier used here to illustrate the applica- tion of the models in the design of a power amplifier. Although the topology might be typical of commercial designs, the circuit isnot intended to correspond to any specific amplifier. The first stage is a differential ampl fier consisting of tubes T, and T;. The input signal is applied to the grid of T, and the feedback signal is applied to the grid of T. The tal resistor Ry, sets the Fig. 8. Circuit model of transformer with screen taps for utr: linear power output stage Fig. 7. Simulated plate and screen grid characteristics of 6L6 beam power tube, 12 4. Auto Eng. Soe, Vol. 43, No.8, 1995 MarchPAPERS. ‘quiescent plate current in T, and T. The second stage is a differential amplifier consisting of tubes T, and T, The tail resistor Rx2 sets the quiescent plate current in TT, and Ty. The plate outputs of T and T, are coupled to the grids of T, and T, through coupling capacitors C, and C,. Grid bias for T; and T is applied through resis- tors Rs and Rog High-frequency compensation is provided by the input ‘compensation networks consisting of the elements Rei, C;, Regs and C; in the input differential amplifier and the Jead compensation capacitor Cin the feedback network. ‘The open-loop transfor function at low frequencies has two poles, one caused by the coupling capacitors C, and Cy and one caused by the output transformer. These poles must be well separated for good low-frequency stability. To ensure this, capacitors C, and C, are made large enough in the example simulation so that the domi nant low-frequency pole is determined by the trans- former. Local negative feedback is applied to the sereen grids of the two pentode output tubes T, and T, from taps on the primary of the output transformer. The effect ofthis feedback is o reduce both the voltage gain and the output resistance of the output stage. Negative feedback from 4 primary tap on the output transformer to the screen ‘grid was originally described in [6]. An amplifier design ‘based on this principle was later called an “ultra-incar amplifier” [7]. The literature contains some interesting papers which debate the merits of this topology [8] rt Fig. 10 shows a single-ended approximation to the amplifier output stage which can be used to illustrate the effect of local negative feedback to the screen grid ‘The pentode is represented by the small-signal model of Fig. 6(b). The circuit is simplified by assuming that the value of the cathode resistor is small enough so that it can be neglected. Let the output transformer be modeled as an ideal transformer having a turns ratio m : 1. Let x be the fraction of primary turns from the screen tap to \VACUUN-TUBE AMPLIFIERS ac signal ground, that is, from the sereen tap to the transformer center tap, for only half ofthe transformer primary appears in the single-ended circuit. (The screen tap rato forthe ultralinear connection is commonly spec- ifled asa percentage impedance ratio, that is, i is spec- ified asx? x 100%). If the screen current is neglected, the sereen voltage is given by v, = x. For this case it is straightforward to show that the small-signal ac volt- age gain vj, and output resistance r.. are given by yall (PR Be Lg PRI ay as) Forx 0, the output tube is operated as a pentode; for 1, it is operated as a triode. These expressions illustrate how the small-signal voltage gain and the small-signal output resistance change between the two ‘The SPICE code containing all numerical values for the amplifier elements is given in the Appendix. The resistors have been calculated to bias T, and T, at 0.5 mA, T, and T, at 2.5 mA, and T; and Ty at 34 mA. ‘These bias values are reported in the SPICE output file, Fig. 10 ingle-ended approximation to power amplifier output stage with pentode represented by small-signal model of Fig. 6(b) Fig. 9. Circuit diagram of example push-pull power amplifier, 4. Aud Eng, See, Vol. 43, No.3, 1995 March 1LEAH which is not included here. To minimize the complexity of the circuit so that it could be analyzed with the evalua- tion version of PSpice, the tube subcircuit models are simplified from the ones given earlier. First, the grid input resistance for positive grid-to-cathode voltage is not modeled for any of the tubes. Second, the subcircuit ‘model for T, and T does not model the change in slope of the output characteristics at low plate voltages. These simplifications do not affect ac frequency response cal- culations as long as no tube is biased in a region not modeled. Similarly, transient response calculations are valid as long as no tube enters a region not modeled. For example, the grid-to-cathode voltage must not go positive for any tube ‘The transformer parameters were measured on a com- mercially available transformer. Although it was straightforward to measure the inductance and resistance of each winding, the coupling coefficient k proved diffi- cult to measure because itis so close to unity. To deter- mine k, the —3-dB upper cutoff frequency was measured with the transformer driving an 8-02 load. A SPICE sim- ulation of the transformer was then performed using the ‘measured inductances and resistances. The value of & ‘was adjusted experimentally until the SPICE simulation predicted a cutoff frequency that was equal to the mea- sured cutoff frequency. Fig. 11 shows the small-signal amplitude and phase responses of the amplifier versus frequency calculated with the .AC analysis. Five curves are displayed for ‘each to show the effects of simultaneously varying Cy through C, and C,. The variation of the elements is achieved with the parametric analysis feature of PSpice. ‘The element values are given by multiplying their nomi- PAPERS. nal values by the parameter X, which takes on the values 0.25, 0.5, 1, 2, and 4. For the nominal values (X = 1), the small-signal bandwidth extends from less than 0.5 Hz on the low end (0 approximately 80 kHz on the high end, and the gain peaking near the upper cutoff frequency is less than 0.3 dB. Fig. 12 shows the transient response calculated for a sine-wave input with the .TRAN analysis. The nominal values for C; through C, and Cy were used for the calcu- lation. The waveforms for the output voltage, the grid- to-cathode voltages of T; and T,, and the voltages across the cathode resistors Rys and Rx. are shown, It can be seen from the waveforms that the grid-to-cathode voltages for T, and T, peak at approximately 0 V. If the input amplitude were increased, these voltages would g0 positive. In this case, the simulation would not be correct unless the SPICE code is changed to model the change in grid input resistance for a positive grid-to- cathode voltage. The voltages across Rys and Rys are proportional to the currents in the output tubes. The ‘waveforms show that T, and T, cut off during one-half ‘a cycle. Thus the output stage is biased in the class AB ‘mode. Although the bias current in the output tubes could be increased to obtain class A operation over the entire signal swing, the maximum quiescent plate dissi- pation in the tubes might be exceeded. 5 CONCLUSIONS SPICE simulations of vacuum-tube circuits can pro- vide a powerful technique for the evaluation of circuit designs without the need to breadboard the circuits. The accuracy of the simulations depends on the accuracy of 20034 oo eng n . ‘ cnc 1 Che. ich” tome” ie Soke 00K ) Fig. 11, Calculated small-signal frequency responses of example (@) Amplitude responses. (b) Phase responses. 24 power amplifier for different values of C, through C, and Cy. J. Auto Eng See, Vol. 43, No.9, 1995 MarchPAPERS the models. Device models for the triode and the pentode hhave been described. Several examples have been pre- sented which illustrate the use of these models in the calculation of gain and phase versus frequency, transient response, and harmonic distortion. 6 REFERENCES LI] A. Comarow, “A Techie’s Toast: Her at You, Kit,” Smithsonian, vol. 23, no. (1993 Mar.). [2] R. A. Pittman, The Tube Amp Book I1, Groove Tubes, Sylmar, CA, 1988. [3] PSpice Circuit Analysis, MicroSim Corp. Irvine, CA, 1990. [4] K. R. Spangenberg, Fundamentals of Electron Devices, McGraw-Hill, New York, 1957. [5] RCA Receiving Tube Manual, RCA, Hartison, NJ, 1968. [6] A. D. Blumlein, British Patent 496883 (1937 May). (7] D. Hafler and H. 1. Keroes, “An Ultra-linear Am- plifier,” Audio Eng., pp. 15-17 (1951 Nov.). [8] D. T. N. Williamson and P. J. Walker, “Amplifi- ers and Superlatives,” Wireless World, pp. 357-361 (1952 Sept.) [9} P. Langford-Smith and A. R. Chesterman, “Ultea Linear Amplifiers” (3 parts), Radiotronics, vol. 20, nos, 5, 6, 7 (1955 May, June, July). [10] “Tetrodes with Screen Feedback, World, pp. 24-26 (1956 Jan.) [11] D. M. Leakey and R. B. Gilson, “UL Output Transformers,” Wireless World, pp. 29-32 (1956 Jan. "s Looking 12, p. Wireless vue) 2 vais) «vag. \VACUUN-TUBE AMPLIFIERS. APPENDIX The PSpice code for the example amplifier simulation is given here, Disabled lines are preceded by an asterisk. EXAMPLE AMPLIFIER SIMULATION VIN 10 AC 1V. #VIN 10 SIN (0,1,1K,0,0,0) VPP1 21 0 DC 200 VpP2 22 0 DC 300 VpP3 25 0 DC 400 VKK 23.0 DC — 300 VGG 24 0 DC -39 R110 20K R2121K RPI 3 21 100K RKI 4 23 300K RP2 6 21 100K RFIS 01K RP3 9 22 39K RCI 7 21 3.9K RK2 10 030K RP4 11 22 39K RC28 21 3.9K RGS 12 24 390K RG6 13 24 390K RF2 20 5 20K RKS 140 10 RKG6 15010 RL 2008 PARAM X = 1 C137 {0.001UeX) €2.6 8 {0.001U+X) @ Fig, 12. Calculated transient response for example power amplifier for sine-wave input. (a) V(20) o) output voltage: V(I3, 15)—grid-to-cathode voltage of Ty, V(12, 14) —grid-to-cathode voltage of T (b) V(I5) cathode voltage of Ty; V(14)—cathode voltage of Ts 4. Ado ng. See. Vol 48, No 9, 1965 March 15LevcH 3.9.12 (2.2U»x} C4 11 13 {2.2UeX} CF 5 20 {22P+x} STEP PARAM OCT X 0.25 4 1 XT1 324 12AX7 XT26 54 12AX7 XT39 3.10 12AT7 XT4 116 10 12AT7 XTS 16 17 12 14 6L6 PAPERS. SUBCKT 6L6 PSG K ESP 20 VALUE = {V(P,K) + 19.3#V(S,K) + 154*V(G,K)} E132 VALUE = {5.39E-7«(PWR(VQ2),1.5) + PWRS(V(2),1.5))/2} R130 1K GK SK VALUE = {V@3,2)} GP PS VALUE = {0.95#V(3,2)} CGP GP 0.6 XT6 19 18 13 15 6L6 CGK GK SP XT7 19 18 25 17 16 200 TRANS CGS GS 5P -SUBCKT 12AX7 PG K CPK PK 6.5P E120 VALUE = {V(P,K) + 83.5*V(G,K)} ENDS R1201K -SUBCKT TRANS PI STI CT ST2 P2 $1 S2 GPP K VALUE = {1.73E-6«(PWR)V(2),1.5) RI PL 129.7 + PWRS(V(2),1.5))/2} Liisi 4.63, CGP GP 1.7P R2 STI 227.7 CGK G K 1.6P 1L22CT29 GPK P K 0.4P R3.CT 326.9 ENDS 13387229 SUBCKT 12ATT P GK Ra $124 37.9 E120 VALUE = {V(P,K) + 44.84V(G,K)} L4 4 P2 4.62 RI 201K R5S1 50.487 GP PK VALUE = {8.31E-6«(PWR(V(2).1.5) KALL LI L2L3 L4 LS 0.9988 + PWRS(V(2),1.5))/2} ENDS CGP GP LSP AC DEC 200.1 IMEG CGK G K 2.2P + TRAN (0.01M,.2M. CPK PK 0.45P PROBE ENDS END THE AUTHOR W, Marshall Leach, Jr, received B.S. and M.S. de agrees in electrical engineering from the University of South Carolina, Columbia, in 1962 and 1964, and a Ph.D. degree in electrical engineering from the Georgia Institute of Technology in 1972 In 1964 he worked at the National Aeronautics and. Space Administration in Hampton, VA. From 1965 t0 1968 he served as an officer in the U.S. Ait Since 1972 he has been a faculty member atthe Georgia Institute of Technology, where he is presently professor of electrical engineering. Dr. Leach teaches courses in applied electromagnetics and electronic design. He is a fellow of the Audio Engineering Society and a senior ‘member of the IEEE. JL Aude Eng. Soe, Vol. 43, No.3, 1995 March